Reframing The Conceptual Change Approach in Learning and Instruction Advances in Learning and Instruction Advances in Learning and Instruction Ad PDF

RE-FRAMING THE CONCEPTUAL
CHANGE APPROACH IN LEARNING AND

INSTRUCTION
ADVANCES IN LEARNING AND INSTRUCTION SERIES
Series Editors: K. Littleton, C. P. Constantinou, L. Mason, W.-M. Roth and R. Wegerif
Further details: www.elseviersocialsciences.com

www.elsevier.com/locate/series/ali
Recently Published Volumes
VERSCHAFFEL, DOCHY, BOEKAERTS AND VOSNIADOU

Instructional Psychology: Past, Present, and Future Trends
TYNJÄLÄ, VÄLIMAA AND BOULTON-LEWIS

Higher Education and Working Life: Collaborations, Confrontations and Challenges
ELEN AND CLARK

Handling Complexity in Learning Environments
HAKKARAINEN, PALONEN, PAAVOLA AND LEHTINEN

Communities of Networked Expertise
TUOMI-GROHN AND ENGESTROM

Between School and Work
DE CORTE, VERSCHAFFEL, ENTWISTLE AND VAN MERRIËNBOER

Powerful Learning Environments
Related journals – sample copies available online from:

http://www.elsevier.com
Learning and Instruction

International Journal of Educational Research
Computers and Education
The Internet and Higher Education
Early Childhood Research Quarterly
RE-FRAMING THE
CONCEPTUAL CHANGE
APPROACH IN LEARNING
AND INSTRUCTION
EDITED BY
STELLA VOSNIADOU
University of Athens, Greece
ARISTIDES BALTAS
National Technical University of Athens, Greece
XENIA VAMVAKOUSSI
University of Athens, Greece
Published in Association with the European Association for Learning and Instruction
Amsterdam ● Boston ● Heidelberg ● London ● New York ● Oxford

Paris ● San Diego ● San Francisco ● Singapore ● Sydney ● Tokyo
Elsevier
The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, UK
Radarweg 29, PO Box 211, 1000 AE Amsterdam, The Netherlands
First edition 2007
Copyright © 2007 Elsevier Ltd. All rights reserved
No part of this publication may be reproduced, stored in a retrieval system

or transmitted in any form or by any means electronic, mechanical, photocopying,
recording or otherwise without the prior written permission of the publisher
Permissions may be sought directly from Elsevier’s Science & Technology Rights
Department in Oxford, UK: phone (+44) (0) 1865 843830; fax (+44) (0) 1865 853333;
email: permissions@elsevier.com. Alternatively you can submit your request online by
visiting the Elsevier web site at http://elsevier.com/locate/permissions, and selecting
Obtaining permission to use Elsevier material
Notice
No responsibility is assumed by the publisher for any injury and/or damage to persons
or property as a matter of products liability, negligence or otherwise, or from any use
or operation of any methods, products, instructions or ideas contained in the material
herein. Because of rapid advances in the medical sciences, in particular, independent
verification of diagnoses and drug dosages should be made
British Library Cataloguing in Publication Data

A catalogue record for this book is available from the British Library
Library of Congress Cataloging-in-Publication Data

A catalog record for this book is available from the Library of Congress
ISBN-13: 978-0-08-045355-2
ISBN-10: 0-08-045355-4
For information on all Elsevier publications

visit our website at books.elsevier.com
Printed and bound in The Netherlands
07 08 09 10 11 10 9 8 7 6 5 4 3 2 1
ADVANCES IN LEARNING AND INSTRUCTION
Series Editors: K. Littleton, C. P. Constantinou, L. Mason, W.-M. Roth and R. Wegerif
Editor-in-Chief
K. Littleton, Centre for Childhood Development and Learning, Open University, MK7
6AA. E-mail: k.s.littleton@open.ac.uk
Editorial Board
P. Boscolo, University of Padova, Italy

E. De Corte, University of Leuven, Belgium
W.-M. Roth, University of Victoria, British Columbia, Canada
Publisher’s Liasion
R. Wegerif, Open University, UK

This page intentionally left blank
Contents
List of Figures xi
List of Tables xiii
Acknowledgements xv
Contributors xvii
Preface xxi
1. The Conceptual Change Approach and its Re-Framing 1

Stella Vosniadou
PART 1: Foundations of the Conceptual Change Approach: Kuhn’s

Influence 17
2. The Philosophical Foundations of the Conceptual Change Approach: An
Introduction 19
Aristides Baltas
3. In the Wake of Thomas Kuhn’s Theory of Scientific Revolutions:

The Perspective of an Historian of Science 25
Lillian Hoddeson
4. Kuhn’s Philosophical Successes? 35

Peter Machamer
5. Conceptual Change and Scientific Realism: Facing Kuhn’s Challenge 47

Theodore Arabatzis
6. Background ‘Assumptions’ and the Grammar of Conceptual Change:

Rescuing Kuhn by Means of Wittgenstein 63
Aristides Baltas
viii Contents
Commentaries 81
7. Reflections on Conceptual Change 83
Stathis Psillos
8. Conceptual Change as Structure Change: Comment on Kuhn’s Legacy 89

Matti Sintonen
PART 2: Personal Epistemology and Conceptual Change 97

9. Personal Epistemology and Conceptual Change: An Introduction 99
Stella Vosniadou
10. Epistemological Threads in the Fabric of Conceptual Change Research 105

P. Karen Murphy, Patricia A. Alexander, Jeffrey A. Greene and
Maeghan N. Edwards
11. Conceptions of Learning and the Experience of Understanding:

Thresholds, Contextual Influences, and Knowledge Objects 123
Noel Entwistle
12. Conceptual Change in Physics and Physics-Related Epistemological

Beliefs: A Relationship under Scrutiny 145
Christina Stathopoulou and Stella Vosniadou
13. Effects of Epistemological Beliefs and Learning Text Structure on

Conceptual Change 165
Lucia Mason and Monica Gava
14. Conceptual Change Ideas: Teachers’ Views and their Instructional Practice 197
Reinders Duit, Ari Widodo and Christoph T. Wodzinski
Commentary 219
15. First Steps: Scholars’ Promising Movements Into a Nascent Field of Inquiry 221
Patricia A. Alexander and Gale M. Sinatra
PART 3: Extending the Conceptual Change Approach to Mathematics

Learning 237
16. Extending the Conceptual Change Approach to Mathematics Learning: An
Introduction 239
Xenia Vamvakoussi
17. When We Clashed with the Real Numbers: Complexity of Conceptual

Change in Number Concept 247
Kaarina Merenluoto and Tuire Palonen
Contents ix
18. How Many Numbers are there in a Rational Numbers Interval? Constraints,
Synthetic Models and the Effect of the Number Line 265
Xenia Vamvakoussi and Stella Vosniadou
19. Students’ Interpretations of Literal Symbols in Algebra 283

Konstantinos P. Christou, Stella Vosniadou and Xenia Vamvakoussi
20. Teaching for Conceptual Change: The Case of Infinite Sets 299
Pessia Tsamir and Dina Tirosh
Commentaries 317
21. Nurturing Conceptual Change in Mathematics Education 319
Brian Greer and Lieven Verschaffel
22. Reconceptualizing Conceptual Change 329

Anna Sfard
Subject Index 335

List of Figures
Figure 10.1: Relational model of epistemological stances. 108

Figure 11.1: Categories used in describing conceptions of knowledge
and learning. 129
Figure 12.1: Only 11 students, all of whom were found to hold
constructivist physics-related epistemological beliefs,
were found to achieve high scores in the Force and Motion
Conceptual Evaluation (FMCE) instrument. 148
Figure 12.2: A skeletal theoretical framework for conceptualizing the
relationship between (physics-related) epistemological
beliefs and physics conceptual understanding. 150
Figure 12.3: Some examples of physics questions and problems used in
the interviews. 153
Figure 13.1: Adjusted mean scores of text-based comprehension by
condition and epistemological beliefs. 179
Figure 13.2: Adjusted mean of composite scores for conceptual change by
Figure 14.1: Steps of the constructivist teaching sequences for
conceptual change. 201
Figure 14.2: Design of the study. 202
Figure 14.3: Results of subcategories A-1 to A-5. 203
Figure 14.4: Results of subcategories B-1 to B-5. 205
Figure 14.5: Results of subcategories C-2 and C-3. 205
Figure 14.6: Results of subcategory D-1 to D-4. 206
Figure 14.7: Average time of each step of the CTS. 207
Figure 14.8: Analyses of the sequences of lessons. 207
xii List of Figures
Figure 14.9: Achievement gains (Rasch parameters and residuals) for

types A to D. 211
Figure 17.1: A simplified model of the extensions of number domains from
natural numbers to real numbers and the changes in the level of
abstraction with each enlargement. 250
Figure 17.2: Interviewed mathematicians in the two-dimensional field of
explanations of their individual thinking and their involvement
in the community of mathematics. 258
Figure 17.3: A tentative “graph” of the radical conceptual changes in the
extensions of the number concept and “areas” of students’ and
mathematicians’ explanations. 260
Figure 20.1: Vertical and Graphical Representations. 309
List of Tables
Table 11.1: Contrasting forms of understanding in revising for final

examinations. 136
Table 12.1: Criteria for identifying students’ approach to learning and studying. 154
Table 13.1: Items loading on the epistemological belief dimension “Certain
and simple Knowledge.” 172
Table 13.2: Frequency/percentage of responses to the open-ended questions
at pretest by condition and epistemological beliefs. 173
Table 13.3: Frequency/percentage of responses to the multiple-choice
questions at pre, immediate, and delayed posttests by
Table 13.4: Frequency/percentage of responses to the text-based
comprehension questions at immediate and delayed
posttests by condition and epistemological beliefs. 180
Table 13.5: Frequency/percentage of responses to the open-ended questions
at immediate and delayed posttests by condition and
epistemological beliefs. 184
Table 14.1: Constructivist-oriented science classrooms (COSC). 200
Table 14.2: Types of teachers’ implicit theories based on the analysis of
teachers’ interviews. 210
Table 17.1: Intensity of Interviewees’ Expressions and Elaborations. 253
Table 18.1: The open-ended questionnaire (QT1). 271
Table 18.2: The forced-choice questionnaire (QT2). 272
Table 18.3: Percentage of answer types, in the total of the answers
given in all six questions, as a function of the type of
questionnaire and of grade. 273
xiv List of Tables
Table 18.4: Number and percentage of students in categories formed with

respect to their performance with and without the number line,
as a function of grade and type of questionnaire. 274
Table 18.5: Percentage of students placed in the five categories, as a
function of type of questionnaire and of grade. 276
Table 18.6: Students’ accounts of the structure of rational numbers intervals:
some examples. 277
Table 19.1: Differences between the natural numbers in the context of
arithmetic and literal symbols in the context of algebra. 286
Table 19.2: An example of the way in which questions were posed in the
forced choice questionnaire. 291
Table 19.3: Percentages of students’ responses to the combined questions. 292
Table 19.4: Frequencies, percentages, and examples of students’ responses
to each question. 293
Acknowledgements
The present book owes its existence to the hard work of many people. We would like
to make reference in particular to a set of students in the inter-disciplinary graduate
programme in ‘Basic and Applied Cognitive Science’ at the University of Athens, whose
help has been instrumental in the organisation of the 4th European Symposium on
‘Conceptual Change’ out of which this volume has emerged: Irini Skopeliti, Nektarios
Mamalougos, Katerina Ligovanli, Kalliopi Ikospentaki, Christina Stathopoulou, Maria
Koulianou, Erifylli Tsirempolou, Vassiliki Siereki, Konstantinos P. Christou, and Natassa
Kyriakopoulou. We also thank all the members of the Special Interest Group on Conceptual
Change of the European Association for Research on Learning and Instruction, and parti-
cularly its coordinators (at that time) Kaarina Merenluoto and Gunilla Petersson, who have
contributed to the success of the 4th European Symposium. Finally we express our sincere
thanks to Spyridoula Efthimiou for her superb secretarial support during the preparation of
the book.
The 4th European Symposium and therefore also this book would not have been possi-
ble without the financial help from the University of Athens, the Greek Ministry of
Education (through its research programmes Pythagoras and Hrakleitos), the Greek
Ministry of Culture, the Greek Center of Educational Research, the Commersial Bank of
Greece and Gutemberg Publications.
The editors
Stella Vosniadou
Aristides Baltas
Xenia Vamvakoussi
Contributors
Patricia A. Alexander
Department of Human Development, University of Maryland, College Park, MD, USA
Theodore Arabatzis
Department of Philosophy and History of Science, University of Athens, Athens, Greece
Aristides Baltas
Faculty of Applied Mathematics and Physics, National Technical University of Athens,
Athens, Greece
Konstantinos P. Christou
Reinders Duit
Institute for Science Education, University of Kiel, Kiel, Germany
Maeghan N. Edwards
Department of Educational Psychology, The Pennsylvania State University, PA, USA
Noel Entwistle
Higher and Further Education Department, University of Edinburgh, Edinburgh, UK
Monica Gava
Department of Developmental and Socialization Psychology (DPSS), University of
Padova, Padova, Italy
Jeffrey A. Greene
University of Maryland, MD, USA
Brian Greer
2632, NE 7th Avenue, Portland, OR, USA
Lillian Hoddeson
Department of History, University of Illinois at Urbana-Champaign, Urbana, IL, USA
xviii Contributors
Peter Machamer
Department of History and Philosophy of Science, University of Pittsburgh, Pittsburgh,
PA, USA
Lucia Mason
Department of Developmental and Socialization Psychology (DPSS), University of
Padova, Padova, Italy
Kaarina Merenluoto
Department of Teacher Education, University of Turku, Turku, Finland
P. Karen Murphy
Department of Educational Psychology, The Pennsylvania State University, PA, USA
Tuire Palonen
Faculty of Education, University of Turku, Turku, Finland
Stathis Psillos
Department of Philosophy and History of Science, University of Athens, Athens,
Greece
Matti Sintonen
Department of Philosophy, University of Helsinki, Finland
Anna Sfard
Department of Education, University of Haifa, Haifa, Israel and Michigan State
University, USA
Gale M. Sinatra
Department of Educational Psychology, University of Nevada, Las Vegas, NV, USA
Matti Sintonen
Department of Philosophy, University of Helsinki, Finland
Christina Stathopoulou
Dina Tirosh
School of Education, Tel-Aviv University, Tel-Aviv, Isreal
Pessia Tsamir
School of Education, Tel-Aviv University, Tel-Aviv, Israel
Xenia Vamvakoussi
Department of Philosophy and History of Science, University of Athens, Athens,
Greece
Contributors xix
Lieven Verschaffel
Department of Educational Sciences, Katholic University of Leuven, Center for
Instructional Psychology & Technology (CIP&T), Leuven, Belgium
Stella Vosniadou
Ari Widodo
Jurusan Pendidikan Biologi, FPMIPA UPI, Jl. Dr. Setiabudhi 229, Bandung, Indonesia
Christoph T. Wodzinski
Institute for Science Education, University of Kiel, Kiel, Germany
Preface
The present volume is an outcome of the meeting of the Special Interest Group (SIG) on
Conceptual Change of the European Association for Research in Learning and Instruction
(EARLI), which took place in Delphi, Greece, from May 19 to May 23, 2004. This was
the 4th European Symposium on Conceptual Change; the previous meetings took place in
Jena (1994), Madrid (1998) and Turku (2002). This is also the third book being published
as an outcome of the conferences held by the EARLI SIG. The first book was entitled New
Perspectives on Conceptual Change [W. Schnotz, S. Vosniadou, & M. Carretero (Eds),
Elsevier Science, 1999] and the second was entitled Reconsidering Conceptual Change:
Issues in Theory and Practice [M. Limon, & L. Mason (Eds), Kluwer Academic
Publishers, 2003].
The purpose of the meetings of the Conceptual Change SIG is to discuss special theo-
retical and methodological issues on conceptual change that are of particular interest to the
SIG members. The Delphi symposium was designed to address the foundations of the con-
ceptual change approach in the philosophy and history of science and, more specifically,
the contribution of Thomas Kuhn. We were interested in examining some of the criticisms
of Kuhn’s theory and in understanding how they apply to conceptual change research in
learning and instruction.
The present volume is the outcome of the presentations and discussions that took place
at the SIG meeting. It has collected some of the papers presented by the philosophers and
historians of science at the Delphi Symposium, which are particularly helpful in illumi-
nating some of the links between philosophy and history of science and the current theo-
retical concerns in the field of learning and instruction. It also includes papers from invited
symposia on Changes on Epistemological Beliefs and Effects of Epistemological Beliefs
on Conceptual Change and Conceptual Change in Mathematics: Theoretical Issues and
Educational Applications because these are particularly appropriate in the context of the
re-framed conceptual change approach that we wanted to present in this book.
More specifically, the book consists of three sections. The first section concerns the
roots of the conceptual change approach in learning and instruction, which lie in the phi-
losophy and history of science. The influential work of Thomas Kuhn is critically exam-
ined by philosophers and historians of science who trace aspects of Kuhnian key constructs
that are still of value today. The second section examines the influence that epistemologi-
cal beliefs can have on conceptual change. The third section deals with mathematics learn-
ing and teaching from a conceptual change perspective. Epistemological beliefs and
xxii Preface
mathematics learning and teaching are new domains of study that, we argue, can be exam-
ined in a beneficial way from a conceptual change point of view. However, this implies that
the conceptual change approach should be re-defined so as to account for those aspects of
learning that have been neglected before, as well as to take into consideration the particu-
larities of a domain (mathematics learning) which differentiates it from science learning.
The general introduction to the book, the introductions to the three parts(Parts I, II and III),
as well as the commentaries that follow each part attempt to capture the important issues
in the three areas of interest, trace the questions that need to be further investigated and
strengthen the coherence among the interdisciplinary contributions. We hope that the pres-
ent book will contribute, in the tradition of the Conceptual Change SIG volumes, to fur-
ther advance our understanding of the problem of conceptual change in learning and its
implications for instruction.
The Editors
Stella Vosniadou
Aristides Baltas
Xenia Vamvakoussi
Chapter 1
The Conceptual Change Approach and its

Re-Framing
Stella Vosniadou
Theories of learning must provide an answer to the problem of how concepts change. Most
theories do that by assuming that learning is cumulative and domain general, and that
concepts change through the enrichment of prior knowledge. All of these theories find it
difficult to explain both the considerable re-organization of conceptual knowledge that takes
place with learning and development, and the difficulties students encounter particularly
when it comes to learning some of the more advanced concepts in science and mathematics.
The conceptual change approach has emerged from an effort to provide more satisfac-
tory answers to questions regarding the radical re-organization of conceptual knowledge
and the understanding of difficult science concepts. It is a constructivist approach that,
unlike previous approaches, considers knowledge to be organised in domain-specific, the-
ory-like structures, and knowledge acquisition to be characterized by theory-like changes.
Although the beginning of the conceptual change approach can be traced to physics edu-
cation, it is not restricted to physics but makes a larger claim about learning that transcends
many domains and can apply, for example, to biology (Hatano & Inagaki, 1994), psycho-
logy (Wellman, 2002) and cognitive development in general (Carey, 1985). Recently the
term ‘controversial conceptual change’ has been used to refer to the debates surrounding
certain topics, such as evolution and environmental changes, or issues that are related to
political and ideological differences and peace education.
The Classical Approach to Conceptual Change
The foundations of the conceptual change approach lie in the attempts by philosophers and
historians of science to explain theory change in science. In his well-known book The
Structure of Scientific Revolutions, Thomas Kuhn (1962), following other philosophers of sci-
ence, like Norwood Russell Hanson, Stephen Toulmin, Paul Feyerabend, Larry Laudan and
Imre Lakatos, questioned the attempts by logical positivists and logical empiricists to treat
scientific theories as sets of axioms that could be formulated in mathematical logic (known as
the Received View) and the related treatment of theory change as theory reduction. According
Reframing the Conceptual Change Approach in Learning and Instruction

Copyright © 2007 by Elsevier Ltd.
All rights of reproduction in any form reserved.
ISBN: 0-08-045355-4
2 Stella Vosniadou
to theory reduction, a theory that enjoys a high degree of confirmation cannot ever be discon-
firmed, but can only be expanded to a theory with a wider scope, or absorbed into a more
inclusive and comprehensive theory (see Suppe, 1977). Kuhn argued that normal science
operates within sets of shared beliefs, assumptions, commitments and practices that constitute
‘paradigms’. Discoveries emerge over time that cannot be accommodated within the existing
paradigm. When these anomalies accumulate, science enters a period of crisis that is eventu-
ally resolved by a revolutionary change in paradigm. According to Kuhn, different paradigms
are incommensurable. Scientific knowledge grows as we move from one paradigm to another,
but it is no longer possible to imagine the results of scientific revolutions as a cumulative,
linear progression.
In their search for a theoretical framework to conceptualize the learning of science,
some science educators turned to the philosophy and history of science as a major source
of hypotheses concerning how concepts change in the process of learning science.
Researchers like Viennot (1979), Driver and Easley (1978) and McCloskey (1983) realized
that students bring to the science learning task alternative frameworks, preconceptions or
misconceptions that are robust and difficult to extinguish. Some of them saw an analogy
between theory change in science and the need for students to change their alternative
frameworks and replace them with the scientific concepts instructed at school (Posner,
Strike, Hewson, & Gertzog, 1982). According to Posner et al. (1982), there are four fun-
damental conditions that need to be fulfilled before conceptual change can happen in
science education: (1) there must be dissatisfaction with existing conceptions, (2) there
must be a new conception that is intelligible, (3) the new conception must appear to be
plausible, and (4) the new conception should suggest the possibility of a fruitful program.
This theoretical framework, which we call the ‘classsical approach’ to conceptual change,
became the leading paradigm that guided research and instructional practices in science
education for many years.
On the basis of this ‘classical approach’, the child is like a scientist, the process of
science learning is a rational process of theory replacement, conceptual change is like a
gestalt shift that happens over a short period of time, and cognitive conflict is a major
instructional strategy for producing conceptual change.
Over the years practically all of the above-mentioned tenets of the classical approach
have been seriously questioned. Some of these criticisms are similar in many respects to
the criticisms voiced of Kuhn’s philosophical approach. Indeed, one of the purposes of the
present volume is to examine the criticisms of the Kuhnian approach that emerged over the
years in the philosophy and history of science, and evaluate their usefulness for theorizing
conceptual change in learning and instruction.
One important set of criticisms, coming from socio-cultural theory, has pointed out that
conceptual change should not be seen as only an individual, internal, cognitive process, but
also, as a social activity that takes place in a complex socio-cultural world (see Hatano,
1994; Caravita & Hallden, 1994). These researchers believe that the situational, cultural,
and educational contexts should be taken into account in trying to explain how people’s
concepts change. It is interesting to note that this point is actually very much in line with
Kuhn’s (1962) own arguments that the notion of ‘theory’, conceived as a set of propositions,
is too narrow to account for the activities of scientists and should be replaced with a ‘para-
digm’. As Machamer (this volume) notes, the introduction of the notion of ‘paradigm’ by
The Conceptual Change Approach and its Re-Framing 3
Kuhn shifts the emphasis from individuals’ minds to the role that the scientific community
and their group commitments, shared examples, and tacit knowledge play in scientific dis-
covery and change. Hoddeson (this volume) also points out how Kuhn’s influence helped
historians of science to think about science as a social activity.
We agree in part with the criticisms coming from the socio-cultural perspective and we
believe that it is important to consider the important role played by socio-cultural factors
in conceptual change. However, a serious limitation of radical socio-cultural (or situative)
perspectives (e.g., Lave, 1996; Rogoff, 1998; Saljo, 1999; Sfard, this volume) is that they
consider only the internalization or appropriation of existing cultural practices, tools, and
artifacts, and do not pay adequate attention to the active role of the individual in under-
standing or constructing new knowledge. As Hatano (1994) aptly expresses, discussing the
work of another of his Japanese colleagues (Kobayashi, 1994), ‘although understanding is
a social process, it also involves much processing by an active individual mind. It is
unlikely that conceptual change is induced only by social consensus. The post-change
conceptual system must have not only coherence but also subjective necessity. Such a
system can be built only through an individual mind’s active attempts to achieve integra-
tion and plausibility’ (p. 195). We believe that the conceptual change approach could be
re-framed to account for the effects of socio-cultural factors without ignoring the
contribution of the constructive individual (see also Vosniadou, in press, for a detailed dis-
cussion of this issue).
Other aspects of the conceptual change approach have also been criticized. One kind of
criticism centers on the nature of conceptual change and how it is achieved. For example,
many researchers believe that conceptual change is a slow and gradual process and not a
dramatic, gestalt shift that happens over a short period of time (Caravita & Hallden, 1994;
Vosniadou, 2003; Hatano & Inagaki, 1994). Others have pointed out that science learning
does not require the replacement of ‘incorrect’ with ‘correct’ conceptions, but the ability
on the part of the learner to take different points of view and understand when different
conceptions are appropriate depending on the context of use (e.g., Pozo, Gomez, & Sanz,
1999; Spada, 1994). Again we agree with these criticisms and we believe that it is possi-
ble to find a solution to the problem of how enrichment-kind of additive mechanisms can
produce radical changes in knowledge over long periods of time. Our proposal, which will
be discussed in the next section, has many similarities to the philosophical arguments
made by Baltas (this volume), and provides a new perspective to the problem of
incommensurability.
Finally, a third kind of criticism concerns the methods of teaching for conceptual
change. It has been argued that cognitive conflict is not a successful instructional strategy
for producing conceptual change, as students tend to patch up local inconsistencies in a
superficial way (Chinn & Brewer, 1993; Smith, diSessa, & Roschelle, 1993; Vosniadou,
1999). Motivational theorists have pointed out that the ‘classical approach’ represents
‘cold’ cognition and does not take into consideration affective and motivational factors.
Effective teaching for conceptual change, they point out, should try to find ways to
increase students’ motivation to change their beliefs and persuade them that conceptual
change is necessary (Pintrich, Marx, & Boyle, 1993; Sinatra & Dole, 1998; Alexander,
2001). Again, we agree with the above-mentioned criticisms and believe that the concep-
tual change approach should utilize but cannot rely on cognitive conflict as an instructional
4 Stella Vosniadou
strategy. In addition, attention needs to be paid to the development of motivated and

intentional learners who have the metaconceptual awareness necessary in order to under-
stand the differences between their naïve beliefs and the scientific theories to which they
are exposed and are capable of using sophisticated mechanisms of hypothesis testing and
deliberate belief-revision that scientists use in the process of scientific discovery
(Nersessian, 1992).
In view of the above-mentioned, there have been a number of attempts to re-frame the
classical approach (e.g., Caravita & Hallden, 1994; Hatano, 1994) including our own
(Vosniadou, 2002, 2003, in press; Vosniadou & Verschaffel, 2004). In this chapter we will
try to collect all these attempts into a proposal for a more coherent, re-framed conceptual
change approach to learning and instruction.
Re-Framing the Conceptual Change Approach
Domain Specificity and Naïve Theories
As mentioned earlier, the conceptual change approach considers conceptual change as

domain-specific theory-like change. In the ‘classical approach’ the theories that needed to
be changed were supposed to be students’ alternative frameworks, or misconceptions.
In the re-framed approach, however, the theories that need to be changed are the naïve,
intuitive, domain-specific theories constructed in early childhood, on the basis of everyday
experience under the influence of lay culture. In this respect, the re-framed approach draws
heavily on domain-specific approaches in cognitive developmental research (Carey, 1985;
Keil, 1990; Wellman, 2002; Keil, 1992; Inagaki & Hatano, 2002). In the sections that fol-
low we will examine in greater detail the assumptions regarding domain specificity and
naïve theories.
Domain specificity Most theories of learning and development, such as Piagetian and
Vygotskian approaches, information processing or socio-cultural theories are domain
general. They focus on principles, stages, mechanisms, strategies, etc., that are meant to
characterize all aspects of development and learning. In contrast, the conceptual change
approach is a domain-specific approach. It examines distinct domains of thought and
attempts to describe the processes of learning and development within these domains.
The idea that human cognition includes domain-specific mechanisms for learning is
based on a number of independent research traditions and sets of empirical findings, some
coming from animal studies (Gallistel, 1990), others based on Chomsky’s work in lin-
guistics (Chomsky, 1988). Many cognitive developmental psychologists see domain speci-
ficity through the notion of domain-specific constraints on learning (Keil, 1981, 1990). It
is argued that such constraints are needed in order to restrict the indeterminacy of experi-
ence (Goodman, 1972) and guide, amongst others, the development of language
(Markman, 1989), numeric understanding (Gelman, 1990), or physical and psychological
knowledge (Wellman & Gelman, 1998).
There is a great deal of debate in the literature as to whether domain-specific constraints
should be seen as hardwired and innate as opposed to acquired, and as having representational
content or not (see Elman et al., 1996). Many researchers prefer to see domain-specific con-
straints as innate or acquired biases or preferences that do not have representational content
but rather mitigate the interaction between a learning system and the environment (e.g., Keil,
1990; Inagaki & Hatano, 1991).
Finally, some domain-specific approaches focus on the description of the development
of expertise in different subject-matter areas, such as physics (Chi, Feltovich, & Glaser,
1981), mathematics (VanLehn, 1990; Mayer, 1985) or chess (Chase & Simon, 1973),
without necessarily appealing to innate modules or constraints.
The conceptual change approach can be applied to any of the above conceptualizations
of domain specificity, focusing on the description and explanation of the changes that take
place in the content and structure of knowledge with learning and development. Domain-
specific approaches should be seen as complementary rather than contradictory to domain-
general approaches. It is very likely that both domain-general and domain-specific
mechanisms and constraints apply to development and learning (Keil, 1990).
Naïve theories A basic characteristic of the conceptual change approach is the assump-
tion that domain-specific knowledge is theory-like. The term theory-like is used to denote
a relatively coherent body of domain-specific knowledge characterized by a distinct onto-
logy and a causality that can give rise to explanation and prediction, and not an explicit,
well-formed and socially shared scientific theory.
For example, many cognitive developmental psychologists would agree that children are
biased to notice differences in objects that move by themselves compared to those that
require the push/pull of an external agent (e.g., Golinkoff, Harding, Carlson, & Sexton,
1984). This early distinction forms the basis for the differentiation of animate from inani-
mate objects and thus the beginnings of the distinction between naïve psychology and naïve
physics. Naïve physics consists of the ontology of inanimate objects in the context of
mechanical causality, while naïve psychology consists of the ontology of animate objects in
the context of intentional, psychological causality. There is a great deal of experimental evi-
dence that by the age of six children have developed a naïve theory of physics and a naïve
psychology and very probably a naïve biology as well (Inagaki & Hatano, 2002).
The importance of the assumption that early knowledge is organized in the form of
naïve theories lies in the fact that theory-like structures are generative. As such, they make
it possible for children to formulate explanations and predictions and to deal with unfa-
miliar problems, thus enabling them to make sense of everyday phenomena. As mentioned
above, naïve theories are very different from scientific theories. They are not well formed,
they are not explicit, they are not socially shared, and they are not accompanied by meta-
conceptual awareness. It appears that children do not fully understand that their beliefs are
hypotheses that need to be tested and may be falsified, and are not very good at co-ordi-
nating theory with evidence in order to revise their explanations (Kuhn, Amsel, &
O’Loughlin, 1988; Vosniadou, 2003).
How does Conceptual Change Happen?
The nature of conceptual change The processes of knowledge acquisition with devel-
opment can proceed either in the direction of enriching existing knowledge structures or
6 Stella Vosniadou
towards restructuring them (Carey, 1985; Vosniadou & Brewer, 1987). Theory-like knowl-
edge structures allow the possibility that developmental change is theory-like change and
this is exactly what conceptual change is meant to be.
There is a great deal of evidence that cognitive development is indeed characterized by
conceptual re-organization. For example, in the domain of biology, cross-sectional devel-
opmental studies show that the biological knowledge of the 10-year old is qualitatively
different from that of the 4–6-year-old child (Carey, 1985; Hatano & Inagaki, 1994),
although there is disagreement as to how exactly this development proceeds. Young chil-
dren explain biological phenomena within the broader framework of a naïve psychology,
whereas for older children biology has become a distinct domain. Conceptual changes in
the domain of biology have been described in terms of three fundamental components:
first, the ontological distinctions between living/non-living and mind/body; second, the
modes of inference that children employ to produce predictions regarding the behavior of
biological kinds; and third, the causal-explanatory framework children employ. For exam-
ple, young children employ intentional or vitalistic causality to explain biological phe-
nomena, reason on the basis of similarity to humans, and consider plants to be non-living.
In contrast, by the age of 10 most children have formed relatively well-defined biological
categories from which they reason about biological phenomena, and have re-organized
their concept of living things to include plants.
Similar re-organizations of conceptual knowledge across early childhood years can be
found amongst others, in children’s theory of mind (Wellman, 1990), theory of matter
(Smith, Carey, & Wiser, 1985), and in astronomy (Vosniadou & Brewer, 1992, 1994). Our
work in observational astronomy has shown that considerable qualitative changes take place
in children’s concept of the earth between the ages of 4–6 and 10–12. Pre-school children
consider the earth to be a stable, stationary and flat physical object located in the center of
the universe. These beliefs about the earth are embedded within the larger, framework the-
ory of physics, that is, a naïve physics. The earth is categorized as a physical object and all
the beliefs that apply to physical objects in general are also applied to the earth. On the con-
trary, most children at the end of the elementary school think of the earth as an astronomi-
cal object, a planet, rotating around itself and revolving around the sun in a heliocentric
solar system. In this process, a significant ontological shift has taken place in the concept
of the earth. From a physical object for most first graders it has become a solar object for
the majority of sixth grader children (Vosniadou & Skopeliti, 2005). Similar ontological
shifts have been pointed out by Chi and her colleagues to occur, for example, in the case of
the concept of force, light, heat, electricity etc. (Reiner, Slotta, Chi, & Resnick, 2000).
The processs of conceptual change In the classical approach, theory changes with
learning and development were considered the result of a rational process of theory
replacement, by a thinking (like a scientist) student. Following Kuhn’s (1962) original pro-
posal, this theory replacement was supposed to be achieved within a short period of time,
like a gestalt-type switch. Again, following Kuhn, the new theory was supposed to be
incommensurable to the old one. This proposal has received a great deal of criticism both
from psychologists and educators.
The empirical studies so far show that the process of conceptual change is slow and
gradual rather than a dramatic gestalt-type shift, by learners who, unlike scientists, lack
metaconceptual awareness of their beliefs and of the process of change or, for that matter,
the need to change (Vosniadou, 2003). Furthermore, as mentioned earlier, it has become
clear that conceptual change is not only an internal cognitive process but one that happens
in broader situational, cultural and educational contexts, and that it is significantly
influenced and facilitated by socio-cultural factors (see Hatano, 1994; also Machamer, this
volume).
Mechanisms of conceptual change We believe that if we consider that existing knowl-

edge is organized in theory-like structures embedded in larger explanatory frameworks,
then conceptual change can be accounted for through the use of simple, additive mecha-
nisms, assuming, of course, that new information is coming in gradually through partici-
pation in socio-cultural activities. For example, everyday experiences with plants, such as
watering plants, seeing them become bigger, or noticing that they can die, can lead chil-
dren to understand that plants are similar to animals in certain properties, such as feeding,
growing and dying. These similarities can eventually lead children to consider plants as
living things, rather than as inanimate objects, despite the fact that they lack self-initiated
movement (see Inagaki & Hatano, 1991; Vosniadou, 1999). This category change can be
described as branch jumping (Thagard, 1988), or as an ontological shift (Chi, 1992). In
other words, it appears that small and gradual changes in the concept of plant, through
enriched experience and participation in cultural activities, together with the ability to
engage in similarity-based reasoning can eventually produce changes in the larger explana-
tory framework within which the concept of plant was originally embedded. While this
type of change can be considered as theory-like, it may be implicit and not accompanied
by metaconceptual awareness.
The use of bottom-up, implicit, additive mechanisms may not be very productive, how-
ever, in producing ‘instruction-induced conceptual change’. Instruction-induced concep-
tual change is the kind of conceptual change that requires systematic instruction in order
to be achieved. Science concepts, like the concepts of force and energy, of heat and of pho-
tosynthesis, require many years of instruction in order to be understood. This is the case
because scientific knowledge has developed over hundreds of years of scientific discovery
into rather elaborate, counter-intuitive theories that differ in their concepts, in their struc-
ture, and in the phenomena they explain from the explanations children construct on the
basis of their experience. Thus, the process of learning science appears to require children
to understand a complex and counter-intuitive scientific theory, which represents an
explanatory framework completely different from their naïve theories. In this process, the
use of implicit, additive mechanisms is bound to produce hybrid or synthetic models
(Vosniadou, in press).
For example, let us take the case where a child is simply told that the earth is a sphere
(and maybe shown a globe) without any further explanation. This information comes in
conflict with the child’s naïve model of the earth, based on everyday experience, namely,
that the earth is a flat physical object that has all the properties of physical objects, includ-
ing the necessity of being supported in order not to fall ‘down’ (what has been called
‘up/down gravity’). Our studies have shown that in such situations, the use of simple,
assimilatory types of mechanisms can give rise to a number of different synthetic models.
One possible synthetic model is the model of a ‘dual earth’ according to which there are
8 Stella Vosniadou
two earths: a flat, supported and stable earth on which people live, and a spherical, rotat-
ing earth, which is a ‘planet’ up in the sky. In forming this synthetic model, children sim-
ply add the information regarding the spherical earth to their existing knowledge without
any other changes. Another synthetic model is that of the disc earth. According to this
model, the information that the earth is spherical has been distorted to mean that the earth
is round but also flat at the same time (Vosniadou & Brewer, 1992).
Synthetic models are created because children revise some but not all of their beliefs
about the earth that need to be changed if the scientific model is to be understood, using
the bottom-up, additive mechanisms described earlier. Synthetic models happen because
the children do not have explicit knowledge of their own beliefs and therefore they under-
stand neither the contradictions between their naïve theories and the scientific explanations
to which they are exposed, nor the distortions of the scientific view that they create.
Such synthetic models can be observed in many subject-matter areas, from astronomy
and mechanics, to history and mathematics (Gelman, 1990; Vosniadou & Brewer, 1992,
1994; Vosniadou & Ioannides, 1998; also Stafylidou & Vosniadou, 2004; Christou &
Vosniadou, this volume). In mathematics, for instance, children misinterpret fractions to
consist of two independent integers and order them from smaller to bigger focusing
either on the nominator or on the denominator. Or, failing to understand the dense struc-
ture of rational numbers, many children think that there are no other numbers between
two decimals like 0.05 and 0.06, or that there is only a finite number of numbers (e.g.,
0.05, 0.051, 0.052, … 0.059, 0.06.) between them. Alternatively, they may understand
that there are infinite numbers between decimals while at the same time believing that
there is only a finite number of numbers between fractions (Vamvakoussi & Vosniadou,
2004, this volume).
In order to avoid the construction of such synthetic models, students must (1) become
aware of the inconsistencies between their naïve theories and the scientific ones, and (2)
use the top-down, conscious and deliberate mechanisms for intentional learning mentioned
earlier. In other words, instruction-induced conceptual change requires not only the
restructuring of students’ naïve theories but also the restructuring of their modes of learn-
ing and the creation of metaconceptual awareness and intentionality (Sinatra & Pintrich,
2003; Vosniadou, 2003). Finally, it should also be added that avoidance of synthetic mod-
els can also be facilitated through the use of appropriate tools and artifacts as well as by
the situational context in which the activity is taking place.
The problem of incommensurability The problem of incommensurability is the claim

that ‘if a scientific revolution has occurred, the meaning of terms in the former paradigm
are incommensurable with (have a different meaning from) the terms of the new revolu-
tionary paradigm’ (Machamer, this volume, p. 11). This is one of the most critical and yet
most contested aspects of Kuhn’s theory. Because of the criticism he received from
philosophers of science, Kuhn changed his position from one of ‘global’ incommensura-
bility to one of ‘local’ incommensurability. The claim of local incommensurability is that
only part of the meaning of the terms in the former paradigm change in the transition to
the new paradigm. As Machamer (this volume) points out, this may be a move in the right
direction, but it certainly weakens Kuhn’s original claim that the history of science is
marked by scientific revolutions among paradigms.
We are stopping to consider this issue here because most would admit that Kuhn’s
incommensurability captures something important about learning. This ‘something’ has to
do, on the one hand, with the terrible difficulty experienced by students or laymen in gen-
eral in understanding the advanced concepts of current science and mathematics, and on
the other with the difficulty we all experience in retrieving our old ways of thinking. While
all of us were children once, it is nevertheless very difficult for us to now think like a child.
Baltas (this volume) proposes a new look at Kuhn’s incommensurability that tries to save
some of these intuitions.
According to Baltas (this volume), paradigms include background assumptions that
provide the glue, or what Wittgenstein would call the ‘grammatical hinges’ on which a
theory and its concepts rest. The discovery of anomalies and inconsistencies in the process
of doing normal science often necessitates a new examination of some of the background
assumptions of the theory that creates this contradiction. Baltas argues that becoming
aware of these assumptions, whose presence accounts for the observed contradictions,
‘opens up the grammatical space’ and allows a new paradigm/theory to occur. The above
account does not claim that a new paradigm is incommensurable with the old, but, rather,
that there is an asymmetry in their relation. The scientist who has made the change is capa-
ble of considering both the old and the new paradigms and understands the assumptions,
which have been taken for granted in the old paradigm and caused the original anomalies.
The scientist who holds on to the old assumptions, on the other hand, continues not to be
able to understand how these fundamental assumptions can be questioned.
Baltas’ (this volume) account of the intuition of incommensurability is consistent in
part, with our account of the process of conceptual change. As mentioned earlier, our
empirical work in the area of young children’s ideas about the shape of the earth and the
day/night cycle, has shown that the concept of the earth is initially embedded within a
naïve physics and that children apply all the presuppositions that are applicable to physi-
cal objects in general, also to the earth. Amongst those, the assumption that gravity ope-
rates up/down poses a particularly resistant obstacle to children’s understanding of the
spherical shape of the earth and to cause many of the misconceptions or synthetic models
that children construct. Lifting this constraint allows children to view the earth in a com-
pletely new way (see Vosniadou, 2003; Vosniadou & Brewer, 1992). Of course, this
process is not under the metaconceptual control of the child who tries to understand a
scientific concept, unlike the case with the scientist engaged in scientific discovery. This
interpretation of the process of conceptual change is consistent with empirical results
coming from many other areas of science. Just to mention one more, it appears that the
assumption that force is a property of physical objects stands as an important constraint
on students’ understanding of Newtonian dynamics (Ioannides & Vosniadou, 2001; Reiner
et al., 2000).
In addition to being consistent with the above-mentioned psychological accounts of
conceptual change, Baltas’ (this volume) proposal also provides a philosophical argument
to support what educational researchers have for a long time been wanting to argue for,
namely, that understanding a scientific concept should not require the replacement of one
theory (the incorrect) with another (the correct), but rather the ability to move on to a new,
wider, broader perspective. From that point of view, conceptual change should be seen as
requiring the ability to take multiple perspectives, examine different points of view and
10 Stella Vosniadou
understand how they relate to different contexts of applicability (see, e.g., Pozo et al.,
1999; Spada, 1994; Vosniadou, 1999; Vamvakoussi & Vosniadou, this volume).
It would of course be wrong to assume that students have the above-mentioned abili-
ties, that is, to take multiple perspectives and understand how they relate to different con-
texts of use. On the contrary, all the information we have points to the conclusion that
students are not even aware of their own beliefs and especially of the hypothetical nature
of their beliefs, not to mention being able to understand different points of view. Rather
these should be seen as areas of future development and important aspects of an instruc-
tional approach for the fostering of conceptual change. All of that is consistent with cur-
rent approaches that place a great deal of emphasis on metaconceptual awareness and
intentional conceptual change, given that understanding the new paradigm is contingent
upon becoming aware of the assumptions or what we call ‘entrenched presuppositions’
(Vosniadou & Brewer, 1992), that were responsible for the contradictions in the old theory
and which must be denied in the new (Sinatra & Pintrich, 2003; Vosniadou, 2003).
We would like to emphasize, however, that the development of metaconceptual awareness
and intentional learning can be best achieved through extensive socio-cultural support
(see Hatano & Inagaki, 2003).
Epistemic beliefs The above considerations regarding the importance of background

assumptions lead us very nicely to epistemic beliefs and their role in conceptual change.
Epistemic beliefs are the beliefs individuals hold about the nature of knowledge and the
process of knowing. Research has shown that personal epistemologies are not static, but
changing and evolving, and there is a relatively well-established trend showing that indi-
viduals move from more absolutist and objectivist views about the nature of knowledge
to views that are more relativist, constructivist, and evaluative. Up to now, research on
epistemic beliefs has mainly concentrated on theoretical and methodological issues.
There are different theoretical approaches on epistemic beliefs (such as the developmental
approach, the multidimensional approach, the theory approach), and different kinds
of instruments have been designed to measure them (e.g., quantitative vs. qualitative
measures).
In the present volume we look at epistemic beliefs in a new way, emphasizing their role
in conceptual change and exploring some of the mechanisms that make this possible. The
theoretical position we want to advance is that epistemic beliefs can have both direct and
indirect influences on conceptual change. For example, beliefs in simple, stable, certain
knowledge can prevent individuals from being open to new information that questions
some of their basic assumptions, while on the contrary, individuals who believe that
knowledge is complex, uncertain and constantly evolving may be willing to ‘open up the
grammatical space’ and allow new paradigms/theories to be seriously entertained.
Epistemic beliefs can influence conceptual change also in indirect ways, for example, by
influencing students’ learning goals, study strategies and self-regulation. Some of these
issues are addressed in the papers on epistemological beliefs and conceptual change
included in the present volume.
Conceptual change in mathematics The re-framed approach to conceptual change that

we propose can be used to predict and explain students’ difficulties in the development of
mathematical knowledge. It also constitutes a valuable framework for analyzing and

reflecting on instructional interventions in mathematics teaching.
The application of the conceptual change approach in mathematics learning and teach-
ing is a relatively new attempt. Although there has been a great deal of research in the
tradition of misconceptions during the 1970s and 1980s, the mathematics education com-
munity has been reluctant to adopt the conceptual change approach, which was developed
mainly in the context of the physical sciences. This is because mathematics has been
traditionally regarded as a discipline with particular characteristics that differentiate it even
from its nearest neighbors, the physical sciences. Thomas Kuhn himself exempted mathe-
matics from the pattern of scientific development and change presented in The Structure
of Scientific Revolutions (see Mahoney, 1997). He did this because mathematics is based
on deductive proof and not on experiment, is proven to be very tolerant to anomalies and
it does not display the radical incommensurability of theory before and after revolution.
Unlike science, the formulation of a new theory in mathematics usually carries mathema-
tics to a more general level of analysis and enables a wider perspective that makes possi-
ble solutions that have been impossible to formulate before (Corry, 1993; Dauben, 1984).
However, if we consider theory change in science not as replacement, as Baltas (this
volume) argues, but as the ability to move to a new, wider theoretical perspective or para-
digm, having understood the confining role of certain background assumptions, then the
conceptual change approach can be equally well applied in science and mathematics. In
fact, from a learning point of view, it appears that students are confronted with similar
situations when they learn mathematics and science. As it is the case that students develop
a naïve physics on the basis of everyday experience, they also develop a naïve mathema-
tics that appears to be neurologically based (through a long process of evolution), and to
consist of certain core principles or presuppositions (such as the presupposition of dis-
creteness in the number concept) that facilitate some kinds of learning but inhibit others
(Dehaene, 1998; Gelman, 2000; Lipton & Spelke, 2003). Such similarities support the
argument that the conceptual change approach can be fruitfully applied in the case of
mathematics learning.
Concluding Statements
The re-framed approach to conceptual change is a constructivist, domain-specific

approach that avoids many of the criticisms of earlier attempts to account for the process
of conceptual change with learning and development. First, the focus is not on miscon-
ceptions as unitary, faulty conceptions, but on an intricate knowledge system consisting
of different domain-specific areas organized in complex theory-like structures. Second, a
distinction is made between naïve explanations, based on everyday experience and lay
culture, and those that result from learners’ attempts to synthesize new, scientific infor-
mation with existing knowledge. We consider synthetic models not to be stable
misconceptions forming alternative theories, but dynamic, situated, and constantly
changing representations that adapt to contextual variables or to the learner’s developing
knowledge. Third, this theoretical position is a constructivist one. It can explain how new
information is built on existing knowledge structures and provides a comprehensive
12 Stella Vosniadou
framework within which meaningful and detailed predictions can be made about the
knowledge acquisition process that can guide instructional interventions. It is also impor-
tant to consider that the re-framed approach to conceptual change takes into considera-
tion socio-cultural factors. It does that by considering as its primary unit of analysis the
individual participating in rich socio-cultural activities, without, however, denying that
knowledge can be acquired and stored in memory in some form. It also considers that
teaching for conceptual change cannot be achieved through cognitive means alone but
requires extensive socio-cultural support. Last, conceptual change is considered not as the
replacement of an incorrect naïve theory with a correct one, but rather as an opening up
of the conceptual space through increased metaconceptual awareness and epistemologi-
cal sophistication, creating the possibility of entertaining different perspectives and
different points of view (see Vosniadou, in press).
Acknowledgement
The present work was financially supported through the program EPEAEK II in the frame-
work of the project ‘Pythagoras — Support of University Research Groups’ with 75%
from European Social Funds and 25% from National Funds.
References
Alexander, P. A. (2001). Persuasion: Rethinking the nature of change in students’ knowledge and
beliefs. International Journal of Education, 35, 629–631.
Caravita, S., & Hallden, O. (1994). Re-framing the problem of conceptual change. Learning and
Instruction, 4, 89–111.
Carey, S. (1985). Conceptual change in childhood. Cambridge, MA: MIT Press.
Chase, W. G., & Simon, H. A. (1973). The mind’s eye in chess. In: W. G. Chase (Ed.), Visual infor-
mation processing (pp. 215–281). New York: Academic Press.
Chi, M. T. H. (1992). Conceptual change within and across ontological categories: Examples from
learning and discovery in science. In: R. Giere (Ed.), Cognitive models of science: Minnesota
studies in the philosophy of science (pp. 129–186). Minneapolis, MN: University of Minnesota
Press.
Chi, M. T. H., Feltovich, P. J., & Glaser, R. (1981). Categorization and representation in physics
problems by experts and novices. Cognitive Science, 5, 121–152.
Chinn, C., & Brewer, W. F. (1993). The role of anomalous data in knowledge acquisition: A theoret-
ical framework and implications for science instruction. Review of Educational Research, 63, 1–49.
Chomsky, N. (1988). Language and problems of knowledge. Cambridge, MA: MIT Press.
Corry, L. (1993). Kuhnian issues, scientific revolutions and the history of mathematics. Studies in
History and Philosophy of Science, 24, 95–117.
Dauben, J. (1984). Conceptual revolutions and the history of mathematics: Two studies in the growth
of knowledge. In: E. Mendelsohn (Ed.), Transformation and tradition in the sciences, essays in
honour of I. Bernard Cohen (pp. 81–103). New York: Cambridge University Press. [Reprinted in
D. Gilies (Ed.). (1992). Revolutions in mathematics (pp. 15–20). Oxford: Oxford University Press.]
Dehaene, S. (1998). The number sense: How the mind creates mathematics. Harmondsworth,
Middlesex, England: The Penguin Press (First published by Oxford: Oxford University Press, 1997).
Driver, R., & Easley, J. (1978). Pupils and paradigms: A review of literature related to concept devel-
opment in adolescent science students. Studies in Science Education, 5, 61–84.
Elman, J., Bates, L., Johnson, M., Karmiloff-Smith, A., Parisi, D., & Plunkett, K. (1996). Rethinking
innateness. A connectionist perspective on development. Cambridge, MA: MIT Press.
Gallistel, R. (1990). The organization of learning. Cambridge, MA: MIT Press.
Gelman, R. (1990). First principles organize attention to and learning about relevant data: Number
and animate-inanimate distinction as examples. Cognitive Science, 14, 79–106.
Gelman, R. (2000). The epigenesis of mathematical thinking. Journal of Applied Developmental
Psychology, 21, 27–37.
Golinkoff, R. M., Harding, C. G., Carlson, V., & Sexton, M. E. (1984). The infant’s perception of
causal events: The distinction between animate and inanimate objects. In: L. P. Lipsitt, &
C. Rovee-Collier (Eds), Advances in infancy research. Norwood, NJ: Ablex.
Goodman, N. (1972). Seven strictures on similarity. In: N. Goodman (Ed.), Problems and projects
(pp. 341–352). New York: Bobbs-Merrill.
Hatano, G. (Guest Editor). (1994). Introduction: Conceptual change — Japanese perspectives.
Special Issue of Human Development, 37(4), 189–197.
Hatano, G., & Inagaki, K. (1994). Young children’s naive theory of biology. Cognition, 50, 171–188.
Hatano, G., & Inagaki, K. (2003). When is conceptual change intended?: A cognitive-sociocultural
view. In: G. M. Sinatra, & P. R. Pintrich (Eds), Intentional conceptual change (pp. 407–427).
Mahwah, NJ: Erlbaum.
Inagaki, K., & Hatano, G. (1991). Constrained person analogy in young children’s biological infer-
ence. Cognitive development, 6, 219–231.
Inagaki, K., & Hatano, G. (2002). Young children’s naive thinking about the biological world.
Oxford: Psychology Press.
Ioannides, Ch., & Vosniadou, S. (2001). The changing meanings of force. Cognitive Science Quarterly,
2(1), 5–62.
Keil, F. C. (1981). Constraints on knowledge and cognitive development. Psychological Review, 88,
197–227.
Keil, F. C. (1990). Constraints on constraints: Surveying the epigenetic landscape. Cognitive Science,
14, 135–168.
Keil, F. C. (1992). Concepts, kinds, and cognitive development. Cambridge, MA: MIT Press.
Kobayashi, Y. (1994). Conceptual acquisition and change through social interaction. Human
Development, 37, 233–241.
Kuhn, T. (1962). The structure of scientific revolutions. Chicago, IL: University of Chicago Press.
Kuhn, D., Amsel, E., & O’Loughlin, M. (1988). The development of scientific thinking skills. San
Diego, CA: Academic Press.
Lave, J. (1996). Teaching, as learning, in practice. Mind, Culture, and Activity, 3, 149–164.
Lipton, J. S., & Spelke, E. S. (2003). Origins of number sense: Large numbers discrimination in
human infants. Psychological Science, 4, 396–401.
Mahoney, M. S. (1997). Revolution in mathematics (Unpublished manuscript).
Markman, E. M. (1989). Categorization and naming in children: Problems of induction. Cambridge,
MA: MIT Press.
Mayer, R. E. (1985). Implications of cognitive psychology for instruction in mathematical problem
solving. In: E. A. Silver (Ed.), Teaching and learning mathematical problem solving: Multiple
research perspectives (pp. 123–138). Hillsdale, NJ: Lawrence Erlbaum Associates.
McCloskey, M. (1983). Naive theories of motion. In: D. Gentner, & A. Stevens (Eds), Mental models.
Hillsdale, NJ: Erlbaum.
Nersessian, N. J. (1992). How do scientists think? Capturing the dynamics of conceptual change in
science. In: R. N. Giere (Ed.), Cognitive models of science (pp. 3–45). Minneapolis, MN:
University of Minnesota Press.
14 Stella Vosniadou
Pintrich, P. R., Marx, R. W., & Boyle, R. A. (1993). Beyond cold conceptual change: The role of
motivational beliefs and classroom contextual factors in the process of conceptual change. Review
of Educational Research, 63, 167–199.
Posner, G. J., Strike, K. A., Hewson, P. W., & Gertzog, W. A. (1982). Accommodation of a scientific
conception: Towards a theory of conceptual change. Science Education, 66, 211–227.
Pozo, J., Gómez, M., & Sanz, A. (1999). When change does not mean replacement: Different repre-
sentations for different contexts. In: W. Schnotz, S. Vosniadou, & M. Carretero (Eds), New
perspectives on conceptual change (pp. 161–174). Amsterdam: Pergamon.
Reiner, M., Slotta, J. D., Chi, M. T. H., & Resnick, L. B. (2000). Naive physics reasoning: A com-
mitment to substance-based conceptions. Cognition and Instruction, 18(1), 1–34.
Rogoff, B. (1998). Cognition as a collaborative process. In: D. Kuhn, & R. S. Siegler (Eds), Handbook
of child psychology. Cognition, perception, and language (Vol. 2, pp. 679–744). New York: Wiley.
Saljo, R. (1999). Learning as the use of tools. A sociocultural perspective on the human-technology
link. In: K. Littleton, & P. Light (Eds), Learning with computers. Analysing productive interaction
(pp. 144–161). New York: Routledge.
Sinatra, G., & Dole, J. (1998). Case studies in conceptual change: A social psychological perspec-
tive. In: B. Guzzeti, & C. Hynd (Eds), Perspectives on conceptual change: Multiple ways to under-
stand knowing and learning in a complex world (pp. 39–53). Mahwah, NJ: Erlbaum.
Sinatra, G. M., & Pintrich, P. (Eds). (2003). Intentional conceptual change. Mahwah, NJ: Lawrence
Erlbaum Associates, Inc.
Smith, C., Carey, S., & Wiser, M. (1985). On differentiation: A case study of the development of the
concepts of size, weight and density. Cognition, 21, 177–237.
Smith, J. P., diSessa, A. A., & Roschelle, J. (1993). Misconceptions reconceived: A constructivist
analysis of knowledge in transition. The Journal of the Learning Sciences, 3, 115–163.
Spada, H. (1994). Conceptual change or multiple representations? Learning and Instruction, 4, 113–116.
Stafylidou, S., & Vosniadou, S. (2004). The development of students’ understanding of the numeri-
cal value of fractions. In: L. Verschaffel, & S. Vosniadou (Guest Editors), Conceptual change in
mathematics learning and teaching; Special Issue of Learning and Instruction, 14(5), 503–518.
Suppe, F. (1977). The structure of scientific theories (2nd ed.). Chicago, IL: University of Illinois Press.
Thagard, P. (1988). Computational philosophies of science. A Bradford Book. Cambridge, MA: MIT
Press.
Vamvakoussi, X., & Vosniadou, S. (2004). Understanding the structure of the set of rational num-
bers: A conceptual change approach. Learning and Instruction, 14, 453–467.
VanLehn, K. (1990). Mind bugs: The origins of procedural misconceptions. Cambridge, MA: MIT
Press.
Viennot, L. (1979). Spontaneous reasoning in elementary dynamics. European Journal of Science
Education, 1(2), 205–221.
Vosniadou, S. (1999). Conceptual change research: State of the art and future directions. In:
W. Schnotz, S. Vosniadou, & M. Carretero (Eds), New perspectives on conceptual change
(pp. 3–13). Oxford: Elsevier Science.
Vosniadou, S. (2002). On the nature of naïve physics. In: M. Limon, & L. Mason (Eds), Reconsidering
the processes of conceptual change (pp. 61–76). Dordrecht, The Netherlands: Kluwer Academic
Publishers.
Vosniadou, S. (2003). Exploring the relationships between conceptual change and intentional learn-
ing. In: G. M. Sinatra, & P. R. Pintrich (Eds), Intentional conceptual change (pp. 377–406).
Mahwah, NJ: Lawrence Erlbaum Associates.
Vosniadou, S. (in press). The cognitive-situative divide and the problem of conceptual change.
Educational Psychologist.
Vosniadou, S., & Brewer, W. F. (1987). Theories of knowledge restructuring in development. Review
of Educational Research, 57(1), 51–67.
Vosniadou, S., & Brewer, W. F. (1992). Mental models of the earth: A study of conceptual change in
childhood. Cognitive Psychology, 24, 535–585.
Vosniadou, S., & Brewer, W. F. (1994). Mental models of the day/night cycle. Cognitive Science, 18,
123–183.
Vosniadou, S., & Ioannides, C. (1998). From conceptual development to science education: A psy-
chological point of view. International Journal of Science Education, 20(10), 1213–1230.
Vosniadou, S., & Skopeliti, I. (2005). Developmental shifts in children’s categorization of the earth.
In: B. G. Bara, L. Barsalou, & M. Bucciarelli (Eds), Proceedings of the XXVII annual conference
of the cognitive science society (pp. 2325–2330).
Vosniadou, S., & Verschaffel, L. (2004). Extending the conceptual change approach to mathematics
learning and teaching. In: L. Verschaffel, & S. Vosniadou (Guest Editors), Conceptual change in
mathematics learning and teaching (Special Issue) Learning and Instruction, 14(5), 445–451.
Wellman, H. M. (1990). The child’s theory of mind. A Bradford Book. Cambridge, MA: MIT Press
Wellman, H. M. (2002). Understanding the psychological world: Developing a theory of mind. In:
U. Goswami (Ed.), Blackwell handbook of childhood cognitive development (pp. 167–187).
Oxford, UK: Blackwell.
Wellman, H. M., & Gelman, S. A. (1998). Knowledge acquisition in foundational domains. In:
D. Kuhn, & R. Siegler (Eds), Cognition, perception and language. Handbook of child psychology
(5th edn., Vol. 2, pp. 523–573). New York: Wiley.
PART 1:
FOUNDATIONS OF THE CONCEPTUAL

CHANGE APPROACH: KUHN’S INFLUENCE
Chapter 2
The Philosophical Foundations of the

Conceptual Change Approach: An
Introduction
Aristides Baltas
Philosophy of science situates itself by definition at the crossroads where general philos-
ophy meets science — mostly natural science — in its results, in its structure and in its
developments. It is the discipline into which general philosophical outlooks and general
philosophical methods are brought to bear on the deeper understanding of science thus aid-
ing the construction of its very self-image. Concomitantly, in a sort of inverse movement
as it were, the more significant scientific achievements never fail to pose the kind of ques-
tions which provoke philosophy to respond, perhaps even at the cost of its drastic internal
reorganization. This is not to say that the relations between science and philosophy of sci-
ence have always been perfectly smooth and perennially harmonious. Episodes of pro-
found misunderstanding and periods of mutual mistrust among the practitioners of the two
endeavors have been witnessed in the not too distant past and will no doubt be witnessed
again in the future. The long and intricate episode having been dubbed “the science wars”
is a case in point.1 Nevertheless even those unhappy moments attest to one fact: philoso-
phy of science remains constitutively open to what lies outside of it.
1
“Science wars” refers to the long and particularly heated debate, involving all kinds of media and attaining inter-
national proportions, which was sparked by an article written by the physicist Alan Sokal and published in the
journal Social Text in the late 1990s. The article appeared as offering a “post-modern” interpretation of some fun-
damental physical theories by seemingly relying on ideas brought forth by famous French intellectuals like
Derrida or Lacan. Immediately after publication, Sokal revealed that his article was a hoax and indicted Social
Text for publishing it. The editors replied and a long and intricate debate ensued implicating scientists, journal-
ists, philosophers and historians of science as well as academics of various disciplines in the humanities not only
in the United States but in many other countries as well. Starting from what the ethics and the criteria for publi-
cation should be, the debate became generalized to issues concerning the differences separating and the relations
holding natural and social sciences and the humanities in general, what “post-modern” thought might amount to,
its influence in American universities and the reasons for this influence and so on so forth. Some of the scientists
participating in the debate went as far as to argue that philosophers of science like Kuhn or even Popper had had
a pernicious influence on the public understanding of science if not on the development of science itself.

ISBN: 0-08-045355-4
20 Aristides Baltas
Such openness has been institutionally acknowledged since the end of World War II.
Many American and European universities have set up departments and programs that
have brought together philosophy of science and history of science in a way tending to turn
the two into an almost single discipline. More recently, the openness in question has been
further ratified by the creation of Science and Technology Studies — or Science,
Technology and Society — departments or programs that add more partners to the study
of science in a truly interdisciplinary spirit. In these departments or programs, philosophy
of science crosses its insights and its methods not only with history of science but also with
sociology of science, with the study of technology, with scientific policy studies, with
engineering studies, with feminist or post-colonial approaches to such issues, with social
and cultural anthropology and the list remains open.
But philosophy of science has proved its open character and its outward-looking pro-
clivities yet in another direction. This is the direction covered by some disciplines or areas
of research that are not concerned directly with natural science and its development as
such. Cognitive and developmental psychology, science teaching, science learning and sci-
ence education in general, all disciplines concerned with the way in which real people,
children and adults, acquire and change their beliefs — scientific, epistemological or
everyday — have discovered in at least some areas of philosophy and history of science
ideas and tools which could help them proceed with their own investigations. Conversely,
significant achievements and original, productive ideas of these disciplines and areas of
research have established decisively some of the more important approaches in the history
and philosophy of science and have thus assisted them to articulate their own guiding ideas
in close connection with what really happens when real people set out to know or are
helped to come to know this or that. The works of Ronald Giere, of Paul Thagard, or of
Nancy Nersessian come here readily to mind.
It is almost uncontroversial that the “historicist turn” in philosophy of science, inaugu-
rated effectively by Kuhn’s publishing the Structure of Scientific Revolutions, was the deci-
sive factor in opening history and philosophy of science in this last direction. However, a
lot of water has obviously run down the river of history and philosophy of science since
then. As it was only to be expected, not all such subsequent developments could have a
direct bearing on what mainly interests the present volume. What does have an important
bearing on it, however, is the fact that, after a period of quasi-oblivion, the main ideas and
insights of what Kuhn has bequeathed us have re-emerged forcefully on the foreground.
Books, special issues of journals, many interesting articles or essays, conferences and sym-
posia in many parts of the world attest to this fact. We should include in the list Hoyningen-
Huene’s (1993) meticulous study, Fuller’s (2000) very controversial historio-sociological
approach which, despite its valiant efforts, did not manage to lay Kuhn’s views to rest, the
balanced account of Sharrock and Read (2002) as well as the collection of particularly inter-
esting papers edited by Nickles (2003). In addition, it should be mentioned that a special
issue of Philosophia Scientiae, edited by Soler (2004), has been devoted to the issue of
incommensurability more or less in Kuhn’s sense of the term, while a volume covering fun-
damentally the same area — to be edited by Howard Sankey, Paul Hoyningen-Huene and
Léna Soler and to be published by Springer — is actually at the last stages of preparation.
As it is usually the case in philosophy, the historical distance covered and the new
insights gained along the way have brought those who work in history and philosophy of
The Philosophical Foundations of the Conceptual Change Approach 21
science to a position where they can situate and assess which parts or which aspects of
Kuhn’s work remain of lasting significance. Among them, we should include, I believe, his
highlighting the role of scientific communities; his considering a paradigm as involving
the theoretical dimension of science as indissolubly linked with the corresponding para-
digmatic applications; his distinguishing within a unified framework how a new paradigm
emerges from how it can be learnt; his connecting, at least indirectly, the incommensura-
bility of succeeding paradigms with the obstacles one encounters in coming to understand
the novel paradigm; and so on and so forth. Given what we have said just above, such sig-
nificance cannot concern only history and philosophy of science narrowly conceived. It is
or it can become of equal value to all the disciplines and areas of research to which phi-
losophy and history of science have become opened by Kuhn’s work because all of them
have proven capable of benefiting from the associated critical potential. In other words, the
critical re-emergence in question can enhance the dialogue between history and philoso-
phy of science, on one hand, and cognitive and developmental psychology and science
education, on the other, still further.
All chapters in Part I bear eloquent witness to this possibility, if not fact. Although all
of them are written within the disciplinary confines of history and philosophy of science,
each one of them not only throws its own particular light on this or that aspect of Kuhn’s
overall approach as it is being currently discussed within that discipline but, much more
importantly, it offers a set of important handles to the chapters coming next in the volume.
Thus Hoddeson (this volume) sets forth how the “sociologist turn” in philosophy and
history of science, sparked partly by Kuhn’s own work, eventually left behind sociology of
science narrowly conceived and returned to history of science proper. This movement
arrived to enlarge our understanding of the social dimension of science, the importance of
which Kuhn’s work has already underscored, and to enlarge it in a way that challenged
Kuhn’s own views. Thus, by taking systematically into account the social determinations
of the identity of scientists (their gender, their race, the particulars of the social position-
ing of their discourse, etc.) these novel historiographic approaches effectively undermined
the distinction between internal and external history that Kuhn had left more or less intact.
This is to say that these approaches ceased to consider scientific concepts and theories as
simply having a proper life of their own (internal history), radically distinguished from the
evolution of social institutions and of the social relations generally wherein science func-
tions (external history), but took the very content of scientific ideas as indebted, at least to
some extent, to the social determinations in question. In this way Hoddeson offers, at least
indirectly, ways and means which cognitive and developmental psychology and science
education might exploit for the purpose of integrating the social element more fully into
their own approaches.
The “clarity and distinctness” of the major components of Kuhn’s conceptual arsenal
(paradigm, puzzle solving, scientific community, incommensurability and so forth) is what
the contribution of Machamer (this volume) focuses on. The ways by which these concepts
have been received, criticized and elaborated upon by many philosophers of science since the
Structure of Scientific Revolutions was first published, as discussed in conjunction with both
their philosophical ancestry and Kuhn’s later views on the relevant issues, allow Machamer
to pinpoint the insights and ideas of Kuhn’s overall approach which he considers have
remained important. Among them, he highlights the relation between knowing a theory and
22 Aristides Baltas
being capable of applying it in novel circumstances as both aspects are tied together in
Kuhn’s notion of “paradigm” (a confusing notion, as Machamer points out and as Kuhn him-
self later admitted), the idea of local incommensurability which Kuhn’s later work brought
to the forefront and the determinative importance of countenancing science in terms of sci-
entific communities rather than in terms of the isolated scientist. In this way Machamer’s
essay helps the readers of the present volume to understand exactly where they are standing
today in respect to Kuhn’s approach and to go on with their own work starting from there.
Arabatzis (this volume) discusses one of the issues that have constituted a major battle-
ground within philosophy of science at least since the 1980s, namely scientific realism.
Arabatzis argues that the many views to the contrary notwithstanding, Kuhn’s approach need
not imply either anti-realism or relativism. Rather, and at least with regard to the objective
existence of entities like the electron, Kuhn’s views align themselves with scientific realism
once they are specified and elucidated in the way Arabatzis proposes. Thus, granting that sci-
entific concepts indeed change, he argues that at least some of them (such as “electron”) may
continue referring to the same entities even after a radical paradigm change affecting pro-
foundly their meaning has come about.2 Although this discussion might appear somewhat
“esoteric” in a volume like the present one, it is my conviction that what Arabatzis has to say
on the issue allows cognitive and developmental psychologists as well as science educators
to lay to rest worries about the “existence of the objective world”, which inevitably crop up
in their own practices, sometimes getting exacerbated by some misreadings of Kuhn’s work
and by some misunderstandings of what this work in fact implies.
Finally, my essay (Baltas, this volume) considers the meaning dimension of a Kuhnian
paradigm change. It relies on Wittgenstein’s conception of grammar to uphold that a novel
paradigm discloses and puts in question some of the grammatical conditions (which I call
background “assumptions”) securing the meaning of some concepts of the old paradigm.
On this basis, the essay tries to relate Kuhnian incommensurability to the grammatical
challenge in question and discusses some of the associated issues arguing that Kuhn’s
approach, once read in this way, can be “saved” from the charges of idealism and rela-
tivism. Moreover, that such a grammatical barrier separates succeeding paradigms explains
why students are impeded in their efforts to understand the novel paradigm and thus the
essay offers, at least indirectly, some handles for developing an efficient teaching strategy.
Science educators as well as cognitive and developmental psychologists may thus find here
some ideas intimately connected to their own work.
Last but not the least, the comments of Stathis Psillos and Matti Sintonen that follow
the chapters of Part I should be read carefully for a number of reasons. First, they illus-
trate in an exemplary fashion the kind of painstaking elucidatory work forming the daily
bread of philosophy of science. Second, the points of agreement or disagreement they
focus on, together with the arguments they present for the purpose, offer much room for
thought to a reader eager to understand more fully the contributions these comments are
making. Third, the work of Sellars, which Psillos invokes, as well as the work of Sneed
and Stegmüller, which Sintonen appeals to, are, at least according to me,3 very important
2
A full presentation of the argument in conjunction with all the associated intricate historical details is to be
found in Arabatzis (2006).
3
In respect to the Sneed–Stegmüller approach, I permit myself to refer to Baltas (1989).
The Philosophical Foundations of the Conceptual Change Approach 23
philosophical contributions in their own right which clarify, even if from very different
angles and among many other things, what science is all about.
Thus, among many other things, the way Sellars distinguishes the “scientific” from the
“manifest image” (Sellars, 1991) as well as the way he treats theory change (Sellars, 1973)
lays out a royal path for the serious philosophical treatment of most of the issues that
Kuhn’s work has opened in direct connection with more general epistemological and meta-
physical concerns. On the other hand and in the opposite direction as it were, the heavy
formalism of the Sneed–Stegmüller way of conceiving the structure of mathematical
physics — a way of conceiving which Kuhn himself was quite sympathetic to (Kuhn,
1976) — allows many of the same issues to become pinned down in a way assisting their
clarification and their assessment. The reader of the present volume would profit greatly, I
believe, if she became acquainted more fully with these not-so-well-known philosophical
approaches, approaches to which Psillos and Sintonen offer us substantial, if short, intro-
ductions.
To close, I can only say that the interdisciplinary dialogue that started a long time ago
can go on more strongly and more productively with the help of the present volume. The
reader is invited to ascertain this and, why not, to offer her own contribution.
References
Arabatzis, T. (2006). Representing electrons: A biographical approach to theoretical entities.
Chicago: The University of Chicago Press.
Baltas, A. (1989). Louis Althusser and Joseph D. Sneed: A strange encounter in philosophy of sci-
ence. In: K. Gavroglu, Y. Goudaroulis, & P. Nicolacopoulos (Eds), Imre Lakatos and theories of
scientific change (pp. 269–286). Dordrecht, The Netherlands: Kluwer Academic Publishers.
Fuller, S. (2000). Thomas Kuhn: A philosophical history for our times. Chicago, IL: The University
of Chicago Press.
Hoyningen-Huene, P. (1993). Reconstructing scientific revolutions: Thomas S. Kuhn’s philosophy of
science. Chicago, IL: The University of Chicago Press.
Kuhn, T. S. (1976). Theory change as structure change: Comments on the Sneed formalism.
Erkenntnis, 10, 179–199.
Nickles, T. (Ed.). (2003). Thomas Kuhn. Cambridge, UK: Cambridge University Press.
Sellars, W. (1973). Conceptual change. In: G. Pearce, & P. Maynard (Eds), Conceptual change (pp.
77–93). Dordrecht, The Netherlands: D. Reidel.
Sellars, W. (1991). Philosophy and the scientific image of man. In: W. Sellars (Ed.), Science, per-
ception and reality (pp. 1–41). Atascadero, CA: Ridgeview Publishing Co.
Sharrock, W., & Read, R. (2002). Kuhn, philosopher of the scientific revolution. Cambridge, UK:
Polity Press.
Soler, L. (Ed.). (2004). Le problème de l’incommensurabilité un demi-siècle après. Special Issue of
Philosophia Scientiae, 8(1).
Chapter 3
In the Wake of Thomas Kuhn’s Theory of

Scientific Revolutions: The Perspective of an
Historian of Science
Lillian Hoddeson
Thomas S. Kuhn was that special teacher who inspired me to change fields, from physics
to history of science. He had made the same change himself shortly after taking his Ph.D.
He had also been trained in theoretical physics but found that studying the historical deve-
lopment of science interested him more than practicing science.
In my own case, this switch began about five years after I took my Ph.D. I remember
stepping with great excitement into what I experienced as a new world, one with different
assumptions about the meaning of knowledge production. Here, the data were no longer
measurements but instead were traces of the past found in letters, scientific papers, reports,
or even the fragile memories of individuals. Problem solving and creativity in the new
domain depended on interpreting historical, rather than physical, events. I never regretted
making the switch from physics to history, and I owe a special debt of gratitude to Kuhn
for serving as a model of one who had the courage to break with familiar teachings and
turn to new ones.
Kuhn taught me many things about history that as a scientist I had not encountered. The
most important was always to view events from the perspective of those who lived in the
period being studied, never from the present vantage. One had to ask what was known
then? What did people believe then? Why? What did the authorities preach? Was it rea-
sonable then? What were the stakes in holding one idea rather than another? Such ques-
tions invariably will lead the historian of science outside the domain of science. Kuhn
encouraged his students to look there too.
For me, working with Kuhn meant returning to the status of a graduate student, even
while I was on a physics teaching faculty. In those years, 1972–75, I took two seminars that
Kuhn offered, both about major conceptual change in physics. One seminar focused on the
scientific revolution that brought us from Newtonian mechanics to quantum mechanics; the
other was on the revolution that yielded the first two laws of thermodynamics. In both semi-
nars all the students, as well as Kuhn, read papers and books from the periods being stud-
ied in the languages in which they had been written, usually German. Their messages were

ISBN: 0-08-045355-4
26 Lillian Hoddeson
couched in the incredibly awkward and complex formulations that scientists use when
struggling to explain phenomena that the current intellectual tools cannot treat.
I had already read most of Kuhn’s prior works in the history of science, including his
masterpiece about the switch from the Ptolemaic to the Copernican picture in astronomy.
In Kuhnian terms, one could speak of an old world giving way to a new and very differ-
ently configured one because scientists shared a new set of assumptions and agreements,
what he called a new paradigm. Many scholars in different fields subsequently adapted
Kuhn’s account of conceptual change, for instance to the problem of understanding how
children’s models of the earth transform (Vosniadou & Brewer, 1992). This part of Kuhn’s
teaching is as valuable today as it was in the years when Kuhn’s theory was still new. His
well-known book, The Structure of Scientific Revolutions, became a standard text in count-
less fields, from history to literature to education (Kuhn, 1962).
But in my field of history of science — my field and Kuhn’s — most scholars presently
consider Kuhn’s Structure too limited for explaining the problems that are of most inter-
est. Even though Kuhn moved from theoretical physics to the history of science, his
approach remained close to that of a physicist, abstracting simplified patterns from
sequences of events in order to seek general structures and laws. Both the strengths and the
limitations of Kuhn’s theory come from this approach.
If we recall the leading features of The Structure of Scientific Revolutions, we can see
how much they depend on sharp conceptual oppositions and disjunctions. The basic oppo-
sition between normal and revolutionary science is the most striking. Normal science oper-
ates within the sets of shared assumptions and agreements that constitute paradigms, and
by working within those constraints, scientists can dedicate their efforts to solving prob-
lems. In Kuhn’s scheme, discoveries emerge over time that cannot be accommodated
within the reigning paradigm. And when these anomalies accumulate beyond a certain
point, science enters a period of crisis that is eventually resolved by a revolution, a change
of paradigm. Though we can trace the developments that produce these revolutions, which
typically take one or more generations to work themselves out, the relation between the
previous paradigm and the one that replaces it is sharply discontinuous, so that scientists
working in the new paradigm inhabit a different world from that of their predecessors.
This claim for the discontinuity between successive paradigms, the distinctive feature
of Structure, is both subversive and enduring. It is subversive because it discredits the
belief that scientific knowledge is cumulative, moving ever closer to a true and complete
account of nature. Instead, Kuhn insists that different paradigms are incommensurable;
scientific knowledge grows as we move from one to another, but it is no longer possible to
imagine the results of scientific revolutions as a cumulative linear progression. This con-
tinues to be Kuhn’s most controversial claim; it was and is still rejected by many scientists
and their allies who maintain a belief in scientific progress. It is also enduring because it
gets some things right. At the level of theory, there really are revolutionary changes and
incommensurable differences between earlier ways of making sense and the ways that
eventually replace them — like the incommensurable pictures of burning in chemistry,
where in the earlier view intangible particles called phlogista stream out of burning objects
and in the later view oxygen combines chemically with the burning materials.
In terms of general cultural impact, the notion of revolutionary paradigm shift is also
Kuhn’s most influential achievement. It has undoubtedly been used too widely and loosely
In the Wake of Thomas Kuhn’s Theory of Scientific Revolutions 27
for describing or promoting changes that are far less radical than those he analyzed. Still,
it has provided a useful model for researchers working in other fields, such as psychology
and education. Such borrowings seem particularly appropriate because Kuhn borrowed, or
was influenced by, psychological models such as the gestalt switch and Piaget’s stages of
cognitive development.
In science studies, Kuhn’s most important influence had less to do with theoretical
models than with the way he helped a generation of scholars who began to think about sci-
ence as a social activity that occurs in a particular period. (For disparate views on this
issue, see, e.g., Keller, 1998; Pickering, 2001; Rouse, 1998). Expanding the field of study
in the history of science from experiments and theories to scientific communities, and to
communication within and between these communities, redirected the history of science
toward considering a wider range of contributing factors and contingencies. Kuhn turned
us in that direction, but he himself did not pursue it very far; his work in the history and
philosophy of science remained strongly focused on the conceptual part of scientific
knowledge production.
Kuhn was certainly aware that there is always much more involved in scientific knowl-
edge production than the scientific work itself. He often referred, implicitly if not explic-
itly, to the real-world factors that fall outside science proper, like communities, education,
values, traditions, perceptions, and cognition, as well as to the material basis of phenom-
ena. Thus in The Copernican Revolution, Kuhn (1959, pp. 14–35, 114–123) mentioned
religious beliefs, scholastic education, and how astronomical observations were made.
Such factors play a far less prominent role in Kuhn’s more famous account, The Structure
of Scientific Revolutions, perhaps because he wanted this work to speak to a wide audi-
ence. I think he would have preferred to go past the first steps he took in considering sci-
ence in social, material, and cognitive terms, as well as in intellectual terms, and build a
more fully articulated theory that encompassed such aspects as gender, class, race and eth-
nicity, or how competitions, jealousies, ambitions, and religious beliefs enter the story of
scientific knowledge production. But he could not work them rigorously or elegantly into
his theory. So he left them out, perhaps hoping other scholars would figure out how to
incorporate them (Keller, 1998).
The fact that Kuhn did not or could not do much with the multitude of contingencies does
not mean that he did not care about them. In fact, his theory’s limited ability to deal with
social and cultural influences caused him considerable anguish. I came to understand this as
a student in his seminar on the history of quantum mechanics during the spring of 1973.
Kuhn’s plan was to retrace Werner Heisenberg’s intellectual route to his formulation of
quantum mechanics. To help place ourselves into the period, we spent the first weeks of the
seminar slogging through countless experimental anomalies that appeared in the work of
physicists who were studying atomic phenomena in the early years of the twentieth century.
Most of the problems arose from the failure to explain the masses of new data available on
the atomic spectra of particular atoms. Arnold Sommerfeld’s Atombau und Spektrallinien
was one of the works we plowed through, of course in the original German. The other texts
included a number of confusing and abstruse journal articles by Werner Heisenberg,
Wolfgang Pauli, Niels Bohr, Hendrik Kramers, John Slater, and others. For the students in
the seminar, and for Kuhn, it was an exciting but extremely hard winter journey through
rugged scientific terrain. For weeks we seemed to be making little noticeable progress.
28 Lillian Hoddeson
At last, about the time that spring finally came to Princeton, we encountered the first
hints of a break. Reading extremely difficult German papers, we began our slow ascent to
the climax of the Kuhnian narrative about quantum mechanics. We struggled as we worked
through the most obscure work of all, in which Heisenberg and Kramers offered their last
wheezing gasp at explaining the stationary states of an anharmonic oscillator using mys-
terious diagrams; it was not clear why these diagrams worked. The patchwork picture that
emerged, neither classical nor quantum-mechanical, harked back to the messy adjusted
Ptolemaic system before the Copernican revolution.
The next step was the “turning point”, Heisenberg’s breakthrough paper in which the
first version of quantum mechanics came into focus. He conceived this work in Helgoland,
a barren island in the North Sea where grass cannot grow. The twenty-three year-old trav-
eled there in June 1925 because he was suffering from a severe attack of hay fever.
At this point in the seminar, Kuhn was also exhibiting some discomfort. He suddenly
announced that we needed to decide “what to do about” a major paper by Paul Forman that
had appeared a year earlier in the journal Historical Studies in the Physical Sciences (now
Historical Studies in the Physical and Biological Sciences) (Forman, 1971). Forman, who
had been one of Kuhn’s Ph.D. students, had recently worked with Kuhn in a massive effort
to preserve the documents of the history of quantum mechanics and also interview leading
figures in that history.1 His paper dealt with acausality and indeterminacy, key components
of the new quantum mechanics picture. Forman argued, radically, that the notion of
acausality arose naturally for physicists in Weimar Germany because of its uncertain and
indeterminate political climate. He hypothesized that in Weimar Germany it was easier for
physicists to give up the secure causal basis they had enjoyed as Newtonians.
As was fairly clear to the students in the seminar, our respected teacher was experiencing
some sort of a crisis of his own. Kuhn repeatedly spoke about Forman’s thesis as “terribly
important”, but he could not fit its social perspective on quantum mechanics into his narrative
about normal science, anomalies, revolution, and paradigm change. This failure disturbed
Kuhn so much that he could not bring himself to discuss the problems raised by Forman’s
paper on the Princeton campus. He called a special evening seminar to talk about it at his
home. As students we spent an enjoyable evening debating whether and how Forman’s thesis
could be reconciled with Kuhn’s structure, but our arguments seemed to go nowhere.
Looking back on that evening, I now believe we came farther than we realized at the
time, for the debate sharpened everyone’s awareness of what Kuhn’s structure could not
explain. And as we went our separate ways in the years that followed, we all arrived at a
richer, if messier, sense of science history.
The idea that science studies must confront all the relevant factors, scientific or not, has
become so widely accepted in the history of science that hardly anyone in the field argues about
it now. But in the early 70s non-scientific factors posed a problem because the field was still
divided into “external” and “internal” history of science. Kuhn was considered an internalist, a
member of the professional sub-community that believed ideas are more important than mate-
rials, values, or social attributes. Focusing on the outside factors was somehow less reputable.
1
This project resulted in the Archives for the History of Quantum Physics available in Copenhagen, the American
Institute for Physics, earlier based in New York City and now in College Park, Maryland, and some other places.
I was among the many of Kuhn’s students who started out as internalists in the history of
science. I began my career in pursuit of cases on which I could impose the Kuhnian theoret-
ical structure. The picture seemed simply true, perhaps because it was so truly, beautifully
simple. One of my first historical papers dealt with the emergence of the quantum theory of
solids. As Kuhn had instructed, I read letters, scientific manuscripts, and published papers
from the period, in this case 1926–1932, in the languages in which they were written. Thanks
to Kuhn and his colleagues, many of the crucial archival documents could be found in the
Archives for the History of Quantum Mechanics, which I examined in Copenhagen in 1976
and later in New York City. In this history, I looked for, and found, paradigms, anomalies,
even revolutions. For example, Pauli’s brilliant attempt to deal with the experimental anom-
aly of a too weak paramagnetism in metals led directly to Sommerfeld’s transitional semi-
classical electron theory of metals, and in time to Felix Bloch’s fully quantum-mechanical
theory of electrons in metals. The Kuhnian structure worked here. My paper on the quantum
mechanical electron theory of metals, co-authored with Gordon Baym, concluded:
This development replayed the stages that the theory of the atom passed
through: a classical period in severe crisis shortly after the turn of the century,
a semi-classical period — the old quantum theory — during which non-clas-
sical assumptions, such as the Bohr–Sommerfeld quantization condition,
were included ad hoc in the classical framework, and the revolutionary mod-
ern period, heralded by Heisenberg’s first paper on quantum mechanics. One
may indeed characterize the development of the electron theory of metals as
a “secondary scientific revolution”.(Hoddeson & Baym, 1980, pp. 19–20)
What I gradually realized was that Kuhn’s structure fit this earliest phase in the history
of the quantum theory of solids because in those years, 1926–1932, the objects of this the-
ory were “ideal” solids, abstract models of metals and insulators.2 But as soon as physi-
cists tried to apply this theory to real materials (sodium, copper, tin, etc.), all sorts of messy
details muddied the formulation. The study of solids, as Pauli3 declared, became a “physics
of dirt”, a physics, he cautioned his students, “one should not wallow in”.
The problem was that each solid has its own multitude of real properties, and to incor-
porate all of them in the analysis would make the theory impossible to work with. An anal-
ogous dilemma was caused by the all-encompassing memory of the central character in the
Jorge Luis Borges (1962) story, “Funes the Memorious.” Because Funes remembered
every detail of everything he ever encountered, or even thought or dreamed of, his mem-
ory was, as he lamented, “like a garbage heap,” and this rendered him almost incapable of
thought (Borges, 1962, p. 64).
Just as we need to forget a good many details in order to think, the physicists who worked
on the problem of extending the theory of ideal solids to real materials were forced, from
about 1933, to make simplifying assumptions that would reduce the clutter of what they
2
One could make an analogous statement about why Kuhn’s theory fit the early stages of the development of
quantum mechanics itself.
3
Letter to R. Peierls (July 1, 1931), cited in Hoddeson, Braun, Teichmann, and Weart (1992, p. 181, 458).
30 Lillian Hoddeson
knew and thus render it useful. And so this field became characterized by a huge number of
approximations, which Pauli viewed as ugly. Many in his generation agreed that studying
real solids made physics messy, just as studying how science really works has made the his-
tory of science messy. Fortunately for the progress of solid-state physics, others in the
physics community (people like John Slater, John Bardeen, Frederick Seitz, Conyers
Herring, and Nevill Mott) found the messy properties of real solids, including the approxi-
mations used in representing them, to be beautiful because they belonged to real, rather than
ideal, solids. This group of physicists formed the first generation of physicists to refer to
themselves as solid-state physicists. Studying their work occupied me for years (Hoddeson,
Braun, Teichmann, & Weart, 1992, pp. 88–235).
And just as a new generation of physicists was happy to work on the messy real solids,
in a new solid-state physics community, which then grew into the largest subfield of mod-
ern physics now called condensed matter physics (Weart, 1992), the generation of histori-
ans of science who worked after Kuhn mostly found the messy contingent factors of real
science practice rewarding to study. The price for their fuller engagement with the world
was, however, abandoning the elegant Kuhnian account of successive paradigms.
To follow my path in the history of science just a bit farther, I found it increasingly nec-
essary to deal with the contingencies of real-world factors in my own work. I wrote about
all kinds of influences that affected scientific change without the apparent guidance of any
grand narrative — wars, patent fights, teachers, schools, families, apparatus, materials, and
places. For example, when I explored the implications of the quantum theory of solids for
the research at Bell Laboratories, I realized that this theory was important not only in con-
ceptually advancing the physics, but in helping to transform the environment into one
where the transistor could be invented. This happened because Bell Labs recognized its
need for physicists trained in quantum mechanics, but to keep such physicists happily
employed — people like William Shockley and John Bardeen — the laboratory had to
offer them a considerable research freedom, at least comparable to that offered by univer-
sities. This social condition was as important to the invention of the transistor as the
ghostly new paradigm of empty spaces (holes) bearing positive charge (Hoddeson, 1981a).
In the transistor story, too, contingencies played a large part. The physics of semicon-
ductors alone would probably not have led to the invention had not the United States gov-
ernment invested lavishly in the development of radar during World War II. This investment
gave rise to contributions to the physics of silicon and germanium, which were necessary
for the transistor’s invention. This wartime radar work also created an interacting network
of groups from different industries and universities who were all in touch with the latest
findings about semiconductors and electronics. A major Bell Labs administrator (Mervin
Kelly) recognized by 1943 that these groups would surely face off against each other after
the war in fierce commercial competition. As one line of defense, Bell Labs organized its
semiconductor research group, which in time invented the transistor (Hoddeson, 1981b;
Riordan & Hoddeson, 1997).
The fact that Kuhn’s Structure does not account for the material base of science has
been discussed by Andy Pickering, who explains in one example how the different appa-
ratus used by William Fairbank and Giacomo Morpurgo in their attempts to explain the
existence or non-existence of free quarks is what was really behind the Kuhnian incom-
mensurability, the “different worlds” separating their explanations, and more generally
between the pre-standard models pictures of the mid-1970s and the later gauge theory.
Attempting to find “a fault line somewhere in the mid-1970s”, Pickering realized:
The incommensurability of the old and the new physics did not seem to me
much like incompatible perceptions of those funny gestalt figures: the duck
and the rabbit, and so on. This incommensurability seemed to me rather to
have, once more, an important material dimension. The theories and models of
the old physics hung together with specific forms of apparatus and the specific
material phenomena they displayed, the same went for the new physics, and
these domains of instrumentation and phenomena were largely disjoint.
Materially, one did the old physics differently, eliciting quite different material
performances, from the new physics, and vice versa. (Pickering, 2001, p. 502)
A similar case where material factors and social contingencies impart a Kuhn-like struc-
ture to a scientific development can be found in the massive effort (involving every major
theoretical physicist working between 1911 and 1957) to create a first-principles theory of
superconductivity. Developing the new paradigm had as much to do with post-war experi-
ments using pure mercury isotopes, available as a result of the Manhattan Project, as it had
with anomalies, crises, and paradigm shifts. As critical, perhaps, was the fact that Bardeen
left Bell Labs in 1951 because he no longer experienced experimental freedom and for this
reason joined a more supportive group at the University of Illinois built by Bardeen’s grad-
uate school friend Frederick Seitz (Hoddeson & Daitch, 2002, pp. 314–329).
The real problem with Kuhn’s theory of scientific revolutions is that it just is not legiti-
mate to extract a vital artery out of the body of science and designate it as the essence of the
whole business. The kinds of questions about theory change that Kuhn was drawn to are
inevitably tangled up with questions that require drawing on the full range of factors that are
at play in any human enterprise. The building of the atomic bomb at Los Alamos, which cre-
ated all sorts of new scientific fields and problems, had as much to do with our fear of Hitler,
the concerns of Jewish refugee scientists, and the material properties of amassing plutonium
and uranium, as it had to do with developing and applying the underlying physics of nuclear
fission (Hoddeson, Henriksen, Meade, & Westfall, 1993). And a solid explanation of the new
theoretical structures emerging from the mushrooming field of particle physics after World
War II has to be related to United States science policy during the Cold War, to the wartime
invention of radar resonance techniques, to the expansion of graduate programs in physics,
and to the development of new materials and technologies (Brown & Hoddeson, 1983).
One implication of taking a more holistic look at scientific development and conceptual
change is the dazzling variety of new sources we need to engage with. John Heilbron,
another of Kuhn’s early Ph.D.s who worked with Kuhn on the Quantum Physics Project,
suggests the rewards of drawing on the full range of sources:
The industrious historian who has looked at private correspondence, gov-

ernment papers, foundation and university reports, newspapers, patent
applications, court records, architectural monuments, scientific apparatus,
motion pictures, painted neckties, and literary T-shirts — as well as the
scientific literature — necessarily sees connections that the people being
studied could not have known. (Heilbron, 1989, p. 48)
32 Lillian Hoddeson
To some leading historians of science, the limitations of Kuhn’s Structure appear tied
with the simple fact that even after Kuhn switched fields, he continued to think like a
physicist. Martin J. Klein, another physicist-turned-historian of physics, who often argued
with his contemporary Thomas Kuhn, was one of the earliest to observe that it is
characteristic of the physicist to want to get at the very essence of a phe-

nomenon, to strip away all the complicating features and see as clearly and
directly as he can just what is really involved. (Klein, 1972, p. 16)
That is why, Klein explained (before most of us were ready to listen), physicists prize
simple conceptual models while historians are drawn instead to the richness of particular
moments. Klein reflected, possibly with Kuhn in mind, that this is why physicists who are
interested in history often
expect to be able to treat historical problems the same way — it is not sur-
prising, but it is not justified.(Klein, 1972, p. 16).
To those who regret the loss of simplicity and generalizability that has accompanied the
developments that came in the wake of Kuhn, it may seem that historians of science now
have no theory to replace his structure, only a welter of particulars. I think we have, if not
a theory, at least a set of tacit agreements that enable our research. To me the most prom-
ising is the view that has made the old division between internal and external accounts
obsolete, a view that considers all factors, whether ideas or materials, individuals or insti-
tutions, as alike in being potential actors in the stories we tell (e.g., Latour & Woolgar,
1986; Pickering, 1995). Instead of the traditional hierarchy in which ideas rule, we try to
envision a democracy of all the actors, where anyone can in principle play a leading role,
and where which one dominates depends both on the circumstances and how we choose to
tell the story. In that vision, conceptual change becomes just one strand in the web, one
dimension of social and cultural history.
Acknowledgment
I thank Peter Garrett, who contributed greatly to this paper.
References
Borges, J. L. (1962). Funes the memorious. In: D. Yates, & J. Irby (Eds), Labyrinths: Selected
stories and other writings (pp. 59–66). New York: New Directions Publishing Corporation.
Brown, L. M., & Hoddeson, L. (Eds). (1983). The birth of particle physics: Particle physics in the
1930s and 40s. New York: Cambridge University Press.
Forman, P. (1971). Weimar culture, causality and quantum theory, 1918–1927: Adaptation by
German physicists and mathematicians to a hostile intellectual environment. Historical Studies in
the Physical Sciences, 3, 1–115.
Heilbron, J. L. (1989). An historian’s interest in particle physics. In: L. M., Brown, M., Dresden, &
L. Hoddeson (Eds), Pions to quarks: Particle physics in the 1950s (pp. 47–54). New York:
Cambridge University Press.
Hoddeson, L. (1981a). The entry of the quantum theory of solids into bell telephone laboratories,
1925–40: A case-study of the industrial application of fundamental science. Minerva, 18(3),
422–447.
Hoddeson, L. (1981b). The discovery of the point-contact transistor. Historical Studies in the
Physical Sciences, 12(1), 41–76.
Hoddeson, L. H., & Baym, G. (1980). The development of the quantum mechanical electron theory
of metals: 1900–28. Proceedings of the Royal Society of London A, 371, 8–22.
Hoddeson, L., & Daitch, V. (2002). True genius: The life and science of John Bardeen. Washington,
DC: Joseph Henry Press.
Hoddeson, L., Braun, E., Teichmann, J., & Weart, S. (Eds). (1992). Out of the crystal maze: Chapters
from the history of solid-state physics. New York: Oxford University Press.
Hoddeson, L., Henriksen, P., Meade, R., & Westfall, C. (Eds). (1993). Critical assembly: A history
of Los Alamos during the oppenheimer years, 1943–1945. New York: Cambridge University Press.
Keller, E. F. (1998). Kuhn, feminism, and science? Configurations, 16, 15–19.
Klein, M. J. (1972). The use and abuse of historical teaching in physics. In: S. G. Brush, & A. L King
(Eds), History in the teaching of physics (pp. 12–27). Hanover, New Hampshire: University Press
of New England.
Kuhn, T. S. (1959). The copernican revolution: Planetary astronomy in the development of western
thought (2nd ed.). New York: Vintage Books.
Kuhn, T. S. (1962). The structure of scientific revolutions. Chicago: The University of Chicago Press.
Latour, B., & Woolgar, S. (1986). Laboratory life: The construction of scientific facts (2nd ed.).
Princeton: Princeton University Press.
Pickering, A. (1995). The mangle of practice: Time, agency, and science. Chicago: The University
of Chicago Press.
Pickering, A. (2001). Reading the structure. Perspectives on Science, 9(4), 499–510.
Riordan, M., & Hoddeson, L. (1997). Crystal fire: The birth of the information age. New York:
W.W. Norton.
Rouse, J. (1998). Kuhn and scientific practice. Configurations, 16, 33–50.
Vosniadou, S., & Brewer, W. (1992). Mental models of the earth: A study of conceptual change in
Weart, S. (1992). The solid community. In: L. Hoddeson, E. Braun, J. Teichmann, & S. Weart (Eds),
Out of the crystal maze: Chapters from the history of solid-state physics (pp. 617–69). Oxford and
New York: Oxford University Press.
Chapter 4
Kuhn’s Philosophical Successes?

Peter Machamer
Introduction
Thomas Kuhn’s The Structure of Scientific Revolutions was published in 1962 (Kuhn,
1962). Since its publication it has sold about 20 million copies and has been translated into
18 languages.1 This probably makes it the best-selling book on history and philosophy of
science ever written. Yet, at the time of its publication, most philosophers hated the book,
much to Kuhn’s dismay, and many social scientists loved it, also to his great dismay.
In this chapter, I will assay some of Kuhn’s philosophical claims and critically assess
what value remains. That it has been widely influential is beyond dispute. Consider a fairly
recent issue of Science magazine, where the 2000 Nobel Prize winner, Arvid Carlsson
(Gothenburg, Sweden) entitled his Nobel address “A Paradigm Shift in Brain Research”
(Carlsson, 2001). Therein he recounted the change in basic thinking about the brain:
As late as the 1950s, it was assumed that communication between nerve

cells in the brain occurred by electrical impulses. A decade later, the theory
of chemical transmission, which until then had been thought to occur only
in the peripheral nervous system, had gained strong entrance for the central
nervous system. This paradigm shift opened up an enormous new perspec-
tive in brain research, not the least by facilitating the study of brain func-
tion by means of chemical tools. Which in different ways could modify
chemical signaling between nerve cells. Moreover, such tools sometimes
turned out to be useful as therapeutic agents. Thus, for the first time, a vari-
ety of disorders in the central nervous system could be treated effectively.
You may recall that Carlsson began his work by finding out that LSD worked on serotonin
receptors and went on to work on dopamine, which led to the now standard treatment for
Parkinson’s disease. But the point I wish to emphasize is that Kuhn’s “paradigm shift” has
made it into the annals of the Nobel Prize winners’ account of their work.
1
These figures come from Michael Matthews in a lecture he gave at IHPST in Denver, Nov. 9, 2001.

ISBN: 0-08-045355-4
36 Peter Machamer
This beginning rumination takes us into the heart of our topic today: what were Kuhn’s
views on the nature of science, and how have they held up almost 50 years of scrutiny?
The Nature of Kuhnian Science and Some Problems
The topics I want to consider include the major philosophical concepts that have passed
down into contemporary usage and which often are attributed to Kuhn’s (1962) Structure:
Normal science, puzzle solving, paradigm, theory-ladenness and, of course, incommensu-
rability.
Science, Kuhn held, for the most part was normal science, as practiced by people,
scientists, who had been specially trained in their field to use certain concepts and practices
in order to solve scientific puzzles. So, normal science was puzzle solving. The puzzles were
set by paradigms, which were basically nonpropositional sets of interconnected concepts
and associated practices for applying those concepts. Paradigms directed the practicing sci-
entists as to what things to look at (and how to conceptualize their looking) and what were
the mathematical, experimental and technical procedures allowable for such explorations
(e.g., what kind of equations could be used to solve which kind of problems, and what
instruments would allow the scientists to measure or observe the right objects or quantities).
Puzzle solving was working out the details and refining of the paradigm.
Paradigm was what was learned by students of science as they were trained to be pro-
fessionals in their field. They learned the proper mathematics by problem-solving exer-
cises (exemplars) in textbooks, the proper use of instruments by apprenticeship in
laboratories, and the vocabulary and preferred forms of expression by writing research
reports for professional journals. It was such common training and learning by groups of
people that created a scientific community of shared beliefs and practices. This emphasis
on the scientific community and their group commitments, consensually shared examples,
and tacit knowledge — as he called them in the 1969 Postscript to the Second Edition
(Kuhn, 1970, pp. 174–210) — was one of Kuhn’s most effective lessons, not for philoso-
phers but for historians of science, sociologists of science and science and technology
studies (STS) people. The themes of science as a social activity and scientific practice are
a major part of Kuhn’s legacy, but I shall not focus on them in this talk.
Kuhn failed to note an interesting contrast when he introduced the idea of puzzle solv-
ing. Puzzle solving much better characterizes much of Medieval and Renaissance “sci-
ence”, but there is a difference. A typical Medieval Quaestiones was nothing but a set of
puzzles.2 They were compendia of various questions and an author’s proffered answers.
Questions would be such as: Why do lions roar? Why do bodies fall? Can there be essen-
tial accidents? Is it possible that there are many worlds? Answers to each question would
be given. But most often there were no overarching set of principles that would be invoked
to make the coherence among the set of questions, neither need there be consistency
2
I remember David Lindberg pointing this out to me in the late 60s. We were discussing optics texts and how
various Medieval writers tried to answer optical questions. Even within a closely constrained topic there was still
no desire for consistency among answers.
Kuhn’s Philosophical Successes? 37
among the answers. So, in Kuhn’s terms there was no guiding paradigm. Each answer was
ad hoc to a specific question. To be sure they shared some general philosophical presup-
positions, basic Latinized Aristotle, but this is far from having a shared paradigm. More
importantly there was no attempt at using a few “fundamental” principles and techniques
to assure consistency among the various answers. Contradictory principles could be used
among questions and sometimes even within discussions of the same question.
Presumably, Kuhnian paradigms would not allow such ad hoc answers, except in para-
digm-saving cases (as Lakatos, 1971 emphasized). Paradigms, by providing constraints
that must be satisfied by all answers, were intended to ensure that the answers have a
mutual consistency. So normal science is not just puzzle solving, it is principled puzzle
solving in that it presupposes some fundamental principles (and practices).
Yet, as noted above, a paradigm is more than a theory conceived as a set principles or
laws stated in the form of a set of propositions, whether axiomatic or not. Paradigms were
conceived as “vehicles” for theories, and theories were taken to provide the basic ontic
commitments and behaviors of the entities of a given science. This point was picked up
and developed by Laudan (1977) in his distinctions between methodological, conceptual
and ontological commitments. Paradigms also were said to be like maps and contain the
principles for orienting oneself in problem space, and so repeated what Toulmin (1953)
had said earlier.
The concept of a paradigm is no doubt the best-known legacy of Kuhn’s Structure. It
also has been one of its most controversial ideas. When Shapere (1964) reviewed
Structure, he spent most of his 11-page review laying out the ambiguities, inconsistencies
and radical equivocations in meaning of this central term “paradigm”. Later, at a confer-
ence at the Bedford College, London in July 1965, Margaret Masterman (1970) carefully
distinguished among 21 senses of the term “paradigm” and then tried to distill them all
down into four basic senses.3 Kuhn (2000b) later would comment she was right.
Even more telling is Kuhn’s remark: “Paradigm was a perfectly good word, until I
messed it up” (2000b, p. 198). Indeed, it was; for example, “experimental paradigm”
meant (and in some circles still means) an experimental design that is a field’s preferred
set up used to investigate a type of problem or phenomenon. Yet this mess was just what
he intended. He wanted a word that would be broader and more multifaceted than
“theory”, especially where “theory” meant something like a set of law-like propositional
axioms. There is no doubt that Kuhn was doing a good thing in trying to shift the locus of
philosophical analysis away from axiomatic theory.
The axiomatic idealization of the concept of a scientific theory usually traced its roots
to Euclid and Newton. Axioms were sometimes lauded by some logical positivists (e.g.,
Carnap, 1939 and Reichenbach, 1958) and later by the logical empiricists and their sym-
pathizers (including Braithwaite, 1953 and Nagel, 1961). Yet, for the most part, axiomati-
zation had no place in the history or practice of actual science. Whoever heard of axiomatic
organic chemistry or axiomatic planetary astronomy (though the latter certainly did use
geometrical methods)? What I find strange is not that Kuhn wanted to liberate philosophy
of science from assuming that theories could be conceived best as systems of axioms, but
3
Shapere (1971) reviewed the 2nd edition of Kuhn’s Structure and the Criticism volume.
38 Peter Machamer
that any previous philosopher should have thought of the axioms as being the ideal form
for scientific theories. Certainly around the turn of the 20th century the attempt of Hertz
in 1894 to axiomatize mechanics (Hertz, 1956), the influence of Hilbert’s (1923) formal-
ist mathematics and the impact of Russell and Whitehead’s (1910–1913) formal logic had
their role in creating this ideal. But in fact, Newtonian or classical mechanics was not rig-
orously axiomatized until 1953 (McKinsey, Sugar, & Suppes, 1953); which was after the
time that philosophers began to debate whether Newtonian mechanics was still really true
or useful in view of its being reduced to or replaced by Einstein’s theory. In fact it is plau-
sibly arguable that no other science has ever been axiomatized in anything approaching a
rigorous way. So this being an ideal that needed attacking becomes stranger.
In the same period, some years before Kuhn, Toulmin4 (1953) and Hanson (1958) brought
the ideas of the late Wittgenstein into the philosophy of science.5 This questioned the idea of
necessary and sufficient conditions as being the goal of analysis, while around the same time
Quine (1951) attacked the analytic–synthetic distinction and later advocated a naturalized
epistemology (Quine, 1959, pp. 68–90). Toulmin (1953), as we noted, held that theories were
like maps, and Hanson (1958) proclaimed that theories were ways of organizing inquires into
the causes by which things happened. So Kuhn publishing 4 years after Hanson was only part
of the new wave in the philosophy of science. This new wave hoped to crest and crush what
were taken to be the bad, old positivist traditions.6 These were heady days for those of us
being trained during this time. But already the straw man, The Positivist, was beginning to
emerge as the archetypal, tradition-bound, moribund bad guy.
Yet Kuhn was certainly right in redrawing the parameters by which we ought to con-
ceive science and in prodding us to move beyond theory, especially when conceived as sets
of propositions. The shift in focus was to a “larger structure”, to what Lakatos (Lakatos
and Musgrave 1970) later would call a “research programme”, and Laudan (1977) a
“research tradition”. Paradigms extended over years, decades or even centuries, contained
many nonpropositional elements, and were subject to some internal changes even while
maintaining their identity as a specific paradigm. They were much more than just sets of
propositions from which a scientist might draw inferences. However, this led to a large
controversy over whether there were any coherent identity conditions for something’s
being a paradigm. What were the criteria? Kuhn (1970) later developed the idea by talk-
ing about a disciplinary matrix that included “constellations of group commitments”,
“shared examples” [“exemplars”] and tacit or intuitive knowledge about how to proceed in
the doing of the science (including how to give proofs). This helped somewhat and again
may be seen as right headed, but it failed to satisfy philosophically strict analytic require-
ments. Later Lakatos’ “hard and soft core” and Laudan’s “problems” would share similar
identification difficulties. These were articulated in ways different than Kuhn but both
4
Toulmin’s (1953) Philosophy of Science was published the same year as the first edition of Wittgenstein’s
(1953) Philosophical Investigations and in the same year came Quine’s (1953) From a Logical Point of View.
5
Mention should be made here too of Paul Feyerabend (1958). For a review see Machamer (1975).
6
That these bad old traditions were not so bad, and that the “new wave” was not so new and shared much with
the bad old has been documented to a great length. See Alberto Coffa (1991) and Michael Friedman (1999).
models, deep down, were completely intellectual, as opposed to Kuhn’s right-thinking

emphasis on knowing as being tied to social acting and doing. Today some philosophers,
like myself, might claim that science is, for the most part, a search for mechanisms and for
developing research techniques by which mechanisms may be discovered and validated,
wheremechanisms are socially and historically delineated and investigated. This search
involves not just thoughts, but centrally experiments, procedures, productions and many
types of tools.
Let me stop Kuhn exposition at this point for a little philosophical excursus. I just
claimed Kuhn was right to tie knowing with acting and doing. This claim is not just a
pragmatist bias of mine. I would argue that tying knowledge to action goes back to
Aristotle (1987a,b). Briefly, the claim is this; knowledge is not just a passive collection
of sensory inputs that are then inductively worked over by the mind to provide general
concepts. Positively, coming to have knowledge, or knowledge acquisition, is an active
learning process on the part of the knower in the form of critical and selective integra-
tions into multiple memory systems. Another distinct aspect of the activity comes in the
recall and accessing stage when putative knowers act on the basis of their memorial
(sometimes, belief) schemata and receive feedback from the environment that selectively
corrects the internal memorial or schematic representations that constitute the knowledge
(cf. Piaget’s accommodation and assimilation). This corrective function is necessary for
intelligence (what one might call knowing well). The difference between intelligent per-
formance and merely mechanical performance is that intelligent performance has the pos-
sibility of self-correction in virtue of being able to critically monitor its own states.
Simple corrective feedback is not sufficient; a homing torpedo or a thermostat can do this.
For intelligence, there must be a discrete internal representation of a goal state or an equi-
librium ‘setting’ a standard for proper functioning, which may be step-wise decomposed
and compared to (or regulated) and used to correct future actions (or outputs which them-
selves may be modified). Even in the case of procedural knowledge (pace some cognitive
neuroscience theories) this must be the case, for to intelligently perform a procedure one
must be able to monitor the procedural action (though this need not include an awareness
of the mechanisms themselves) and correct it if it goes wrong. This is what servomech-
anisms and many evolutionarily wired bits of “knowledge” cannot do. Elaboration of this
claim would require saying more about how goal states or norms get constructed, by evo-
lution in certain limiting cases but more often by social training and learning. But this
would take us too far beyond Kuhn and our topic.
Knowledge may be conceptual or declarative, it may be discriminatory and noncon-
ceptual, or it may be procedural. This last is often called knowing how, and it is this that
Kuhn attributes to and endorses in Polyani’s idea of tacit knowledge. One has to learn to
how to drive a car, properly taste a wine, develop strategies for solving jigsaw puzzles, or
differentiate two changing quantities. But once one has learned, using such knowledge
becomes, more or less, “automatic”. Yet in each case it is checkable and corrigible, that is
correctable if the activity seems to be going, or has gone, wrong. In fact even in the puta-
tively most passive visual cases, for learning to be successful and for knowledge to be
gained, there must be an active component. Further, a third aspect, must be present for
knowledge to be knowledge, it must be able to be used, at least sometimes (Machamer &
40 Peter Machamer
Osbeck, 2004). This use occurs in the public domain and entails that the criteria for knowl-
edge always have a social component.7 I might, only somewhat misleadingly, say that
activity is necessary to gain knowledge, to preserve knowledge in memory, to access
knowledge and to utilize knowledge. These reflections on activity and use should be a
guide to making clear the educator’s key phrase, “active learning”.
Let us shift our attention now to another Kuhnian conundrum. Like the logical empiri-
cists before him, Kuhn takes up the topic of the meaning of theoretical terms. The topic
arises during the discussion of what he takes to be the hallmark of scientific revolutions,
incommensurability. The stark form of Kuhn’s claim would be: If a scientific revolution
has occurred, then the meaning of the terms in the former paradigm are incommensurable
with (i.e., have a different meaning from) the terms in the new revolutionary paradigm. I
want to explore this claim about meaning for it was the cause of the most vigorous philo-
sophical controversy that surrounded Structure. But before doing so please note one prob-
lem: Kuhn, in discussing incommensurability, slipped back into talking, like the positivists
and logical empiricists, about terms and the meaning of terms. It seems we are no longer
concerned with practices or activities. So I leave it to you, when one paradigm supplants
another, do some or all practices change too? Does the way of solving differential equa-
tions change from Maxwell to Einstein? Did the basic way of using a microscope change
after Pasteur did his revolutionary work? Of course, the answer to both questions is
“Maybe yes, maybe no”.
Augmenting this problem, Kuhn went on to develop the concept of “incommensura-
bility” by conflating changes in a term’s meaning with differing theory-laden perceptions
and even with changes in world views. Somehow for him these were all the same phe-
nomenon, maybe just different aspects. But before criticizing Kuhn, let us note his influ-
ence. When the Max Plank Institute for the History of Science opened in 1989, it held a
conference of which the proceedings were later published as Biographies of Scientific
Objects (Daston, 1999). In this volume, Latour’s essay elaborates a “theory of ‘relative
existence’”:
When a phenomenon ‘definitively’ exists this does not mean that it exists
forever, or independently, of all practice and discipline, but that it has been
entrenched in a costly and massive institution that has to be monitored and
protected with great care…. (Latour, 1999, p. 255)
He says:
It is impossible to pronounce the sentence ‘Ramses II died of tuberculosis’

without bringing back all the pragmatic conditions that give truth to this
sentence. (Latour, 1999, p. 251)
7
It is this need for linking activity and knowledge that shows why one must read Aristotle’s remarks on epagoge
(induction) in Posterior Analytics II. 19 in conjunction with what he says about active knowledge in the Nicomachean
Ethics (for example at 1141B13ff.) This active link has been noted also by many others including the pragmatists
(especially Dewey, 1938); it was noted by Wittgenstein (1956) in Remarks on the Foundations of Mathematics and is
part of what Wittgenstein ought to have meant when he said, “look not for the meaning but for the use”.
Latour here might mean either the use of “tuberculosis” is anachronistic, or there are
only transitory ontic commitments at any time, no fact of the matter. Latour suggests we
only may talk about two different objects.
Pouchet and Pasteur do not define the same physical elements […] nor do
they live in the same social and historical context. Each chain of convic-
tions defines not only different links with the same elements, but different
elements as well. (Latour, 1999, p. 260)
This should sound familiar, for the same was said, about the Ptolemaic, Tychonic and
Copernican universes, about phlogiston and oxygen chemical paradigms and about many
other major changes in science. They are not just different explanations, they are different
universes, different scientific objects. Latour echoes what was claimed by Kuhn (1962) 30
years before and by Hanson (1958) 4 years before that: Plus ça change, plus la meme
chose. But is this claim any more plausible now with Latour than it was 40 years ago with
Hanson and Kuhn (and we should not forget Feyerabend, 1962)?
It is important to distinguish as Kuhn (2000a) later did, between “global” incommen-
surability and “local” incommensurability. The global is where one speaks about world-
views, weltaunshaungen and large-scale, social (cultural, economic political, ideological)
contexts. Local refers only to small, proscribed changes that occur within some, often ill-
defined, identity constraints, and such changes may occur even, for example., within an
individual’s concept as learning occurs. This later locution has become a preferred way of
speaking with some learning psychologists (e.g., Carey & Spelke, 1996) who describe how
children learn science, in terms of changing concepts that produce local incommensura-
bilities. For such descriptions we might as well have stayed with Festinger’s (1962) cog-
nitive dissonance theory.
Yet, surely the intuition behind local incommensurability is basically right. When anyone
learns anything, some of the content of what was believed before becomes otiose, undergoes
modification, and what results may well be in conflict with what had been before. So as a
child learns about gravity as a force, she must relinquish some of her animistic beliefs about
the cause of motion in inanimate bodies. More generally, of course, as language evolves
meanings change. During the course of a human life many metaphors turn literal, extend
their contexts, and domains of reference expand and contract. All these changes add new
implication patterns and so a form of incommensurability.
The incommensurability here is just a failure of one part of a concept to map fully onto
another. The mapping relates the properties of the entities or activities and the implica-
tions licensed by the concept’s network, schema, or script. This is not a troublesome
extension of the geometrical concept of “incommensurable”. The new does not fit onto
the old without remainder. The new concept will have some different properties or impli-
cations that cannot be held coherently together with the old. These may be in contradic-
tion (this was the way Baltas 1997, tried to handle incommensurability), or they may
introduce new entities and activities that “redefine” the mechanism. An example of the
first is allowing for a wave without requiring a medium, of the second modeling brain
mechanisms not as an electrical system but as a set of chemical activities. “Chemical”
refers not to different properties of the electrical objects and processes, but to different
42 Peter Machamer
objects or processes. Yet both purport to describe many of the same anatomical parts and
to explain the same phenomena, for example, memory or perception, by providing the
neurocognitive mechanisms (as measured or operationalized by behavioral tests) by
which the brain works. There is no problem of meaning or of translation in this latter type
of case. One cannot translate electrical talk into chemical talk, though one may relate
them in certain ways, for some chemical phenomena sometimes have or depend upon
electrical properties.
Global incommensurability is much more problematic. Surely Kuhn and Latour are
right in pointing out that there was no such concept as tuberculosis at the time of Ramses
II. So, if miraculously someone at that time had used that word, no historical actor would
have understood. Tuberculosis was not an actor’s category in Ramses’ time. Or take
Kuhn’s example: if an Aristotelian saw a constrained stone swinging on a string, he would
not have seen a pendulum. No such pendulum category existed at that time, though the
concepts of equal time and equal spaces did. But would this mean that we could not teach
the historical actor, in her different time and different culture, the “new” concept? Again
we must give the Greek answer: sometimes yes, sometimes no. One certainly could have
taught the concept of isochrony to Aristotle. He had all the conceptual resources he needed
to grasp this. What he did not have was a “context” that made time a more important vari-
able than distance. Why such a context came about in 1603–1604 for Galileo, and not
before, is a good historical question. It may be a form of the ultimate historical question:
Why did this event happen now, and not at some other time? But it raises no particular
philosophical difficulties except in terms of historical correlative causation.
Philosophical difficulties of the kind claimed above only arise when one combines
such historical claims with specific theories of meaning for example, meaning as intrin-
sically defined by interrelated concepts that all ‘in the head’. However, philosophically,
there is no reason to accept such accounts unless there would be some evidence that
makes them compelling, or, at least, plausible. The accounts assumed by Kuhn (and
Feyerabend and Latour) have no such warrant. Let us consider quickly the famous case
of the meaning of the term “mass”. In classical mechanics mass is conserved, in rela-
tivistic mechanics mass is not conserved (for it may be transformed into energy.)
Therefore, it is claimed the term “mass” means something different in the two theories
(in the two paradigms). Now there are, of course, differences. I have just noted one in the
two sentences above. And this difference is not badly thought of as a difference in the
meaning of the term. But to get incommensurability we also need to claim that there is
no overlap in the meaning of the terms used in the different theories. Yet, to draw this con-
clusion one must hold that the meaning of a term is wholly and only specified by the total
set of connections or implications that are allowable or licensed by the theory in which it
occurs. The totality of the set is constitutive of the meaning. Partial sets of implications
are not allowed.
Now first of all this theory takes meaning to be an all or nothing property (Shapere
1966. remarked on this some years back). A term either keeps it meaning constant or it
changes. Any change is total and complete or nonexistent. Worse, on this theory, meaning
resides wholly internally in the head and has nothing to do with the world. The meaning
of a term, as were, resides solely within the relations in a dictionary (or mental lexicon).
Katz and Fodor (1963) actually elaborated such a theory of meaning, and it later played a
part in Searle’s (1985) Chinese Room argument (of which Casimir Lewy of Cambridge
had an earlier version). This is a preposterous theory of meaning. It is meaning without the
world. Such a theory of meaning leaves out procedures of reference as having anything to
do with the meaning of a term, which is most strange.
What are the cognitive presuppositions of such a theory of meaning and “seeing” or
conceptualizing? Simply put the cognitive theory assumed here is one where meaning is
completely internal and modularly closed (like Katz and Fodor’s theory, of which Fodor
still held a version as late as 1987). Now there is no doubt that conceptual schemes or cat-
egorical structures direct perceptual attention, and further there is no doubt that concep-
tual networks may lead to reasoning that further directs attention toward represented
features. But this is far from claming that conceptual schemes are isolated cognitively,
have no external relations to the world, or are not, by virtue of their parts, semimappable
on the schemes of other people. Kuhn should have known of this nonindependence of
meaning from Bruner & Goodman (1947) and Bruner & Postman (1949) work that had
shown how value and need (as measured by external variables) affect perception and
meaning, and how subjects can learn to discriminate and recognize incongruous suits and
colors of playing cards when given external feedback. Both of these experiments implied
the inadequacy of modularity and internalism. Further, for Kuhn, meanings must be
shared by those in a paradigm and so cannot be wholly individual. A theory of the global
incommensurability of meaning entails a relativistic solipsism should be taken as a reduc-
tio ad absurdum argument against any theory that implies it. Kuhn later recognized this
as an implication, which is why he changed his mind and reduced his claim to local mean-
ing incommensurability (Kuhn 2000a, 91 ff.).
That Kuhn changed his mind about many things is to his great credit. He came to see
many of the problems discussed above and made moves in his later works to try to address
and correct them. Nonetheless, many problems remain. But the biggest by far is how could
he reconcile the philosophical changes he later saw as necessary with his view of the his-
tory of science as proceeding by way of incommensurable revolutions among paradigms?
If this latter cannot be defended, what point could there be in trying to modify the under-
lying philosophy to make it more plausible? More generally, what point could there be in
talking about Kuhnian revolutions at all?
Acknowledgements
This chapter is based on a lecture delivered at University of La Coruna, in Spain in March,

2002, and published as “El éxito de Kuhn, 40 años después”, in Gonzalez, W. J. (Ed.),
Análisis de Thomas Kuhn: Las revoluciones científicas, Madrid: Trotta, 2004. Thirty years
earlier, I took a very different line on Kuhn; see my “Understanding Scientific Change”
(1975). I most happily thank Wenceslao Gonzalez for his help and able assistance on the
earlier version of this chapter. Further, discussions over the years with my dear old friends
and fellow curmudgeons, Aristides Baltas and J. E. McGuire, and my delightful and
constant co-author Lisa Osbeck, have always been beneficial, when they have not led
me astray by their wild ways. No doubt they have influenced me in ways that I dare not
even recall.
44 Peter Machamer
References
Aristotle (1987a). Nicomachean ethics. In: J. L. Ackrill (Ed.), A new Aristotle reader. Princeton
University Press.
Aristotle (1987b). Posterior analytics. In: J. L. Ackrill (Ed.), A new Aristotle reader. Princeton
University Press.
Baltas, A. (1997). Some grammatical points about incommensurability. Paper presented at Kuhn
Memorial Symposium, MIT.
Braithwaite, R. (1953). Scientific explanation: A study of the function of theory, probability and law
in science. London: Cambridge University Press.
Bruner, J. S., & Goodman, C. D. (1947). Value and need as organizing factors in perception. Journal
of Abnormal Social Psychology, 42, 33–44.
Bruner, J. S., & Postman, L. (1949). On the perception of incongruity: A paradigm. Journal of
Personality, 28, 206–223.
Carey, S., & Spelke, E. (1996). Science and core knowledge. Philosophy of Science, 63, 515–533.
Carlsson, A. (2001). A paradigm shift in brain research. Science, 294(2), 1021–1024.
Carnap, R. (1939). Foundations of logic and mathematics. In: O. Neurath, C. Morris, & R. Carnap
(Eds), International encyclopedia of unified science (Vol. 1(3), pp. 139–214). Chicago, IL:
University of Chicago Press.
Coffa, A. (1991). The semantic tradition from Kant to Carnap. Cambridge: Cambridge University
Press.
Daston, L. (Ed.). (1999). Biographies of scientific objects. Chicago, IL: University of Chicago Press.
Dewey, J. (1938). Experience and education. London: Macmillan.
Festinger, L. (1962). A theory of cognitive dissonance. Stanford, CA: Stanford University Press.
Feyerabend, P. (1958). An attempt at a realistic interpretation of experience. Proceedings of the
Aristotelian Society, 58, 143–170.
Feyerabend, P. (1962). Explanation, reduction and empiricism. In: H. Feigl, & G. Maxwell (Eds),
Scientific explanation, space and time (pp. 28–97). Minnesota Studies in the Philosophy of
Science (3). Minneapolis, MN: University of Minnesota Press.
Fodor, J. A. (1987). Psychosemantics. Cambridge, MA: MIT Press.
Friedman, M. (1999). Reconsidering logical positivism. Cambridge: Cambridge University Press.
Hanson, N. R. (1958). Patterns of discovery: An inquiry into the conceptual foundations of science.
Cambridge: Cambridge University Press.
Hertz, H. (1956). The principles of mechanics, presented in a new form. In: D. E. Jones, & J. T. Walley
(Trans.). New York: Dover Publications (Original work published 1894).
Hilbert, D. (1923). Die Logische Grundlagen der Mathematik. Mathematiche Annalen, 88, 151–165.
Katz, J., & Fodor, J. (1963). The structure of a semantic theory. Language, 39, 170–210.
Kuhn, T. S. (1962). The structure of scientific revolutions. Chicago, IL: University of Chicago Press.
Kuhn, T. S. (1970). The structure of scientific revolutions (2nd ed.). Chicago, IL: University of
Chicago Press.
Kuhn, T. S. (2000a). The road since structure. In: J. Conant, & J. Haugeland (Eds), The road since
structure: Philosophical essays, 1970–1993, with an autobiographical interview (pp. 90–104).
Chicago, IL: University of Chicago Press.
Kuhn, T. S. (2000b). A discussion with Thomas S. Kuhn. In: J. Conant, & J. Haugeland (Eds),
The road since structure: Philosophical essays, 1970–1993, with an autobiographical interview
(pp. 253–324). Chicago, IL: University of Chicago Press.
Lakatos, I. (1971). History of science and its rational reconstruction. In: R. C. Buck, & S. Cohen
(Eds), In memory of Rudolf Carnap: Proceedings of the 1970 second biennial meeting of the
Philosophy of Science Association (pp. 91–135). Boston Studies in the Philosophy of Science, 8.
Dordrecht, The Netherlands: Reidel.
Lakatos, I., & Musgrave, A. (Eds). (1970). Criticism and the growth of knowledge. Cambridge:
University of Cambridge Press.
Latour, B. (1999). On the partial existence of existing and nonexisting objects. In: L. Daston (Ed.),
Biographies of scientific objects. Cambridge: University of Chicago Press.
Laudan, L. (1977). Progress and its problems. Berkeley: University of California Press.
Machamer, P. (1975). Understanding scientific change. Studies in History and Philosophy of
Science, 5, 373–381.
Machamer, P. (2004). El éxito de Kuhn, 40 años después. In: W. J. Gonzalez (Ed.), Análisis de
Thomas Kuhn: Las Revoluciones Científicas (pp. 137–153). Madrid: Trotta (in Spanish).
Machamer, P., & Osbeck, L. (2003). Scientific normativity as non-epistemic: A hidden Kuhnian
legacy. Social Epistemology, 17, 3–12.
Machamer, P., & Osbeck, L. (2004). The social in the epistemic. In: P. Machamer, & G. Wolters
(Eds), Science, values and objectivity (pp. 78–89). Pittsburgh, PA: University of Pittsburgh Press.
Masterman, M. (1970). The nature of a paradigm. In: I. Lakatos, & A. Musgrave (Eds), Criticism
and the growth of knowledge (pp. 59–90). Cambridge: Cambridge University Press.
McKinsey, J. C. C., Sugar, A. C., & Suppes, P. (1953). Axiomatic foundations of classical particle
mechanics. Journal of Rational Mechanics and Analysis, 2, 253–272.
Nagel, E. (1961). The structure of science. New York: Harcourt, Brace and Co.
Quine, W. v. O. (1951). Two dogmas of empiricism. Philosophical Review, 60, 20–43.
Quine, W. v. O. (1953). From a logical point of view. Cambridge, MA: Harvard University Press.
Quine, W. v. O. (1959). Ontological relativity and other essays. New York: Columbia University
Press.
Reichenbach, H. (1958). The rise of scientific philosophy. Berkeley, CA: University of California
Press.
Russell, B., & Whitehead, A. N. (1910–1913). Principia mathematica (Vol. 3). Cambridge:
University of Cambridge Press.
Searle, J. (1985). Minds, brains and science. Cambridge, MA: Harvard University Press.
Shapere, D. (1964). The structure of scientific revolutions. Philosophical Review, 73, 383–394.
Shapere, D. (1966). Meaning and scientific change. In: R. G. Colodny (Ed.), Mind and cosmos
(pp. 41–85). Pittsburgh, PA: University of Pittsburgh Press.
Shapere, D. (1971). The paradigm concept. Science, 172, 706–709.
Toulmin, S. E. (1953). The philosophy of science. An introduction. London: Hutchinson’s University
Library.
Wittgenstein, L. (1953). Philosophical investigations (G. E. M. Anscobe, Trans.). Oxford: B. Blackwell.
Wittgenstein, L. (1956). Remarks on the foundations of mathematics (G. E. M. Anscobe, Trans.).
Oxford: B. Blackwell.
Chapter 5
Conceptual Change and Scientific Realism:

Facing Kuhn’s Challenge
Theodore Arabatzis
Introduction
The impact of Thomas Kuhn’s work on history and philosophy of science has been unpar-
alleled. His classic The Structure of Scientific Revolutions (Kuhn, 1970) has changed per-
manently the historiographical and philosophical landscape. Its influence on historians of
science has been pervasive but, for the most part, indirect. It is a striking fact that very few
historians have employed the conceptual apparatus of The Structure (i.e., “paradigm”,
“normal science”, “crisis”, “revolution”, “gestalt switch”, “incommensurability”) to illumi-
nate past scientific developments. Philosophers of science, on the other hand, have been
directly affected by Kuhn’s historical philosophy of science and have debated endlessly the
issues that he and his fellow traveler Paul Feyerabend raised. Those issues (e.g., the ration-
ality of theory-choice, the reality of the ontology of science) remain at the forefront of con-
temporary philosophical reflection on science.
One of the most far-reaching claims of Kuhn and Feyerabend concerned the nature of
scientific concepts. They promoted a contextual view of concepts, according to which con-
cepts obtain their meaning from the theoretical framework in which they are embedded. It
follows that when a theoretical framework changes the concepts embedded in it change too
(Feyerabend, 1962; Kuhn, 1970). Given that theory change has been quite common in his-
tory of science, conceptual change must have been a ubiquitous phenomenon. Several
problematic philosophical consequences seemed to ensue from conceptual change, most
notably a relativist view of theory-choice and an anti-realist stance toward the ontology of
science. It was widely believed that rational choice between scientific theories could take
place only against a stable, shared conceptual framework. With such a framework in place
the claims of competing scientific theories could be formulated in a common language and
then subjected to comparative evaluation. On the other hand, in the absence of such a
framework objective theory-choice seemed impossible, since there would be no common
language for formulating and comparing rival scientific theories. Furthermore, the realist
requirement for a stable scientific ontology was hard to reconcile with conceptual change.

ISBN: 0-08-045355-4
48 Theodore Arabatzis
If scientific concepts change, then it is not clear how they can continue to refer to the same
entities, properties, or processes.
This chapter focuses on the latter issue, namely the implications of conceptual change
for scientific realism.1 The debate on scientific realism has taken place along several lines.2
The most important, for my purposes here, concerns the grounds that we have for believ-
ing in the reality of the unobservable entities postulated by contemporary science (atoms,
electrons, photons, fields, etc.). These entities are represented by theoretical concepts. If
these concepts are evolving, then what sense does it make to believe that they continue to
refer to the same, and ipso facto real, entities?3 In what follows, I will discuss some of the
main developments concerning this question in post-positivist philosophy of science. I will
then argue that the evolution of theoretical concepts need not throw doubt, under certain
conditions, on their stable identity. When theoretical concepts change they may continue
to refer to the same entities, properties, or processes.
Historicizing Concepts: Kuhn’s and Feyerabend’s Anti-Realist Theses
As I noted above, Kuhn and Feyerabend launched a novel and powerful attack against a real-
ist reading of the historical development of science, that is, the view that science aims at
a permanent fixed truth, of which each stage in the development of scien-

tific knowledge is a better exemplar. (Kuhn, 1970, p. 173)
On this view,
the progress of science … consist[s] in ever closer specification of a single

world, the actual or real one. (Kuhn, 1989, p. 24)
Here I will focus on Feyerabend’s views, since they were more radical and developed
in more philosophical detail than Kuhn’s early views.4
Feyerabend (1962) presented his views on conceptual change in a seminal paper. There
he argued that the concept associated with a scientific term
is not an intrinsic property of it, but is dependent upon the way in which the
term has been incorporated into a theory. (Feyerabend, 1962, reprinted in
Feyerabend, 1981, p. 74)5
1
What follows draws on material I have presented in more detail in Arabatzis (2006). I thank the University of
Chicago Press for granting me permission to adapt and use portions of Chapter 9 of this book, entitled
“Identifying the electron: Meaning variance and the historicity of scientific realism”. © 2006 by the University
of Chicago. All rights reserved.
2
For a comprehensive analysis of this debate, from a realist point of view, see Psillos (1999).
3
cf. van Fraassen (2002, p. 56).
4
I have discussed Kuhn’s recent and more philosophically articulated views on scientific realism in Arabatzis (2001).
5
Note that Feyerabend framed the issue in terms of meanings, as opposed to concepts.
Conceptual Change and Scientific Realism 49
Since scientific concepts are dependent on the theory in which they are embedded, con-
ceptual change is a necessary consequence of theory change. Concepts “change their
meaning with the progress of science”, since
extension of knowledge leads to a decisive modification of the previous

theories both as regards the quantitative assertions made and as regards the
meanings of the main descriptive terms used. (Feyerabend, 1962, reprinted
in Feyerabend, 1981, pp. 80–81)
What are the implications of conceptual change for scientific realism? To use a concrete
example, emphasized by both Kuhn and Feyerabend, if the
physical referents of these Einsteinian concepts [space, time, and mass] are
by no means identical with those of the Newtonian concepts that bear the
same name (Kuhn, 1970, p. 102)
then what should we conclude about the ontological status of the referents of those concepts?
There are two options available to the realist. First, he or she could accept the discrep-
ancy between Newtonian and Einsteinian concepts and argue that only the Einsteinian con-
cepts have genuine physical referents. However, if the difference between Newtonian and
Einsteinian concepts implies that the former do not have a referent then, similarly, future
theoretical developments may reveal that Einsteinian terms are non-referential too. Second,
the realist could deny that Kuhn and Feyerabend have provided an accurate characterization
of conceptual change in science and, thus, dispute their claim that Newtonian terms do not
have a referent. An argument along these lines would be that, despite the differences
between Newtonian and Einsteinian concepts, there are sufficient similarities between the
two sets of concepts to enable us to maintain that Newtonian terms refer to the manifesta-
tions of relativistic entities at low velocities compared with the velocity of light.6
Finally, another realist response would be to point out some difficulties in the view of con-
cepts favored by Kuhn and Feyerabend and thus to cast doubt on their conclusions. As I men-
tioned above, according to that view, the concepts associated with scientific terms are theory
dependent. One needs to know then what counts as part of a theory and what kinds of the-
ory change amount to a transformation of the concepts of the theory. In this respect, as has
been pointed out by Dudley Shapere, Kuhn’s and Feyerabend’s views are open to objection
(see Shapere, 1966).7 “Theory” is usually understood as a systematic and articulated body of
knowledge. Feyerabend on the other hand has a more inclusive conception of “theory”:
The term “theory” will be used in a wide sense, including ordinary beliefs
(e.g., the belief in the existence of material objects), myths (e.g., the myth
of eternal recurrence), religious beliefs, etc. In short, any sufficiently
general point of view concerning matter of fact will be termed a “theory”.
(Feyerabend, 1965a, p. 219)
6
cf. Shapere (1991, p. 656).
7
The critique of Feyerabend that follows is based on Shapere’s paper. The attempt, however, to improve on
Feyerabend’s views so as to meet Shapere’s criticism is mine.
Given this notion of “theory” and the theory dependence of concepts, it is not clear what
aspects of the theoretical context are relevant to specifying a concept. Feyerabend, to the
best of my knowledge, has not provided a satisfactory resolution of this difficulty.
Nevertheless, this difficulty can be, at least partly, alleviated if we narrow Feyerabend’s
conception of “theory” so as to exclude ordinary beliefs, myths, religious beliefs, etc. and
to retain scientific theories only. In such a case a concept would be specified by the set of
properties ascribed to its referent (a purported entity in nature) by the scientific theory in
which the concept is embedded.8
One may object that not all the properties attributed to an entity are constitutive of the
concept associated with it. The attribution of some of those properties, on such a view, is
a merely factual assertion about the entity and is, therefore, irrelevant to the concept asso-
ciated with it. A satisfactory rebuttal of that objection would require an extensive discus-
sion of the analytic/synthetic distinction, a task that is beyond the scope of this chapter.
Here I can only appeal to the widely accepted view, originating from Quine, that a clear-
cut distinction of this kind is not possible.9 Furthermore, the main realist opponent of
Feyerabend, Hilary Putnam, has emphasized
the impossibility of separating, in the actual use of the word, that part of the
use which reflects the ‘meaning’ of the word [or, I would add, the concept
associated with a word] and that part of the use which reflects deeply
embedded collateral information. (Putnam, 1962, pp. 40–41)
I will have more to say about this issue below.

To return to my proposed modification of Feyerabend’s notion of concepts, a difficulty
remains: What is the relationship between theory change and conceptual change? In other
words, what kinds of modifications in a theory affect the concepts that the theory speci-
fies? One could think of changes in a theory that would be too minor to affect the concepts
embedded in it. For example, a refinement of the gravitational constant in Newton’s law of
gravity would hardly change the meaning of any Newtonian terms. Feyerabend tried to
alleviate this difficulty by proposing that
we shall diagnose a change of meaning [conceptual change] either if a new

theory entails that all concepts of the preceding theory have zero extension
or if it introduces rules which cannot be interpreted as attributing specific
properties to objects within already existing classes, but which change the
system of classes itself. (Feyerabend, 1965b, reprinted in Feyerabend,
1981, p. 98)
The former “diagnostic” procedure is far too strict. After all, one would want to diag-
nose a case of conceptual change if a new theory entails that some of the concepts of the
preceding theory are vacuous and refer to nothing at all.
8
A similar view is held by Dudley Shapere. The only difference is that, in Shapere’s view, the concept associ-
ated with a term is identified with the set of properties attributed by the group of scientists who are using the term
to the entity that the term denotes (see Shapere, 1982, p. 21).
9
For some useful reflections on this issue, see Papineau (1996).
The latter concerns the taxonomic function of scientific theories and it presupposes the
existence of unambiguous rules of classification, which can be used to collect objects,
events, entities, etc. into sets. If a new theory attributes additional properties to the entities
denoted by the concepts of the previous theory but preserves its taxonomic rules, one
should conclude, according to Feyerabend, that no conceptual change has taken place.
First, one might want to dispute his claim that the properties of an entity, to the extent that
they are not related to any rules of classification, are irrelevant to the concept associated
with the corresponding term. However, even if every feature associated with an entity is
relevant to the specification of the corresponding concept, it would still be the case that any
conceptual change brought about by the addition of new properties to a fixed set of enti-
ties would lack any anti-realist connotations. On the contrary, conceptual change of this
kind would support a realist position, since it would have taken place against a stable onto-
logical background.
Furthermore, the existence of unambiguous taxonomic rules in science has been denied
by Shapere (1966). Scientific classifications, in his view, are usually based on pragmatic
criteria, which do not reflect any intrinsic properties of the entities classified. For exam-
ple, the question
Are mesons different “kinds of entities” from electrons and protons, or are
they simply a different subclass of elementary particles? … can be
answered either way, depending on the kind of information that is being
requested for there are differences as well as similarities between electrons
and mesons … [It] can be given a simple answer (“different” or “the same”)
only if unwanted similarities or differences are stipulated away as inessen-
tial. (Shapere, 1966, pp. 51–52)
Thus, an appropriate choice of taxonomic criteria might leave the concepts associated
with the entities in question unaffected by theory change.
Shapere’s criticism is based on his emphasis on the pragmatic character of scientific
classifications, an emphasis that, I think, is overstated. Physics, for instance, has produced
an unambiguous classification of the unobservable realm into various particles (electrons,
protons, neutrons, etc.). It is true that their occasional similarities enable the classification
of otherwise different entities under the same category. For example, electrons, protons,
neutrons, and quarks are all classified as fermions, since they obey Fermi–Dirac statistics.
But these entities, despite their occasional similarities, are clearly distinguished from one
another according to their intrinsic properties (e.g., the magnitude of their charge, their
mass, etc.). At least some classifications are predicated on the intrinsic properties of the
entities classified and, thus, Feyerabend’s notion of conceptual change could escape
Shapere’s criticism if one limited its applicability to classifications of that kind. With these
qualifications, the above difficulties in the criterion of conceptual change favored by
Feyerabend disappear and, thus, cannot be employed to support a realist position.
An anti-realist stance is implicit in those of Feyerabend’s views that I have discussed so
far. In his early work he did not adopt an explicit position on the ontological status of unob-
servable entities, an aspect of the realism debate with which he was not directly concerned.
He was occupied, rather, with the phenomenon of meaning change and its implications for
the then-current theories of scientific explanation and reduction. In his later writings, how-
ever, he formulated unambiguously the anti-realist perspective that was implicit in his
early papers. The phenomenon of conceptual change at its most extreme (incommensura-
bility) implies, in his view, that past scientific theories, which have been replaced by their
incommensurable successors, were based on a non-existent ontology. For example,
prerelativistic terms … are pretty far removed from reality (especially in

view of the fact that they come from an incorrect theory implying a nonex-
istent ontology). (Feyerabend, 1970, p. 87)
Feyerabend’s view of conceptual change, a view strikingly parallel to Kuhn’s, presented

a serious challenge to scientific realists. It seemed that the instability of scientific concepts
had as an immediate corollary the collapse of a realist attitude toward their referents. Given
the then-prevalent belief (inspired by Frege) that a concept is specified by a set of condi-
tions that are necessary and sufficient for its correct application, the slightest change in
those conditions (conceptual change) would imply that the concept as previously used was
vacuous, that is, it referred to nothing at all. Hence Kuhn’s recent remark that “the history
of science is the history of developing vacuity” (Kuhn, 1989, p. 32).
Furthermore, even if one rejects the ‘necessary and sufficient conditions’ view of con-
cepts, as most philosophers (and, as far as I know, all psychologists) nowadays do, a prob-
lem remains. If a concept evolves over time it is not clear how it can refer to the same
entity.10 Why is it, for example, that Faraday’s field concept and Einstein’s field concept
have the same counterpart in nature? (cf. Nersessian, 1984). If they do not, as Kuhn and
Feyerabend argued, then one is forced to admit that there are no entities in nature with the
properties of Faraday’s field. Only the Einsteinian field concept may have a genuine ref-
erent. And even that possibility may be unlikely. Consider the following alarming scenario,
due to Hilary Putnam:
What if all the theoretical entities postulated by one generation (molecules,

genes, etc., as well as electrons) invariably do not exist from the standpoint
of later science? … One reason this is a serious worry is that eventually the
following meta-induction becomes overwhelmingly compelling: just as no
term used in science of more than fifty (or whatever) years ago referred, so
it will turn out that no term used now refers (except maybe observation
terms, if there are such). (Putnam, 1984, pp. 145–146)
Thus, Putnam saw clearly the, prima facie, anti-realist implications of conceptual
change. Let me now turn to his significant attempt to evade those implications.
Putnam on Conceptual Change: A Realist Way Out?
In the early 1970s Putnam tried to disentangle concepts from their referents by suggesting
that the latter are independent and essential constituents of the former. That is, a concept
10
Remember that the requirement for a stable ontology is crucial for realism.
is determined, to a large extent, by its referent and not vice versa. Whatever changes the
other components of a concept may undergo the “referential” component will remain unaf-
fected. Putnam was not particularly concerned with theoretical concepts denoting unob-
servable entities. Following Kripke, whose ideas were developed in the context of modal
logic, he was occupied, rather, with natural kind concepts in general. The purpose of his
arguments was to show that the referents of natural kind concepts have not been affected
by the development of scientific knowledge. Even though
the concept [of a natural kind, e.g., fish] is not exactly correct (as a
description of the corresponding natural kind)…that does not make it a
fiction. …The concept is continually changing as a result of the impact of
scientific discoveries, but that does not mean that it ceases to correspond
to the same natural kind … . (Putnam, 1973, p. 197)
I have two comments here. First, I do not think that one can exclude the possibility that
future changes in the concept of, for example, fish will affect its reference. The concept
may change so that organisms that are now classified as fish will cease to be thus classi-
fied. Second, the example of “fish”, or any other observable natural kind concept, is not
very helpful when it comes to questions of scientific realism. Even if the reference of that
concept changed, the existence of the individual organisms that we classify as fish would
not be questioned. And that is because the existence of those organisms is established on
grounds independent of our system of classification. That is not the case, however, when
the natural kind concept refers to unobservable entities. Besides the fact that several such
concepts have been abandoned,11 even when they have been retained, the lack of inde-
pendent, physical access to the entities denoted by those concepts makes problematic the
claim that, in case of conceptual change, their referents remain stable. Whatever the merit
of Putnam’s arguments for the referential stability of observable natural kind concepts, I
will argue that, when it comes to theoretical concepts, which denote unobservable entities,
his theory of reference does not remove the threat that conceptual change poses to scien-
tific realism.12
Putnam’s view of concepts was motivated by a basic realist intuition, namely “that there
are successive scientific theories about the same things: about heat, about electricity, about
electrons, and so forth”. For instance,
Bohr would have been referring to electrons when he used the word “elec-
tron”, notwithstanding the fact that some of his beliefs about electrons were
mistaken, and we are referring to those same particles notwithstanding the
fact that some of our beliefs — even beliefs included in our scientific “def-
inition” of the term “electron” — may very likely turn out to be equally
mistaken. (Putnam, 1973, p. 197)
11
For a very long list of concepts that have been abandoned, see Laudan (1981).
12
Several critics of Putnam have also argued, for different reasons than I will suggest below, that his theory does
not work for theoretical concepts (see Enç, 1976; Kroon, 1985; Nola, 1980).
To flesh out this realist intuition Putnam had to reject the ‘necessary and sufficient con-
ditions’ view of concepts. In his view, membership in the extension of a term is not deter-
mined by a set of necessary and sufficient conditions, which constitute the concept
associated with that term. Rather the reference of a term (concept) is fixed through our
causal interaction with the world:
things which are given existentially … help to fix reference. Actual things,
whatever their description, which have played a certain causal role in our
acquisition and use of terms determine what the term refers to. (Putnam,
1983, p. 73)
Thus, the reference of a concept is an independent component of it and is not affected

by (even drastic) changes the other components might undergo.13
One of the main arguments that Putnam directed against the necessary and sufficient
conditions view of concepts was the following thought experiment.14 Consider our famil-
iar concept of water. The reference of the concept of water, on the necessary and sufficient
conditions view, is determined by the phenomenological properties we attribute to water
(transparent, tasteless, thirst-quenching, etc.). Putnam, however, pointed out that the dis-
covery of a substance with the same phenomenological properties as our familiar water but
with a totally different structure (XYZ) is conceivable. In that case the reference of our
concept of water would not include the novel substance (XYZ). Thus, the “stereotypical”
features of an entity or substance, despite being part of the concept associated with it, do
not determine the concept’s reference. It follows that the evolution of a concept need not
affect the stability of its reference.
Putnam’s rejection of the necessary and sufficient conditions view neutralizes, to some
extent, the anti-realist implications of conceptual change. If a concept could refer to the
same entity(ies) despite the fact that it has evolved, anti-realism would not be an
inescapable consequence of the instability of scientific concepts. I put the qualifier “to
some extent” because a belief in the existence of, say, electrons would make sense only if
a core of the concept of the electron (i.e., some of our beliefs about electrons) has remained
stable since the initial proposal of the electron hypothesis. Cataclysmic changes of scien-
tific concepts, which would amount to the abandonment of every belief about the corre-
sponding entities, would be indeed a threat to scientific realism. Barring such radical cases,
Putnam’s rejection of the necessary and sufficient conditions view of concepts should be
an indispensable ingredient of any realist position.
Nevertheless, Putnam’s success in disentangling the reference of a concept from the rest
of it does not extend to cases where a concept refers to an unobservable entity (or a class of
13
Here it is worth pointing out that Putnam’s views on reference have remained intact, even though his more
recent views on scientific realism depart considerably from the realist intuitions that motivated his earlier
attempts to articulate a theory of meaning. In a recent discussion of this issue, for example, he insists that “the
reference of the terms ‘water,’ ‘leopard,’ ‘gold,’ and so forth is partly fixed by the substances and organisms them-
selves… [while] the ‘meaning’ of these terms is open to indefinite future scientific discovery” (Putnam, 1990,
pp. 109–110).
14
See Putnam (1975a, pp. 223ff).
such entities). In order for the referent of a concept to be an essential and stable constituent
of the concept in question, as Putnam’s theory requires, it should be possible to identify the
referent without relying on the other constituents of the concept. To that effect, one should
have independent, physical access to the entities denoted by a concept, independent of
(most of) our beliefs about them. This independent access guarantees, extreme skepticism
excluded, that those entities are real. However, in the case of unobservable entities no such
direct epistemic access exists. All that we have is a set of indirectly confirmed beliefs about
those entities and it is not clear that we can identify them independently of those beliefs.15
Furthermore, since the problem under examination is the ontological status of unobservable
entities, it goes without saying that a theory like Putnam’s, which presupposes their exis-
tence, cannot be employed without circularity to support a realist position.
My evaluation of Putnam’s theory as a potential realist tool was based on the assump-
tion that unobservable entities are beyond direct empirical access. This assumption has
been forcefully challenged by Ian Hacking, whose views on scientific realism are the sub-
ject of the next section.
Hacking’s Entity Realism
Hacking (1983) has advanced a realist position, which is based on a close examination of
experimental practice. A satisfactory resolution of the problem of scientific realism would
be possible, Hacking claims, only if we stopped being preoccupied with scientific theoriz-
ing and shifted instead our focus of analysis toward experimentation. Such a shift of
emphasis would be enough to make us all realists with respect to some unobservable enti-
ties, but would not weaken our anti-realist convictions with respect to the theories that pos-
tulate those entities. This peculiar mix of realism about entities and anti-realism about
theories follows from two central aspects of experimental practice. On the one hand, the
manipulation of unobservable entities in the laboratory provides sufficient grounds for
believing in their existence. On the other hand, the fact that experimentalists use, accord-
ing to the purpose in hand, a variety of sometimes incompatible theoretical models of those
entities, generates strong doubts that any of those models accurately represent reality. All
these models, however, have some aspects in common, namely a core of statements about
the causal properties of the corresponding entities, properties which we have come to know
by manipulating those entities in various experimental contexts. One can (should) be a
realist about this common core, which, however, does not deserve to be called a “theory”.
Hacking’s entity realism can be summarized in his aptly chosen slogan: “If you spray [e.g.]
electrons then they are real” (Hacking, 1983, p. 24). Notwithstanding the charm of Hacking’s
slogan, it fails to impress philosophers in the empiricist tradition. Van Fraassen, for instance,
when asked to evaluate Hacking’s argument, responded in the following way: “If they are
15
I will suggest below a recipe for identifying the reference of theoretical concepts denoting unobservable enti-
ties that would allow one to include reference as a component of a concept. However, it is not applicable to enti-
ties that have not been (and perhaps cannot be) subject to experimental investigation (e.g., black holes).
Furthermore, even where it is applicable, the referential stability of a concept requires historical investigation and
should not be taken for granted prior to such an investigation.
real then you spray them”.16 Van Fraassen does not imply, of course, that everything real
can be sprayed. Rather, his point is that one can use the expression “spraying of electrons”
(as the best available description of a given experimental situation) without committing
oneself to believing in the existence of electrons. The coherence of his position stems from
the possibility that a new theory, which would not include electrons in its ontology, could
adequately account for the experimental situation that Hacking, following contemporary
experimentalists, describes as “spraying of electrons”. It is conceivable that the process
now described as “spraying of electrons” might be re-described in terms of an alternative
theory based on a different ontology.
Another objection has been raised against Hacking’s claim that manipulability is a neces-
sary and sufficient criterion for establishing the existence of an unobservable entity, namely
that it does not do justice to actual scientific practice (see Morrison, 1990). In particular, his
exclusive emphasis on manipulability fails to capture the variety of evaluative criteria that are
employed within scientific practice for demonstrating the reality of an unobservable entity.17
The most important of those criteria seems to be the empirical adequacy of the theory that pos-
tulates the entity in question. Even though manipulability is one of the criteria that scientists
often employ, it fails to carry conviction in all contexts. Furthermore, scientists need to have
a fairly definite idea of what it is that they are manipulating before invoking manipulability as
a demonstrative principle. In many experimental contexts we know that we are manipulating
something without knowing what it is that we are manipulating. Manipulability, however, was
supposed to provide adequate grounds for the existence of, for example, electrons and not
merely for the existence of an “I know-not-what” something.
Thus, Hacking’s manipulability criterion fails as a descriptive account of the plurality
of principles that scientists employ in the construction of “existence proofs” for unobserv-
able entities. It also fails as a normative account, since even within actual scientific prac-
tice manipulability does not (and should not) provide adequate reasons for belief in the
entities that are supposedly manipulated.
A further problematic aspect of Hacking’s realist position is associated with his “home
truths” (low-level generalizations) about, for example, electrons that we supposedly know
independently of any high-level theory. Hacking does not specify what kind of electron
properties he has in mind but one could guess that his “home truths” would include well-
known causal properties of electrons, like their charge, mass, and spin, which enable us to
manipulate them in order to investigate other less well known aspects of nature. It is diffi-
cult to see, however, how one could isolate the concepts denoting those properties from the
background theory in which they are embedded. To use a concrete example, it is difficult
to obtain an understanding of “charge” independently of any high-level theory, especially
in view of the fact that “charge” meant different things for different scientists.
16
Personal communication.
17
One could deny that Hacking’s criterion needs to capture the richness of scientific practice, by interpreting it
as a normative as opposed to a descriptive criterion. However, Hacking himself stresses the descriptive (as well
as the normative) dimensions of his criterion. For him, the main reason that scientists themselves believe in the
reality of some unobservable entities is their ability to manipulate those entities: “The vast majority of experi-
mental physicists are realists about some theoretical entities, namely the ones they use. I claim that they cannot
help being so” (Hacking, 1983, p. 262).
Hacking could argue that all those different and incompatible conceptions of charge had
certain aspects in common, that is, the causal properties that we have associated with elec-
tric charge all along (e.g., the ability of charges to attract, or repel each other). But even
those properties may vary across theoretical frameworks. In classical electromagnetic the-
ories, for instance, one of the causal properties of a charged particle was that it emitted radi-
ation when performing accelerated motion. In the old quantum theory of the atom, on the
other hand, charged particles in accelerated motion (electrons orbiting around the nucleus)
did not radiate. Hacking’s attempt to isolate the causal properties of electrons from any
background theory was motivated by the plurality of incompatible models about electrons.
However, it turns out that the causal properties themselves have been interpreted via sev-
eral, incompatible theories. He could, of course, search again for a common core shared by
those theories, but it is not clear that there would be an ending point to this process.
A closely associated problem is Hacking’s selective realism. Let us grant for the sake
of argument that belief in the existence of the “well-known causal properties” of some
entity does not depend on how we interpret those properties theoretically. As I already
mentioned, in his view, one should maintain a realist position with respect to those prop-
erties, but, nevertheless, should not commit oneself to believing that the theories involved
in the interpretation of the relevant experiments are true.18 However, as was pointed out by
Duhem, experimental results falsify (or, I would add, confirm) whole chunks of knowledge
consisting of high-level theory(ies), an understanding of the instruments involved in the
given experimental context, statements about initial conditions, etc. In view of this very
plausible holistic epistemology, which Hacking has not by any means discredited, his
selective realism does not make much sense. To put it another way, it is far from clear why
our ability to make sense of certain experimental situations confers a privileged status on
the entities supposedly involved in the experiments but, nevertheless, cannot be employed
as an argument for a realist stance toward the theory(ies) involved in the interpretation of
those experiments. The following example would suffice to illustrate my point: The manip-
ulation of electrons requires knowledge of their behavior in various experimental contexts
(e.g., in a cloud chamber). To control that behavior one has to know, in addition to the
causal properties of electrons, several background theories (electromagnetism, mechanics,
etc.), which predict how an entity with those properties would behave in a given context.
It is those background theories along with Hacking’s “home-truths” that enable interven-
tion. Thus, if our ability to intervene is an argument for realism, that realism should be of
a very broad kind covering entities and theories alike.
Having discussed some significant positions that have been advanced with regard to the
implications of conceptual change for scientific realism, I will now put forward my own
approach to this problem.
How to Rescue Scientific Realism from Conceptual Change
To present my own view on conceptual change and scientific realism, it is necessary to

adopt a particular view of scientific concepts. In what follows, I take a concept to be the
18
Perhaps one should be a realist about the phenomenological theories of the instruments involved.
set of features that are ascribed to its putative referent, by the theory in which the concept
is embedded. In case that no such systematic theory has been developed, a concept would
consist of the characteristics attributed to its counterpart in nature, by the group of scien-
tists who are using it. These characteristics consist of two interrelated kinds: the properties
of the entity in question and the laws obeyed by it. The former are usually specified via the
latter. For instance, what charge is depends on the laws obeyed by charged bodies, for
example, the laws of classical electromagnetism. Conversely, the laws obeyed by an entity
depend on its properties. For example, the electrons qua charged particles are supposed to
obey Coulomb’s law.
As I already mentioned, one may find objectionable the suggestion that all the charac-
teristics ascribed by a theory to an entity are part of the corresponding concept. In partic-
ular, one may call for a distinction between the concept-constitutive and the merely factual
aspects of the information concerning the referent of a concept. I think, however, that such
a distinction cannot be drawn in a satisfactory way. To begin with, no distinction of this
kind is inherent in the scientific representation of an entity. For example, electrons are rep-
resented as subatomic particles, with a specific amount of charge, a specific rest mass,
spin, etc. There is nothing in this representation, however, that explicitly distinguishes
some of these properties as constitutive of the concept of the electron.
Could one impose such a distinction and on what grounds? I can think of two alterna-
tives, none of which seems adequate. First, one could argue that only those properties that
are relevant to the classification of an entity are part of the corresponding concept. For
example, only the subset of the properties of electrons that enables us to distinguish them
from other elementary particles should be considered part of the concept of the electron.
This recipe has the undesirable consequence that a concept may change if the domain of
classification is enriched. Consider, again, the case of electrons. Until the early 1910s there
was no need to distinguish them from other charged particles, because they were the only
particles postulated. When the concept of the proton was introduced, however, the elec-
tron’s negative charge, or the small value of its mass became crucial for distinguishing it
from a proton. Should we say that with the postulation of protons the concept of the elec-
tron changed? Would someone who learned the concept before the proposal of protons
learn something different from someone who learned about electrons after that proposal?
Second, one could argue that only those properties that are used for identifying the ref-
erence of a term belong to the associated concept. These are properties, as I will suggest
below, which enable the identification of an entity in an experimental situation. In the case
of electrons, their charge-to-mass ratio was crucial for deciding whether various phenom-
ena (cathode rays, Zeeman effect, etc.) were their observable manifestations. Other prop-
erties, for example, the size of electrons, were not significant in that respect (see Arabatzis,
2006). The problem with this suggestion is that the properties employed for identifying the
reference of a term may not suffice to convey the associated concept. The concept of the
electron, for instance, is not exhausted by the electron’s charge-to-mass ratio. One has to
know much more than that to know what an electron is.
The view of concepts I have suggested differs both from the necessary and sufficient
conditions view and from Putnam’s notion. It differs from the former in that the properties
associated with an entity and the laws it is supposed to obey are neither necessary nor suf-
ficient conditions that any such entity should fulfill. The advancement of science may very
well eliminate several of those characteristics, reveal further properties of the entity in
question, or alter the laws it is supposed to obey without thereby affecting the reference of
the corresponding concept.
The view of concepts I advocate also departs from Putnam’s in that it does not require
the referent of a term to be a stable and independent component of the associated concept.
Dropping this requirement has two advantages. First, no problem arises when a concept
denotes unobservable entities with no experimental manifestations, whose reference,
therefore, cannot be identified independently of a description. Second, the stability of a
concept’s reference is left open and, as I will argue below, becomes an issue that has to be
settled through historical work.
Given the above view of concepts a realist position is compatible with the phenomenon
of conceptual change to the extent that a core of the evolving concept in question has sur-
vived changes in theoretical perspective.19 Thus, histories of scientific concepts can play a
seminal role in evaluating the tenability of a realist attitude toward the corresponding enti-
ties. On the one hand, a realist stance toward a particular entity would be discredited if the
historical record revealed that the corresponding concept has undergone a radical trans-
formation, which has affected even the most central features previously associated with
that entity. On the other hand, if historical analysis shows that a core of the concept has
remained unaffected by changes in theoretical perspective, this would be an argument
(albeit not a conclusive one) for maintaining a realist position with respect to the entity
under examination. Such a position would involve the inductively grounded belief that the
core of properties attributed to the entity will survive changes in high-level theory.
One might want to challenge the view that the survival of a conceptual core enables a
realist perspective to get off the ground. In particular, one might ask for the specific crite-
ria that privilege this set of beliefs (the “core”) over the rest of a concept that proved highly
unstable. I will not attempt to respond to this line of criticism, because I think that there is
another way out for the aspiring realist. In particular, there is a way to identify the refer-
ence of a theoretical concept, denoting an unobservable entity, without relying heavily on
that entity’s description. Theoretical concepts are usually introduced and articulated to
make sense of experimentally obtained phenomena. The concept of the electron, for
instance, was initially developed to account for the behavior of cathode rays in a vacuum
tube and for the magnetic splitting of spectral lines (see Arabatzis, 2006). When presented
with such a concept, one has to identify the experimental situations that are taken to mani-
fest the presence of (are causally attributed to) its referent, the corresponding unobservable
entity. Thus, the experimental situations associated with a concept provide a way to track
its referent.
Putnam has anticipated this proposal by suggesting that we can identify a magnitude
(e.g., charge)
by, for example, singling it out as the magnitude which is causally respon-
sible for certain effects (Putnam, 1975b, p. ix)
19
A similar suggestion has been defended by Putnam (1974, p. 275). For some other options that are open to the
realist, see below. cf. also Psillos (1999, p. 295).
However, he takes for granted that the reference of scientific concepts does not change
over time. In my view, on the other hand, the realist intuition that (mature) science has
developed against a stable ontological background has to pass the test of historical
scrutiny. It is an open question whether the reference of, say, the concept of charge has
remained stable. Having provided a criterion for identifying reference, the stability of the
reference of a specific concept over time should be investigated historically. Historical
research, thus, obtains an essential role for evaluating the tenability of a realist attitude
toward the ontology of science.
Furthermore, Putnam does not address a question that, I think, is important: What binds
certain effects together as manifestations of a single entity? (cf. Kuhn, 1979, p. 411).
Different effects can be attributed to the same entity only if they have some qualitative or
quantitative features in common. First, they may share some common qualitative feature,
which indicates that a single entity is “behind” all of them. For example, it was known
since the 17th century that the processes of combustion, calcination (oxidation), and res-
piration took place only in the presence of atmospheric air. That fact was explained by the
hypothesis that air was necessary for absorbing phlogiston, the entity that was given off in
all those processes. Second, from the quantitative features of an experimental situation it
may be possible to infer the value of some property of the hidden entity involved in it.20 If
inferences from different situations converge to the same results, then this agreement binds
them together as effects of the same entity. Furthermore, the value in question provides a
way of identifying the entity in novel experimental situations. For example, at the turn of
the 19th century several experimental phenomena (the Zeeman effect, cathode rays, !-rays,
etc.) were attributed to the presence and action of hypothetical charged particles. It was not
a priori evident that the particles involved in all those different phenomena were the same.
Physicists were led to that conclusion when it turned out that the particles that were
responsible for the Zeeman effect had approximately the same charge-to-mass ratio with
the particles which constituted cathode rays. From that point on, the charge-to-mass ratio
of the electron functioned as a criterion for identifying it in novel experimental situations
(for details see Arabatzis, 2006).
To be a realist about a particular entity one has to establish the referential continuity of
the corresponding concept. That is, to show that over the history of the concept the exper-
imental situations that were assumed to be the observable manifestations of its referent
have remained stable or, at least, exhibited a cumulative development (i.e., the previous set
of situations associated with the concept was a subset of the current set). This expansion
may happen in three ways. First, new situations may get attributed to a familiar entity,
without any modification in the theoretical description of that entity. In that case the cor-
responding concept will remain unaffected. Second, the accommodation of new situations
as manifestations of a familiar entity may require the attribution of additional properties to
that entity, which, moreover, do not conflict with its previously established properties. In
that case, the concept will expand in a cumulative way and its reference will remain the
same. Third, the accommodation of novel situations may require the attribution of novel
20
This type of “backward” inference is sometimes called “deduction from the phenomena”. I think that this
expression is misleading, because these deductions always involve additional theoretical assumptions about, for
instance, the entity’s properties or the laws it obeys. See the very enlightening discussion in Worrall (2000).
properties to the entity, which, however, are incompatible with some of its previously
accepted properties. The latter will, in the process, get rejected. In that case, the concept
will change in a non-cumulative fashion. Despite that change, the concept may still refer
to the same entity, provided that the experimental situations previously associated with the
concept continue to be attributed to its referent. As long as that “family” of experimental
situations remains invariant (or expands in a cumulative fashion) a realist stance toward the
entity involved in them is not threatened by changes in that entity’s description.
Thus, the question whether realism is a defensible approach vis-à-vis a certain entity
requires a historical reconstruction of the evolution of the corresponding concept (which
requires considering the experimental situations that track its reference). Unless the his-
torical and philosophical community undertakes successfully the enormous historiographi-
cal task of showing the referential continuity of most scientific concepts, any realist
position will make sense only with respect to specific entities (local realism).
Acknowledgment
I thank Aristides Baltas, Stathis Psillos, Matti Sintonen, and Stella Vosniadou for their
helpful comments.
References
Arabatzis, T. (2001). Can a historian of science be a scientific realist? Philosophy of Science, 68,
supplement, S531–S541.
Arabatzis, T. (2006). Representing electrons: A biographical approach to theoretical entities.
Chicago, IL: The University of Chicago Press.
Enç, B. (1976). Reference of theoretical terms. Nous, 10, 261–282.
Feyerabend, P. (1962). Explanation, reduction and empiricism. In: H. Feigl, & G. Maxwell (Eds),
Scientific explanation, space and time (pp. 28–97). Minnesota Studies in the Philosophy of
Science (Vol. 3). Minneapolis, MN: University of Minnesota Press.
Feyerabend, P. K. (1965a). Problems of empiricism. In: R. G. Colodny (Ed.), Beyond the edge of
certainty (pp. 145–260). Englewood Cliffs, NJ: Prentice-Hall.
Feyerabend, P. K. (1965b). On the ‘meaning’ of scientific terms. Journal of Philosophy, 62 (10),
266–274.
Feyerabend, P. K. (1970). Against method: Outline of an anarchistic theory of knowledge. In:
M. Radner, & S. Winokur (Eds), Analysis of theories and models of physics and psychology
(pp. 17–130). Minnesota Studies in the Philosophy of Science (Vol. 4). Minneapolis, MN:
University of Minnesota Press.
Feyerabend, P. K. (1981). Realism, rationalism and scientific method: Philosophical papers (Vol. 1).
Hacking, I. (1983). Representing and intervening: Introductory topics in the philosophy of natural
science. Cambridge: Cambridge University Press.
Kroon, F. W. (1985). Theoretical terms and the causal view of reference. Australasian Journal of
Philosophy, 63 (2), 143–166.
Kuhn, T. S. (1970). The structure of scientific revolutions (2nd ed.). Chicago, IL: The University of
Chicago Press.
Kuhn, T. S. (1979). Metaphor in science. In: A. Ortony (Ed.), Metaphor and thought (pp. 409–419).
Kuhn, T. S. (1989). Possible worlds in history of science. In: S. Allén (Ed.), Possible worlds in
humanities, arts and sciences (pp. 9–32). Berlin: Walter de Gruyter.
Laudan, L. (1981). A confutation of convergent realism. Philosophy of Science, 48, 19–49.
Morrison, M. (1990). Theory, intervention and realism. Synthese, 82, 1–22.
Nersessian, N. J. (1984). Faraday to Einstein: Constructing meaning in scientific theories.
Dordrecht, The Netherlands: Martinus Nijhoff.
Nola, R. (1980). Fixing the reference of theoretical terms. Philosophy of Science, 47, 505–531.
Papineau, D. (1996). Theory-dependent terms. Philosophy of Science, 63, 1–20.
Psillos, S. (1999). Scientific realism: How science tracks truth. London: Routledge.
Putnam, H. (1962). The analytic and the synthetic. In: H. Putnam, Philosophical papers
(Vol. 2, pp. 33–69). Cambridge: Cambridge University Press.
Putnam, H. (1973). Explanation and reference. In: H. Putnam, Philosophical papers (Vol. 2,
pp. 196–214). Cambridge: Cambridge University Press.
Putnam, H. (1974). Language and reality, In: H. Putnam, Philosophical papers (Vol. 2,
Putnam, H. (1975a). The meaning of ‘meaning’. In: H. Putnam, Philosophical papers (Vol. 2,
Putnam, H. (1975b). Mind, language and reality: Philosophical papers (Vol. 2). Cambridge:
Cambridge University Press.
Putnam, H. (1983). Reference and truth. In: H. Putnam, Realism and reason: Philosophical papers
(Vol. 3, pp. 69–86). Cambridge: Cambridge University Press.
Putnam, H. (1984). What is realism? In: J. Leplin (Ed.), Scientific realism (pp. 140–153). Berkeley,
CA: University of California Press.
Putnam, H. (1990). Realism with a human face. Cambridge: Harvard University Press.
Shapere, D. (1966). Meaning and scientific change. In: R. G. Colodny (Ed.), Mind & cosmos: Essays
in contemporary science and philosophy (pp. 41–85). Pittsburgh, PA: University of Pittsburgh
Press.
Shapere, D. (1982). Reason, reference, and the quest for knowledge. Philosophy of Science, 49,
1–23.
Shapere, D. (1991). Leplin on essentialism. Philosophy of Science, 58, 655–677.
Van Fraassen, B. C. (2002). The empirical stance. New Haven, CT: Yale University Press.
Worrall, J. (2000). The scope, limits, and distinctiveness of the method of ‘deduction from the
phenomena’: Some lessons from Newton’s ‘demonstrations’ in optics. British Journal for the
Philosophy of Science, 51, 45–80.
Chapter 6
Background ‘Assumptions’ and the Grammar

of Conceptual Change: Rescuing Kuhn by
Means of Wittgenstein
Aristides Baltas
Introduction
It has been variously noticed but mostly not discussed at the length required that no sci-
entific concept and no scientific theory can work only at the explicit level. Since no inquiry
can start from the zero point of an absolute beginning, all concepts and all theories, even
the most formalized or highly counterintuitive, cannot but relate to some aspects of com-
mon wisdom and hence cannot but rely on an amorphous plethora of assessments,
appraisals, commitments, connotations, prejudgements and so forth that are taken for
granted one way or another. That they are taken for granted means that the process of
inquiry has not come to dispute their established roles and hence it has allowed them to
continue performing their work more or less silently, inexpressibly, hiding their function
from view. Following standard terminology, we can say that such assessments, appraisals,
commitments etc. form the inquiry’s background. The ‘elements’ of such background have
been variously called: among other things, they have been named “idola” by Bacon, “hid-
den lemmas” by Lakatos, “natural interpretations” by Feyerabend, “ontological commit-
ments” by Laudan, ingredients of “practical ideology” by Althusser, pieces of “tacit
knowledge” by Polanyi, or “historically changeable Kantian categories” by Kuhn. Let us
give them a name in our turn and call them, for reasons that will appear in a moment, back-
ground ‘assumptions’ with quotes.
The present investigation starts by endorsing Kuhn’s insight that the unavoidable exis-
tence of such background ‘assumptions’ offers an additional argument as to why the notion
of paradigm — in the sense of “disciplinary matrix” — should replace the traditional notion
of theory: a notion richer than “theory” is required if we are to take into account factors
instrumental for the functioning of scientific concepts but not involved in the explicit level.
We can usher Wittgenstein into the picture if we try to move a step further in our efforts to
understand the exact role played by such ‘assumptions’. In §4.002 of the Tractatus Logico-
Philosophicus, Wittgenstein (1986) talks of the “enormously complicated silent adjustments

ISBN: 0-08-045355-4
64 Aristides Baltas
(Abmachungen)” at work in our understanding of everyday language. And in his On

Certainty (Wittgenstein, 1969) he returns, I take it, to the same idea talking now of the
“hinges” that “have to stay put” for the door of understanding to move (pp. 341–343).
Without entering into a Wittgenstein exegesis, I just take the background ‘assumptions’ in
question as simply another name for either these “silent adjustments” or those “hinges”. This
is to say that I consider them as mutely adjusting the usage of the concepts at work in any
given language game so as to assure their legitimate sense. In other words, these background
‘assumptions’ are the “hinges” sustaining the concept’s manifest meaning by noiselessly dic-
tating the conditions under which it makes sense; it is they that make up the latent gram-
matical conditions of the concept’s meaningfulness in the corresponding context of its use.
What follows is an attempt to draw the implications of this idea for conceptual change.1
It will be shown that Kuhn’s overall account can, with some qualifications, be rescued
from the charges of idealism and relativism while some implications for education will
also be drawn.
Background ‘Assumptions’ and Paradigm Change
Since on most views the meaning of a concept derives from the propositions engaging it,
it is to propositions we have to turn in order to assess concretely the role of background
‘assumptions’. Thus some such ‘assumptions’ may render a proposition’s truth perfectly
obvious while hiding themselves and the role they are silently playing in doing this behind
the glare, so to speak, of that very obviousness. For example, on some such ‘assumptions’,
the earth cannot but be immobile for anybody possessing her senses. On the other hand,
some other background ‘assumptions’ may lie well concealed beneath our taking another
proposition as totally inconceivable. For example, again on some background ‘assump-
tions’, no action at a distance can be conceived for how can bodies react instantaneously
to the presence of another faraway body when, in addition, there is nothing in between?
The development of classical mechanics has unearthed the ‘assumptions’ lying silently
beneath the above propositions. And it has done that in a way making both this old obvi-
ousness and that old inconceivability appear today as astonishingly short sighted to almost
everybody, that is, to all those having successfully gone through an elementary physics
course. Teachers of physics are well aware of this, for overcoming the obstacles to such
changes in understanding provides their daily bread.
Moreover, such ‘assumptions’ may also work under a tautology, a contradiction, a deduc-
tive consequence or an analytic truth, mutely assuring the logical status of the propositions
concerned while again hiding themselves and their role from view. Putnam’s (2000) exam-
ple is that somebody can be either naked or not naked but only on the silent background
‘assumption’ that she is not wearing a net. It is well known that Lakatos’s (1976) “hidden
lemmas” mess up rigorous deductive chains while it is analytically the case that the earth
cannot be a planet, if the concept “planet” is indeed defined as referring to celestial bodies
1
What follows is borrowed heavily from Baltas (2004). In that paper I focused on radical scientific discovery,
while here I shift the emphasis toward the concomitant radical conceptual change. In addition, I try to clarify
some formulations and forward a few remarks intending to open the discussion along different paths.
Background ‘Assumptions’ and the Grammar of Conceptual Change 65
moving around the earth. All kinds of examples can be added, implicating everyday or highly
elaborated concepts.
As Wittgenstein (1969) has shown in his On Certainty, the ‘assumptions’ in question
are not proper assumptions — hence the quotes. They are neither a priori and indubitable
nor a posteriori and open to doubt. In normal circumstances,2 the role they are playing
remains veiled and they cannot be moved around in the space of justifications. This is
another way of saying that they are the “hinges” that have to stay put if the inquiry is to
proceed with its questions, its reasons and its doubts. And it is in this sense that they con-
stitute the latent quasi-logical, that is, precisely, the grammatical, conditions allowing the
concepts they sustain to have the meaning they do.
In forming the latent grammatical conditions allowing the concepts involved in a
scientific inquiry to make sense, these ‘assumptions’ underwrite the “natural interpreta-
tion” (Feyerabend’s way of referring to their function) of the corresponding conceptual
system and, by means of it, everything this interpretation determines as, for instance, the
sense in which the corresponding experimental transactions and their results are to be
taken. Thus they are involved in all the pictures, the analogies, the metaphors etc., which
scientists are based on in order to understand the concepts their own work produces, and
to communicate their results. In this way they assure the overall grammatical consistency
of the process of inquiry, channeling it away from the ungrammatical. Given this, we can
take an additional step that Wittgenstein, I believe, fully licences although he does not
explicitly take (since he is not particularly concerned with conceptual change) and states
that while channeling the investigation away from the ungrammatical, the very same back-
ground ‘assumptions’ form the meaning bounds of the concepts implicated thus surrepti-
tiously closing the corresponding horizon.
On this basis, a paradigm change can be understood as centred on the disclosure of such
an ‘assumption’ (or a set thereof) and to the effects of such a disclosure. The pattern is as fol-
lows: First, a process of inquiry stumbles on a Kuhnian “anomaly” which, in the most severe
cases, is a total deadlock or contradiction — what confronts the scientists implicated makes
no sense to them and hence they find themselves confronting a complete impasse. On the
point above, we can say that, one, such a deadlock or contradiction surfaces when the gram-
matical bounds of the concepts involved tend to be trespassed and that, two, it is the back-
ground ‘assumptions’ implicated that silently determine these grammatical bounds, since
their function is precisely to steer the process of inquiry clear of the ungrammatical. Hence
it is they that silently determine what is a deadlock or contradiction in the given context.
Given this, the second step of paradigm change amounts to the illumination of the back-
ground ‘assumption’ at issue and to its emergence from out of the background. Such a dis-
closure transgresses the bounds in question, amounting thus to a leap into the
ungrammatical for whatever concerns the old paradigm, and retraces accordingly — both
widens and modifies — the grammatical space available to the inquiry. This becomes the
grammatical space underwriting the novel paradigm.
Thus an episode of paradigm change turns an ‘assumption’ that, as we have said, is
formlessly taken along as a matter of course and to which, accordingly, questions could
2
For reasons to appear later, “normal” here should be taken as pointing toward Kuhn’s “normal science”.
66 Aristides Baltas
not be addressed (i.e., an ‘assumption’ with quotes) into a distinct proposition that can be
doubted and thence conceptually and experimentally examined (i.e., a proper assumption
without quotes). This proposition thus becomes open to rejection, revision, justification
and so forth.3 It is these new possibilities for the development of the process of inquiry that
activate the novel grammatical space opened by the disclosure in question and it is in this
way that the novel paradigm becomes articulated. Thereby, the conceptual system con-
cerned is itself not only enriched but also transformed substantially.4 We have a case of
conceptual change.
We should stress that the disclosure of a background ‘assumption’ accomplishes when
it happens to two distinct things. On the one hand, it clears, as we have just said, novel
grammatical space. As the horizon of inquiry is no longer closed by the dumb existence
of this ‘assumption’, new avenues of research are opened, new questions are asked and
new answers are given, the ones, precisely, that deploy the novel paradigm. On the other
hand, such a disclosure creates a novel vantage point from where the preceding state of the
investigation can be looked at anew. This is a vantage point whereby the anomalies whose
unaccountable existence had induced the disclosure in the first place appear as miscon-
struals due to the work that the disclosed ‘assumption’ had been silently performing.
Hence reasons can be adduced post hoc for the change that has occurred, reasons that
appear as correcting previous inadvertences or oversights and hence reasons that tend to
assess this change as invariably progressive. The spontaneous whiggism of practising sci-
entists, rehearsed at length in the literature, is based precisely on this.
Examples of Conceptual Change
Although interesting debates have been taking place on whether the ideas of conceptual
change, paradigm change and revolution can apply to mathematics,5 with Wittgenstein
apparently having laid the issue to rest in advance for it would have been misleadingly
formulated,6 I believe that examples from mathematics do help in clarifying what is here
at stake. Accordingly, my first example will be Cantor’s theory of the infinite, understood
as a deep conceptual change concerning enumeration.
We know that Cantor’s theory of sets is the result of his efforts to harness mathemati-
cally the infinite while the necessary ground was opened by Cantor’s famous definition: a
set is of infinite cardinality (Cantor’s novel concept whose meaning widens and modifies
what the old concept of enumeration used to cover) if and only if its elements can be put
into one-to-one correspondence with the elements of one of its proper subsets. What
3
And not just to a simple dismissal and replacement, as would have happened, for example, in the ideological
change involved in a religious conversion. This distinction opens various important issues, which I cannot enter
into here.
4
No scientific concept can perform its function in isolation from the other concepts (Baltas, 1988). Hence back-
ground ‘assumptions’ form the invisible grammatical ‘glue’ allowing the conceptual system to be ‘naturally’
understood (see Baltas, 2004).
5
An interesting volume on the issue is Gillies (1992).
6
For Wittgenstein’s overall conception of mathematics, see Shanker (1987). The author, I take it, leaves
Wittgenstein’s would-be position on paradigm change in mathematics interestingly inconclusive.
relates Cantor’s definition to what we have been saying is the fact that our author, before
offering his definition, was up against a total deadlock, which, in hindsight, we can name
“the anomaly of enumeration”. On the one hand, since the set, say, of even numbers is a
proper subset of the set of natural numbers, it obviously contains ‘fewer’ elements while,
obviously again, the two contain an ‘equal’ number of elements, since these can be put into
one-to-one correspondence. What Cantor did in order to escape the anomaly can then be
considered very ‘simple’: he did not try to resolve or in any way overcome the contradic-
tion but merely turned the very deadlock into a definition! Let us try spelling this out.
The deadlock Cantor encountered was determined as that by some background
‘assumption’ working under the time-honoured conception of the extremely close connec-
tion (if not identity) between the part/whole and the smaller/greater relations. Within the
then emerging language of set theory, the first relation is defined like this: a “proper sub-
set” of a given set is one included in that set without exhausting it; a non-empty remain-
der is left. Bringing in the second relation, we can then hold that the set, say, of even
numbers, is smaller, and cannot be but smaller, than the set of natural numbers, since the
latter includes the odd numbers as well. On the other hand, within the same language and
again by definition, sets have elements, that is, they are the sets of their elements. And
another time-honoured idea is that the even numbers can be put into one-to-one corre-
spondence with the natural numbers: the even number 2n corresponds one-to-one to the
natural number n. This is exactly what gives rise to the “anomaly of enumeration”, for then
the set of even numbers has the same number of elements as the set of natural numbers
although a moment ago we demonstrated that the former is smaller than the latter — How
else could “smaller” be cashed in here, if it is only elements that sets can have? — and
hence it cannot but have fewer elements! Cantor’s change of paradigm, the deep concep-
tual change he effected, amounts then to his accepting this contradiction and go on defin-
ing infinite sets as precisely those for which both of the contradiction’s horns hold
simultaneously. The well-entrenched background ‘assumption’ disclosed through this
move is that always and without exception, analytically so to speak, a collection of items
that is smaller than another, one forming a proper part of its corresponding whole, cannot
but include fewer items. Cantor’s definition opens up the grammatical space where it is not
necessarily the case that the relation “properly included in” coincides with the relation
“has fewer elements than”: for finite sets the two relations are equivalent but for infinite
sets they are not. Giving room to this distinction — a distinction literally unthinkable
before — sets up this novel grammatical space wherein the old contradiction becomes
automatically resolved. The articulation of the new paradigm of enumeration, that is,
Cantor’s theory of cardinality, is nothing but the exploitation of this novel grammatical
space through the deployment of that distinction’s far-reaching consequences. A radical
conceptual change in mathematics has thus occurred.
We have to note, however, that after all is said and done, after the radical novelty of
Cantor’s proposal has become absorbed and assimilated into the common wisdom of logi-
cians and mathematicians, his move can appear only as simple and natural: Why expect
that the infinite should obey the same “laws” as the finite? Post festum, but only post fes-
tum, the question is, of course, totally justified in its disarming candidness: Why indeed?
My second example proposes a summary reconstruction of Bohr’s theory of the atom
as a decisive step toward the formulation of quantum mechanics. Schematically speaking,
68 Aristides Baltas
Bohr’s revolutionary proposal may be reconstructed as follows. For reasons too complex
to enter into here, Bohr could not, or would not, question either the electromagnetic the-
ory of his time or the planetary model of the atom. But the two together demonstratively
lead to an unstable atom: the accelerated motion of the revolving electrons makes them
radiate and thus forces them to fall onto the nucleus. If we deem unstable atoms unac-
ceptable at any price, we have again an anomaly in the form of a contradiction. Bohr’s
overcoming the anomaly can then be considered as similar in structure to Cantor’s. He
refuses succumbing to the contradiction but goes forward by accepting both its horns in
the form of a positive definition of an entirely new physical concept: atoms in the “ground
state” do not radiate and hence they are stable. But they do not radiate and are stable only
by fiat, that is, merely because they are defined in precisely this way by Bohr.7
Since, now, the trajectories of the electrons in this newly defined state do not obey the
classical electromagnetic laws, these can be no ‘real’ trajectories’, whatever this might mean
or imply. In other words, Bohr’s new definition opens up the novel grammatical space capa-
ble of hosting the new distinction his definition comes up with, namely that between mov-
ing ‘classical’ bodies, which do emit radiation when following particular trajectories, and
newly defined “moving quantum bodies” that do not. Distinguishing in this way ‘classical’
from quantum motion unavoidably puts under fire the fundamental concept of “trajectory”,
leading thereby, grammatically if not historically, to the follow-up distinction between
‘classical’ trajectories, where position and velocity (or momentum) can be determined
simultaneously with arbitrary precision, and quantum ‘trajectories’ where they cannot.
Given that “trajectory” is, by definition, the sequence in time of the determined positions
and velocities (the tangents of the corresponding curve) of a moving body, this further dis-
tinction is tantamount to the disclosure and the subsequent questioning of a quasi-elemen-
tal background ‘assumption’: it is not necessarily the case, as it had been silently ‘assumed’
up to then, that the trajectories of all physical bodies do possess this, their defining, prop-
erty. Although this makes quantum ‘trajectories’ impossible to strictly represent in space,
cloud chambers and other experimental artefacts demonstrate that the concept should not be
altogether jettisoned and hence that classical and quantum concepts bear very complex rela-
tions to one another. Concurrently, the disclosure of this ‘assumption’ opens the grammati-
cal connection between quantum ‘trajectories’ and the particle/wave duality thus placing the
articulation of the new paradigm of quantum mechanics well on its track. The radical con-
ceptual change associated with quantum mechanics finds here one of its roots.
But again, after Bohr’s initial move and its various grammatical implications have
become assimilated into the common wisdom of at least the practising physicists, the kind
of candid question we encountered in the case of Cantor inevitably surfaces: Why expect
that the exceedingly small should obey the laws governing the medium-sized bodies of our
everyday experience? The whole practice of teaching introductory quantum mechanics is
founded on the apparent naturalness of precisely this question.
7
Evidently, this simple picture does not intend to exhaust Bohr’s logical, conceptual or experimental motivations
for proposing his model of the atom; it does not seek to trace the history of this proposal, be it even in such a
summary fashion; nor does it aim at reconstructing Bohr’s way of thinking on the issue. It only wants to suggest
that paradigm change and conceptual change are concomitant with deep grammatical changes, the understanding
of which may help us coming to grips with the philosophical issues involved.
My last example concerns the other major revolution in the physics of the 20th century,
namely relativity theory.
It is well known that Maxwell’s electromagnetic theory achieved, among other things,
the unification of electricity, magnetism and optics. One of the more salient features of that
achievement was the prediction of electromagnetic waves, the existence of which was later
confirmed by Hertz. Given that, classically speaking, the concept “wave” is defined as the
propagation of a medium’s disturbances, within the grammatical space supporting this def-
inition — that is, within the grammatical space available to classical physics — the con-
cept “wave” is analytically related to the existence of a material medium. Therefore such
a medium, called “the ether” after a long tradition, could not but exist and it was the physi-
cists’ duty to pin down its properties. However, all efforts to such effect failed systemati-
cally, the ether’s character becoming thereby all the more elusive and mysterious. A crisis
situation thus settled in wherein the ether started to appear as the source of a whole set of
anomalies. It was in this context that Einstein published his “On the Electromagnetics of
Moving Bodies”, introducing the special theory of relativity.
Speaking very roughly, the theory of relativity can be seen ex post facto as resulting
from the disclosure of some of the background ‘assumptions’ silently underlying the clas-
sical concepts of motion and the concomitant postulation — again a case of overcoming
an anomaly through a novel definition — of a new kind of entity, different from both par-
ticles and media, the electromagnetic field.8 The disclosure of such ‘assumptions’ opens
the grammatical space permitting the existence of ‘non-classical’ waves that can propagate
without any material medium being disturbed, a space wherein the ether thus becomes
indeed “superfluous” (Einstein’s word). And this implies that the anomalies associated
with its purported existence disappear for good. But, on the other hand, since such ‘non-
classical’ waves propagate in vacuum, that is, in absence of the material medium whose
disturbance they are, amounting thus to a ‘disturbance’ of nothing that ‘propagates’ in
itself (!), the disclosure in question transforms radically the whole classical conception of
motion. It is this conceptual change that forms the basis for the deployment of the new par-
adigm of relativistic physics.
And once again, and always post festum, we can repeat with both the practising physi-
cists and the teachers of modern physics that follow in their wake: Why expect that the
exceedingly fast should obey the laws governing the kinds of motion we encounter daily?
We see that in this example as well the spontaneous whiggism of both physicists and teach-
ers of physics finds room to thrive.
Rescuing Kuhn
In the above sections we have been using the term “conceptual change” rather loosely.
However, much of the criticism against Kuhn has revolved around the exact meaning of
this term, especially as regards concepts that keep the same name across paradigm change
(e.g., classical and relativistic mass). We are therefore obliged to ask: What exactly do we
8
This is not historically accurate in the details. Remaining however close to the historical facts, it permits us to
bring out the grammatical structure of Einstein’s breakthrough.
70 Aristides Baltas
mean when we say that such concepts change? Is there a kind of continuity that can jus-
tify their keeping the same name or, as Kuhn has mostly been taken to imply, does their
incommensurability make their meanings totally unconnected and hence their keeping the
same name merely a matter of convention? Or is the very idea of meaning change and
hence of incommensurability a non-issue? To try answering, we should have a close look
on how what we have been saying bears on Kuhn’s ideas on paradigm change. We will
attest then that the main charges against Kuhn’s account can be dispelled.
Idealism
First, from the psychological point of view, the disclosure of a background ‘assumption’
does amount, as Kuhn would have it, to a “Eureka!” experience inducing a gestalt shift that
makes one see the world under a new light. Given that the main ingredients of paradigms
cannot but be ideas, Kuhn’s maintaining that people residing in different paradigms “live
in different worlds” seems to imply that it is ideas in general that determine the world and
hence are primary in respect to it. As this is the core thesis of idealism, Kuhn has been
charged, apparently with good reasons, for offering an idealist account of science.
However, on a closer look, this charge is totally misguided. For, as given above, Kuhn’s
account of paradigm change can only imply that it is the world, not only as fully independent
of ideas but also as absolutely primary in respect to them, that can be the only agency respon-
sible for anomalies, that is for the deadlocks or contradictions on which paradigms stumble.
This is to say that it is only the world that can display the inadequacy of a conceptual system,
the fact that such a system does not possess the resources for coping with the relevant dead-
locks or contradictions, and hence for accounting for the world in the corresponding respect.
It is only the world, as fully independent of the ideas articulating the various paradigms,
which, by resisting them, can force us into accepting that those ideas fall short of it and thence
are subordinate to it and fully reliant upon it, for better or for worse. Ideas are at peace with
the world, allowing scientific, or even naïve, realism to appear as compelling, only as long as
no dire anomalies surface to exert their full pressure.9 It follows that Kuhn’s position on par-
adigm change not only cannot be charged as idealist but, on the contrary, that it is a position
vindicating with a vengeance the sovereignty of the world in respect to ideas, that is, the
absolute primacy that practically all forms of naturalism demand of it.
Communication Breakdown
To go on, let us look at the phenomenon Kuhn calls “communication breakdown”. On the
present view, communication between scientists invariably breaks down in the relevant respect
when these reside in succeeding paradigms. Why this is the case should by now be obvious.
The scientist residing in the old paradigm has not undergone the pertinent “Eureka!” experi-
ence and still holds fast to the corresponding background ‘assumption’, while remaining
unaware that a questionable assumption is involved in the first place.
9
Such resistance of the world to the theories or paradigms trying to account for its various aspects should form,
at least according to me, the starting point for a viable realist position, a position which, for reasons related to the
above argument, I call “negative realism”. I have tried to explore this idea in Baltas (1997). See, however,
Psillos’s (2002) interesting and still unanswered objections.
Given the way we explicated the function of such an ‘assumption’, it follows that con-
tinuing to hold fast to it implies that the scientist implicated cannot understand how it could
be possibly questioned. Such questioning is literally unthinkable from the old vantage point,
as it would amount, for example, to the destruction of an analytic relation between concepts,
that is, of a relation that is inconceivable otherwise. Accordingly, the scientist still residing
in the old paradigm can literally not understand the effects of the disclosure in question and
hence neither the concepts the colleague residing in the new paradigm is using nor much of
her way of talking.
It should be clear, nevertheless, that no kind of simple-mindedness is involved here. To
say it provocatively, Lorentz’s, say, continuing to admit till the end that he never came to
understand relativity theory may even make him more ‘rational’ than Einstein. The old
man had precise, well-articulated reasons — logical, conceptual, empirical and historical
— for clinging, even unawares, to the background ‘assumptions’ assuring the meaning of
the old concepts while his young colleague only made a senseless leap into the ungram-
matical. That Lorentz never underwent the appropriate “Eureka!” experience in no way
demonstrates some kind of intellectual inferiority.
In any case, the communication breakdown under consideration is no global breakdown.
Although the grammatical change involved cannot but rebound to a greater or lesser extent
throughout the old conceptual system (since the relevant concepts are closely intercon-
nected), the two parties continue sharing the enormous grammatical space of their common
language within which the blind spots of their communication can be circumscribed, even
if very hazily. For example, both parties can agree that explaining phenomena is what they
are after while remaining completely at odds, not only on what such explanations should
exactly amount to but also on which are the phenomena to be explained.10
We should add that the communication situation is not totally hopeless even in respect to
such ‘dialogues’. Since the disclosure of a background ‘assumption’ creates, as we have said,
a novel vantage point whereby one can see what prevented this new way of looking before,
one is armed with important rhetorical ammunition.11 Although such ammunition cannot
compel logically those holding the old paradigm to undergo the pertinent “Eureka!” experi-
ence, it can serve to surround, so to speak, the corresponding background ‘assumption’ so as
to force those opponents to realize, at least indirectly and in the long run, that they are unwit-
tingly harbouring unwarranted biases. Our pedagogical practices when we are teaching
counterintuitive theories do bear an eloquent witness to the need for exerting such violence.
Asymmetry
The next point we should note is that the relation between the two succeeding paradigms
is not symmetrical, and this is in many senses. For one, it is not grammatically and hence
epistemically indifferent within which paradigm one resides. Since the grammatical space
available to the new paradigm includes the possibility of examining (of negating, of mod-
ifying, of accepting etc.) an additional assumption unavailable to the old, namely the one
10
The communication gap between Galileo and the Aristotelians, or even that between Frege and Hilbert on
what a mathematical definition should amount to, are cases in point.
11
See, for example, Pera (1994).
72 Aristides Baltas
resulting from the disclosure of the ‘assumption’ that had been silently taken for granted,
it is a grammatical space objectively (i.e., independently of the relevant beliefs and con-
victions) wider than that available to the old paradigm.
It is crucial to underline, however, that the assessment of an objectively greater width
is possible only from the vantage point determined by the new paradigm. Those clinging
to the old continue to be unwittingly constrained by the background ‘assumption’ in ques-
tion, to remain blind to the possibility of its interpellation and, therefore, incapable of sur-
veying the greater width attained. Accordingly, if we keep Kuhn’s metaphor of the “gestalt
switch”, we should note that, unlike the duck/rabbit case and at least thus far in our dis-
cussion, this is not a switch that allows one to go back and forth. It is a semiconducting
switch, directing from the old paradigm to the new through the relevant “Eureka!” experi-
ence.12 This is to elaborate the asymmetry we noted: if we continue residing within the old
paradigm, we are blocked in our understanding at least parts of what our interlocutors are
talking about. In fact we even consider what they are telling us as nonsensical, since
ungrammatical. If we reside within the new paradigm, we can understand both what our
interlocutors are talking about and why they are talking this way. We are not blocked
because, to go one step further, the grammatical space available to us can accommodate
an explicit interpretation in our terms of the old conceptual system as a whole, an inter-
pretation concomitant with the reinterpretation of the empirical phenomena involved.
In the case of post-Galilean physics, where a mathematical structure always underpins the
physical conceptual system thus rendering concepts and conceptual relations more precise,
the interpretation in question is based on an explicit, although imperfect, translation of at
least a crucial part of the old conceptual system in terms of the new.13 The case of mathe-
matics where no empirical phenomena are involved should be treated separately and will
not concern us in the remainder of the chapter.
What I mean can be spelled out as follows. In the passage from the old paradigm to the
new, what first14 opens up is the grammatical space tied to the particular concepts located
at the core of the deadlock or contradiction on which the old paradigm stumbled in the first
place. This opening up blows up and changes dramatically the meaning of these concepts.
Some of them may become fully discarded (as phlogiston or caloric) while some others
acquire a radically new meaning, even as their names, for reasons we will see in a moment,
remain the same. The novel grammatical space is such that three closely connected things
become possible within it.
12
Hence the asymmetry in question should not be identified with Kuhn’s “incommensurability”. The latter con-
cerns the relation between the languages of the two paradigms and does allow a movement back and forth as, say,
between English and French. Thus “incommensurability” can be experienced only from the vantage point of the
novel paradigm and, in this sense, it presupposes the asymmetry we are discussing (see Baltas, 2004).
13
Two remarks are in order: 1. My running together “translation” and “interpretation” here against Kuhn’s
(2000) careful distinction of the two terms is allowed by the asymmetry we are discussing while Kuhn has in
mind only his “incommensurability”. See note above as well as Baltas (2004). 2. My formulation seems to imply
that, to the extent that phenomena become ‘merely’ reinterpreted across paradigm change, they, pace Kuhn,
remain per se invariant. I will try arguing below why this implication need not follow although the kind of con-
tinuity suggested by my formulation seems to me inescapable.
14
Not necessarily in the temporal or historical sense. But what we are concerned with here is only the recon-
struction of episodes of paradigm change, not an accurate historical account of their occurrence.
First, one can understand on its basis the role the disclosed ‘assumption’ had been play-
ing in assuring the coherence and self-consistency of the old conceptual system. Thereby
that old system can be interpreted in a way making clear the reasons for both its successes
and its failures in accounting for the empirical phenomena in its domain.
Second and concurrently, the novel grammatical space can host a reinterpretation in the
new language of the empirical phenomena that had been countenanced, and as they had been
countenanced, by the old. The relevant parts and aspects of nature are being understood now
in terms of the new conceptual system and of the novel grammatical space supporting it.
Third, the new conceptual system can account successfully for at least some aspects
of some of the phenomena (as reinterpreted) that were at the heart of the deadlock or
contradiction from which the whole process started, that is, those that the old conceptual
system found impossible to handle — as well as for at least some aspects of at least some
phenomena (as reinterpreted) that had been considered up to then as being successfully
accounted for by the old conceptual system.
Continuity
Given this, let us turn to the crucial question of whether some kind of continuity can be
said to exist across paradigm change and of how this can be fleshed out. If, as we have
said, the new paradigm resolves the deadlock or contradiction on which the old paradigm
had stumbled, the question is exactly what is resolved in this way and, concomitantly, what
relation the two succeeding paradigms might entertain with each other through this reso-
lution. To answer, we have to distinguish the relation between concepts at the two ends of
the passage in question from the corresponding relation between phenomena.
Let us start from concepts. We have said that paradigm change occurs when a new pos-
itive definition is proposed that trespasses the grammatical bounds of some old concept(s),
instituting thereby the grammatical space underlying the new paradigm. As our examples
have shown, the newly defined concept (cardinal number, quantum trajectory, electromag-
netic field) can host a distinction between itself and the interpretation/translation in terms
of the novel conceptual system of the concept(s) whose meaning bounds have been thus
trespassed. If this were not the case, the newly defined concept would not bear at all on
the anomaly at issue; it would be irrelevant and thus remain perfectly idle in respect to it.
On the other hand, that the old concept(s) have to be interpreted/translated for becoming
thus accommodated is a consequence of their impossibility of overstepping these bounds.
It follows that this interpretation/translation, in being precisely an interpretation/transla-
tion, retains the memory of this impossibility. But the fact that the positive definition in
question institutes a novel grammatical space, wherein the ensuing new concept functions
fully grammatically in tandem with the interpretation/translation of the old, makes the old
ungrammaticality disappear from sight and represses its memory.15 The survival of this
15
It follows that this ungrammaticality becomes eventually forgotten, fully covered under what the resultant
grammatical space starts compelling everyone to take for granted. Thus, in the long run, the achievements of the
superseded paradigm appear retrospectively as trivial, as obviously mistaken or as fully incomprehensible. It is
then up to the historian of science to reveal the revolutionary character of the corresponding ungrammatical leap
and to dispense intellectual justice.
74 Aristides Baltas
impossibility, even merely in the guise of a repressed memory, keeps the new conceptual
system linked to the old, thus rendering it continuous with it in that precise sense.
Going now to the phenomena implicated, we can say that, repressing the memory of
this impossibility goes hand in hand with the reinterpretation of these phenomena by the
new conceptual system in terms of the novel grammatical space. But in what way do we
have here a reinterpretation of the phenomena exhibiting the anomaly? If we do not raise
the issue we cannot but be silently assuming that these are merely reinterpreted and thus
that they retain as such their identity, which inevitably begs the question.
To answer, we repeat that, on our account, the appearance of a deadlock or contradic-
tion in respect to some phenomena means only that the world manifests its resistance to
the conceptual system intending to capture them. This is to say, first, that it is the force of
this resistance that the ungrammatical move into the new paradigm has to overcome and,
second, that the success of this move appeases this resistance by making the new concep-
tual system conform to the world in the relevant respect. But this can only mean that
‘something’ of the world, ‘something’ of the materiality of the phenomena concerned, is
carried invariant across paradigm change.16 For, appeasing such resistance and making the
new conceptual system conform to the world can only mean that this new system manages
to capture adequately the very same ‘something’ that resisted its capture by the old con-
ceptual system. In other words, the same ‘something’ is being conceptualized in terms of
an inadequate system at one end of the relevant passage and in terms of a correspondingly
adequate one at the other end.17
The novel grammatical space is itself stamped by the persistence of this ‘something’: the
fact that what is at issue is an impossibility having concerned the phenomena exhibiting the
anomaly, an impossibility; moreover, that continues to survive — even only as a repressed
memory — within the grammatical space having overcome the resistance they had exhibited
marks precisely the elusive passage of the same ‘something’ across the ungrammatical leap.
We can add that the fact that this ‘something’ remains invariant is faithfully reflected in
the pertinent “Eureka!” experience, for this is an experience that cannot engage but a
single thing at both its ends: after having undergone it, we understand exactly what we
were incapable of understanding before. In one word, it is the invariance of this ‘some-
thing’ across paradigm change that allows maintaining that the new conceptual system
indeed reinterprets the phenomena exhibiting the Kuhnian anomaly. It is precisely this that
makes the new paradigm, not merely different from the old, but the one that succeeds it.
16
This is ‘something’ that any brand of naturalism would minimally require, if the world is to be independent of
ideas and primary in respect to them. For an argument why this ‘something’ need not amount to a Kantian “thing
in itself”, see Baltas (1997).
17
I should note that no “convergence to the truth” is implied by such a passage from the inadequate to the ade-
quate. As we will explicate below, a kind of progress is involved, but it amounts only to a relation between two
grammatical spaces while the world remains, so to speak, gravely silent in the background, permanently refus-
ing to pronounce itself otherwise than through resisting inadequate conceptual systems. This is to say that, as con-
cerns the world, the kind of progress involved is ‘negative’: the new conceptual system merely shows that the
world in the relevant respect is not in fact as the old system takes it to be. And that is that. The novel grammati-
cal space inevitably carries background ‘assumptions’ of its own that another, perhaps much more radical, para-
digm change may eventually disclose, showing thereby that the new conceptual system too is grossly inadequate.
For more details see below as well as Baltas (1997).
The grounds of this continuity provide also the reasons why it is not only historically but
also grammatically required that the names of the old concepts be preserved in the novel
grammatical space as far as its continuity with the old calls for: these names constitute the
necessary grammatical reminder of this continuity, that is, of its inescapability.
We have to acknowledge, however, that such continuity through interpretation/translation
and reinterpretation makes the phenomena captured by the old conceptual system appear as
‘at bottom’ identical with their reinterpretation in the new system. And such a form of con-
tinuity is close enough to a continuity of reference to almost compel us, together with most
practising scientists, to say that the two paradigms talk about the same phenomena tout court.
Yet on the above this cannot be strictly correct. If, as we have said, the world manifests
itself only through its resistance to our conceptual systems, then the phenomena we are con-
sidering cannot provide a firm hold to the relation of reference, at least as that relation has
been standardly discussed. Viewed from another angle, this is to say — as Kuhn too would
have it — that no phenomena can be given to our experience and to our perception inde-
pendently of the concepts capturing them,18 for, to use received terminology, they are
always, in a sense, theory-laden.19 In our terms, this simply means that the ‘something’ of
their materiality that remains invariant across paradigm change cannot be captured inde-
pendently of some conceptual system or other, with all the corresponding background
‘assumptions’ inevitably entering into play and the attendant threat of an anomaly always
lurking in the shadows.
The above entail that the continuity of succeeding paradigms through interpretation/
translation and reinterpretation is always a vital feature of paradigm change, accomplish-
ing at least two purposes. First, such continuity is essential for justifying the use of the
old conceptual system when translatable counterparts exist or when the corresponding
limit conditions are satisfied:20 after relativity theory and/or quantum mechanics became
the reigning paradigms, classical mechanics can be used without many scientific or philo-
sophical qualms (which does not mean straightforwardly or unproblematically), thanks to
the continuity assuring the corresponding interpretation/translation and reinterpretation.
Second, the same continuity allows assessing the role the disclosed ‘assumption’ had
been playing within the old conceptual system and hence accounting for the successes and
failures of the old paradigm as well as for the coming to being of its successor. As specif-
ically in the case of physics, this translation saves — sanctions21 — as limiting cases the
translatable, precisely, concepts and conceptual relations of the old paradigm, its existence
permits us to understand how that paradigm could have been successful in the relevant
respects despite the role being played by the ‘assumption’ in question. Concurrently, that
not all parts of the old conceptual system are saved through this translation22 permits us to
18
As McDowell (1994) phrases it, the “conceptual goes all the way down”.
19
For an interesting way to distinguish the conceptual from the theory-laden that is consonant with the present
account and on the basis of which the continuity we are talking about can perhaps be spelled out more fully, see
Pagondiotis (2004).
20
For details see Baltas (2004).
21
The use of this term intends an allusion to Bachelard’s (1975) “histoire sanctionnée”. It lies outside the scope of
the present paper to develop the proximity (and the differences) between the approaches of Bachelard and of Kuhn.
22
Bachelard’s (1975) “histoire périmée” finds its correspondent here.
76 Aristides Baltas
understand the ways in which the horizon of inquiry had been closed by the silent work
performed by the same ‘assumption’. In particular, we can understand why the new para-
digm could not have come into being without the ungrammatical leap represented by the
novel definition, a kind of leap; however, that preserves the continuity between the two
paradigms precisely through the interpretation/translation and reinterpretation at issue.
Surreptitiously shifting this logical “without” to a temporal “before” obliterates such
imperfection in translation as well as other telling differences between two succeeding par-
adigms, and makes us perceive the old paradigm tout court as either totally discardable or
as just a limiting case of the new. As we have been repeating, this misconception lies at the
roots of the ways we tend to teach modern science while simultaneously feeding the spon-
taneous whiggism of practising scientists.
Thus, such continuity spells out the ways in which the grammatical space attached to
the new paradigm is indeed objectively wider than that attached to the old. Kuhn’s alleged
relativism can be offset at precisely this point.
Relativism
First of all it is imperative to emphasize that attaining this objectively greater width is an
irreversible achievement. Once a new grammatical possibility becomes available through
the disclosure of a background ‘assumption’ and once the corresponding implications have
become domesticated through the successes of the new paradigm, scientific reason cannot,
grammatically if not historically, wipe out this possibility, forget its existence, and act as
if it were not there. Forsaking the “Eureka!” experience, retracing the steps of paradigm
change and making the disclosed assumption re-enter the amorphous background sup-
porting grammatically the old conceptual system is obviously impossible. This is to say
that the route for regaining the lost innocence, the route leading back to conceiving the old
concepts strictly the way they were being conceived before the disclosure, is blocked. At
least part of the grammatical glue assuring the coherence and self-consistency of the old
conceptual system has been found out and, to that extent, it has lost for good the corre-
sponding gripping power. The two succeeding paradigms are asymmetrical also in the
sense that the old paradigm has become definitively superseded.
It follows that no room is left for an extra-paradigmatic vantage point, that is, a point from
where one could impartially assess the relative merits and demerits of paradigms, biased by
none. As Kuhn would have it, nothing at all can be conceived outside a paradigm and, as
McDowell (1994) would formulate it, no view “from sideways on” can ever be available: we
can reason, only in terms of a paradigm, on the appropriate more general understanding of the
term. But if we always find ourselves within a paradigm without the possibility of acceding
to extra-paradigmatic neutral ground, it seems to follow that a paradigm is as good as any
other. And if that is indeed the case, rationality receives a lethal blow while all kinds of rela-
tivist positions, from social constructivism to various forms of alleged ‘post-modernism’, find
a privileged soil to thrive on: if rationality cannot be saved even within science, then it cannot
be saved anywhere and hence we are free to choose the paradigm that suits best our conven-
tions, our interests, or even our whims.
However, this is a non sequitur and the corresponding charge is, once again, totally
misguided. Once we accept that we cannot access God’s standpoint, where no latent
grammatical conditions need exist and all background ‘assumptions’ without exception
lay bare to the gazing, no peril whatsoever to rationality can be forthcoming and no kind
of relativism can be implied from our account. The change from a paradigm to its suc-
cessor is as rational a procedure as any in human thought and action. The catchword is
what we have been repeating almost ad nauseam above: there is an inherent asymmetry
between succeeding paradigms in the sense that the grammatical space available to the
new paradigm is objectively wider than that available to the old. That the new paradigm
supersedes definitively the old means that, after having undergone the “Eureka!” experi-
ence, we necessarily reside within the novel grammatical space with no possibility of
going back and hence that no real choice between the two paradigms can be at issue.23 In
that precise sense, objectivity need not imply neutrality in respect to paradigms and
the attendant impartiality; their tie may appear unbreakable only from the point of view
of God.
Scientific Progress
But if that is the case, then our account appears as open to the opposite charge: Do we
endorse cumulative progress? The answer is no, for that the old paradigm has become
definitively superseded does not imply that the new paradigm can do better than its pred-
ecessor in respect to all the (reinterpretations of the) empirical issues that the old had con-
fronted, successfully or not, on its own.
After the ungrammatical leap represented by the novel definition and after the initial
empirical successes securing the irreversible character of the expansion in grammatical
space, the new paradigm may indeed solve (reinterpretations of) some outstanding old
puzzles, it may dissolve and completely discard through the corresponding interpretation/
translation of some others, or it may stumble in its efforts to solve (the reinterpretations
of) some of the puzzles that the old paradigm had successfully tackled in its own terms.
It may even prove the case that, apart from the inaugurating initial successes, no sub-
stantial claim of the new paradigm can survive future empirical trials or that the inaugu-
rating successes themselves need to be reinterpreted. A new paradigm change may be
called for, a paradigm change that will disclose additional background ‘assumptions’,
those that will appear, ex post facto and from the resulting novel point of view, as respon-
sible for the trouble this new paradigm had been encountering. This implies that there is
no paradigm-free way to assess the successes and failures of paradigms so as to come
up with the conclusion that the new paradigm performs unqualifiedly better than
the old.
For the same reason as above, namely that God’s standpoint is unavailable, no neutral
scale can exist, as the idea of cumulative progress would have it, on which to place such
successes and failures, count them impartially and draw the balance. Nevertheless,
progress does occur. As we have been repeating, the novel grammatical space is objectively
23
Theory comparison, theory choice, incommensurability etc. can become issues only after the novel paradigm
has been established and from its own vantage point (see Baltas, 2004).
78 Aristides Baltas
wider and the old paradigm has become definitively superseded. Progress concomitant
with paradigm change amounts ‘merely’ to this.
Concluding Remarks: Implications for Education
We have said above that the definition instituting the grammatical space of the new para-
digm can be perceived only as senseless from the vantage point of the old paradigm. This
is inescapable and most cases of the relevant communication breakdown rehearsed in the
literature are due precisely to this. How then, can we teach the conceptual ingredients of
the new paradigm to those still clinging to the old? How can we bridge the corresponding
gap in understanding? The answer that follows from the above is simple: we can only start
from openly acknowledging that what we are about to teach to our students cannot but
sound senseless to them and go on by judiciously employing appropriate pictures and
metaphors that do not hide but, on the contrary, bring out the nonsensical, precisely, char-
acter of the basic concepts at issue.
Here is an example. Stephen Weinberg, Nobel laureate in physics and grand master of
exposition, in addressing himself (Weinberg, 1977) to those who do not know relativity
theory, ‘defines’ the electromagnetic field as “a taut membrane without the membrane”.
This is a self-destructing and hence nonsensical ‘definition’, openly admitting that the
concept of the electromagnetic field can make no sense at all from the point of view of
classical physics. At the same time, however, this ‘definition’ proposes a kind of ‘picture’
intending to ‘depict’ the only kind of ‘meaning’ the concept can have in the context of the
old paradigm. Indeed, from the vantage point of classical physics, the electromagnetic
field can be conceived, as we have said, only as the disturbances of nothing that propagate
in themselves, that is, only in terms of this or some equivalent nonsensical ‘picture’ or
metaphor. We seem therefore entitled to generalize: teaching effectively the ingredients of
a new paradigm cannot but be based on the astute use of nonsense.
And this brings me to my last remark. Volumes of inconclusive discussions have been
devoted to the conception of nonsense that Wittgenstein (1986) promotes in the Tractatus
Logico-Philosophicus. Although this is certainly not the place to launch yet another line
of approach to the issue, the above allows me to forward, very tentatively, the following.
If we take the famous penultimate paragraph of that work (§ 6.54: “My propositions are
elucidatory in this way: he who understands me finally recognizes them as senseless...”) to
mean basically that Wittgenstein uses nonsense as a basic tool for making his views under-
stood by his intended interlocutors, then nonsense does amount for him too to a kind of
pedagogical device. And this seems to imply that paradigm change need not be that foreign
to him either. If this simple observation can be developed into an account that holds water,
then it would perhaps be worthwhile trying to read the Tractatus Logico-Philosophicus
retrospectively on the basis of the Kuhnian arsenal as we strived to clarify it here.24 Some
more light might be shed then on both Wittgenstein’s and Kuhn’s ideas, ideas that remain
important for they continue to inform our views of language, of science and of the world.
24
See in this respect Baltas (2005) in Greek, with an English draft available on request.
References
Bachelard, G. (1975). Le nouvel esprit scientifique. Paris: PUF.
Baltas, A. (1988). On the structure of physics as a science. In: D. Batens, & J. P. van Bendegem
(Eds), Theory and experiment, recent insights and new perspectives on their relation
(pp. 207–226). Dordrecht, The Netherlands: Reidel Publishing Co.
Baltas, A. (1997). Constraints and resistance: Stating a case for negative realism. In: E. Agazzi (Ed.),
Realism and quantum physics (pp. 74–96). Amsterdam-Atlanta, GA: Rodopi.
Baltas, A. (2004). On the grammatical aspects of radical scientific discovery. Philosophia Scientia,
8(1), 169–201.
Baltas, A. (2005). On paradigms in philosophy: Showing, elucidation, nonsense, philosophy, silence,
ethics in Wittgenstein’s Tractatus”. In: K. Ierodiakonou (Ed.), The use of paradigms in philoso-
phy. Athens: Ekkremes (in Greek).
Gillies, D. (Ed.). (1992). Revolutions in mathematics. Oxford: The Clarendon Press.
Kuhn, T. S. (2000). Commensurability, compatibility, communicability. In: J. Conant, &
J. Haugeland (Eds), Thomas S. Kuhn, the road since structure (pp. 33–57). Chicago, IL: Chicago
University Press.
Lakatos, I. (1976). Proofs and refutations. Cambridge, UK: Cambridge University Press.
McDowell, J. (1994). Mind and world. Cambridge, MA: Harvard University Press.
Pagondiotis, K. (2004). Can the perceptual be conceptual and non-theory-laden?. In: A. Raftopoulos
(Ed.), Cognitive penetrability of perception: An interdisciplinary approach. New York: Nova
Scotia Publishing Co.
Pera, M. (1994). The discourses of science. Chicago, IL: Chicago University Press.
Psillos, S. (2002). What is the positive content of negative realism. Neusis, 11, 85–96 (in Greek).
Putnam, H. (2000). Rethinking mathematical necessity. In: A. Crary, & R. Read (Eds), The new
Wittgenstein (pp. 218–231). London: Routledge.
Shanker, S. G. (1987). Wittgenstein and the turning point in the philosophy of mathematics. Albany,
New York State: The State University of New York Press.
Weinberg, S. (1977). The search for unity: Notes for a history of quantum field theory. Daedalus,
107, 17–35.
Wittgenstein, L. (1969). In: G. E. M. Anscombe, & D. Paul (Trans.), On certainty. Oxford:
Blackwell.
Wittgenstein, L. (1986). In: C. K. Ogden (Trans.) Tractatus Logico-Philosophicus. London:
Routledge and Kegan Paul.
COMMENTARIES
Chapter 7
Reflections on Conceptual Change

Stathis Psillos
When discussed among philosophers of science, the issue of conceptual change brought
in its tow incommensurability. Kuhn and Feyerabend both suggested that in the transition
from the old to the new paradigm (or theory, or conceptual scheme) there is a deep
conceptual asymmetry. Hoddeson (this volume) narrates elegantly the shock that the
community felt and explains the ways that several members of it dealt with the post-
Kuhnian trauma.
In Structure, Kuhn made notoriously extravagant claims about the proponents of com-
peting paradigms practicing their trades in different worlds, or being transported to differ-
ent worlds etc. When the dust settled, Kuhn drew a distinction between global and local
incommensurability and defended a version of the latter. Machamer (this volume) dis-
cusses this issue in his piece and makes what I take it to be the exactly right point about
global incommensurability: it entails relativistic solipsism (which, I add, a moment’s
reflection shows that it is either incoherent or absurd); hence, any theory that entails global
incommensurability is reduced to absurdity.
What then of local incommensurability? In his mature thought, Kuhn argued that it
occurs when the competing theories have locally different taxonomies of natural kinds
(what Kuhn called lexical structures). It amounts to a claim of local untranslatability,
due to a mismatch between the lexical taxonomies associated with the two theories.
Kuhn, actually, was very careful in characterising the situations in which incommensu-
rability arises. The obvious objection to his claim is that mismatches in the lexical struc-
tures may well be there and yet, the lexica might be sufficiently similar to enable setting
up at least rough-and-ready correspondences among their nodes. To this he replied that
not any “old difference” yields incommensurability; rather, there must be differences
that test the compatibility of nodes in the lexical structures. He talks, for instance, about
the no-overlap principle which captures the old Lockean thought that there are gaps or
chasms in nature: the members of two natural kinds that are not related to each other as
genus and species are disjoint. Incommensurability, then, arises when lexical structure A
classifies entity x under kind K and structure B classifies entity x under the disjoint kind
L; or when lexical structure A classifies entity x under kind K and structure B classifies
entity y — but not x — under the kind K.

ISBN: 0-08-045355-4
84 Stathis Psillos
Machamer (this volume) thinks Kuhn is basically right about local incommensurability.
But I do not see why this should give us pause. As stated, the thesis of local incommensu-
rability merely recapitulates the claim that there is conceptual change. To talk of concep-
tual change as opposed to conceptual replacement presupposes that something remains
unaltered while something else changes. Conceptual replacement does occur but gives no
rise to incommensurability. Conceptual change, on the other hand, reassigns referents
(extensions) to new kinds or reshuffles the referents (extensions) of old kinds. But this
would be a problem only if concepts were identified extensionally. Since they are not,
there is no problem.
The real issue, I take it, concerns the similarities and differences in the conceptual roles
assigned to them. These can offer criteria for conceptual similarity and difference. To clar-
ify this point I will appeal to Sellars (1973). He defended a kind of nominalism about con-
cepts. He denied that concepts are abstract entities (types or universals) to which thought
is somehow related. Concepts are not the kind of thing one is related to by having certain
thoughts — e.g., the concept DOG is not the kind of thing one is related to when one has
dog-thoughts. He took the line that concepts are (or are related to) dispositions and abili-
ties (hence, concepts are not things of any sort), viz., dispositions to think thoughts of cer-
tain things — e.g., to have the concept DOG is to have the ability to think thoughts about
dogs, and in particular to think thoughts to the effect that such and such thing is a dog.
Sellars conjoined this dispositional account of concepts with a functionalist account of
thought: thoughts have the content they do because they function in a certain way and two
thoughts with the same function have the same content. If that is what is important about
thoughts (as Sellars says, apart from their functional role, thoughts are neurophysiological
processes), concepts — being constituents of thoughts — can be identified functionally as
well. Identical are those concepts that function in exactly the same way. This move opens
up the space of comparing concepts according to their functional similarity. How a con-
cept functions will depend on its connection to other concepts (that is, to its conceptual
role) as well as the rules (formal or material, as Sellars would have it) that determine its
contribution to a conceptual framework. Similarity of function is then the means to build
bridges among concepts — even if these concepts are not identical (despite the fact that
they might have the same names.) For instance, one could say that the function of a pitcher
in a baseball game is similar to the function of a bowler in a game of cricket. (So: the con-
cepts PITCHER and BOWLER are similar — this is illuminating, I take it, since we can
now focus on where they differ. Pitching is primarily defensive — it aims to prevent the
other team from scoring runs — while bowling is offensive — it aims to remove the bats-
man). Similarly, (mutatis mutandis) one could say that the function of relativistic mass in
special relativity is similar to the function of classical mass in classical physics.
All this presupposes that the functional role of a concept is not determined in a holistic
and undifferentiated way by the framework in which it belongs. Arabatzis (this volume)
disagrees. He defends conceptual holism (that is, the view that a concept is identified holis-
tically by means of all properties possessed by, and all law-like connections that charac-
terise, the entities to which it applies). But conceptual holism cannot explain the robustness
(or invariance) that many concepts possess, viz., that they do not change very easily, and
certainly they do not change when small alterations take place in their conditions of appli-
cation. Besides, conceptual holism cannot explain what the possession of a concept consists
Reflections on Conceptual Change 85
in since it cannot consist in the possession of a full theory. The sensible thing to say is that
not everything (in a conceptual scheme) is constitutive of its concepts and not all belief
changes constitute concept changes. That a line should be drawn seems imperative.
Arabatzis seems to be saying this: concepts are individuated holistically, but there can be
conceptual stability provided there are independent ways to identify what the concept
stands for as the same again. These independent ways are tied to experimental procedures
in which a stable referent for a concept is identified. Therefore, Arabatzis makes experi-
ments the locus of conceptual stability.
But there seems to be a problem with this. Let us take theories to be (or to fix the)
modes of presentations (senses) of concepts and let us take the experimental procedures to
fix the reference of a concept (the type of entity for which the concept stands for). How
are these related to each other? It seems that, on Arabatzis’ story, these two semantic vehi-
cles move parallel to each other and it is a fortunate coincidence that they, if at all, cross
each other’s path. There is no reason to think that the referent identified in a series of sta-
ble experiments is the referent of a concept whose content is identified holistically by a
theory unless the theory informs the experiments and the experiments expand the theory.
Here again, if the whole of the theory informs the experiment, when the theory changes,
the experiment changes too and the referent that has been identified experimentally
changes too. I do not doubt that experiments offer useful ways to identify entities; the issue
is that they have to identify an entity as the referent of a concept. If this concept is indi-
viduated (or is introduced) in the process of the experiment, well and good. If this concept
is individuated by a theory, we need an extra assumption that it is this concept’s referent
that is identified by the experiment.
With this in mind, let us move back to Sellars (1973). I take him to have argued that it
is precisely because some concepts are similar, that is, they play a similar functional role
in their respective frameworks, and that not all parts of the framework are equally respon-
sible for determining the content of concepts. But how can we classify concepts as being
similar? Sellars’ deepest thought in this connection was that it is a mistake to think that
there is just one single concept of X. Conceptual change (not replacement) occurs when
we move from one concept of X to another concept of X and this, of course, is compatible
with the thought that these concepts of X are relevantly similar to each other (hence, there
is conceptual continuity of the sort that allows for conceptual change).
To illustrate this, Sellars considers the following cases:
(A) ISOSCELES TRIANGLE vs. SCALENE TRIANGLE
(B) EUCLIDEAN TRIANGLE vs. RIEMANNIAN TRIANGLE
In case (A), the two concepts operate within a certain (Euclidean) framework and are
species of a genus. Using type-hierarchies, we could say that EUCLIDEAN TRIANGLE
is a supertype, having ISOSCELES TRIANGLE and SCALENE TRIANGLE as subtypes.
These two subtypes are similar in specific respects (since they inherit the properties of
their common supertype) and different in others (in virtue of which they are classified in
different subtypes). In case (B), the two concepts operate within different frameworks.
Using type-hierarchies, we could say that both EUCLIDEAN TRIANGLE and RIE-
MANNIAN TRIANGLE are subtypes of a supertype TRIANGLE. What differentiates the
subtypes is the fact that EUCLIDEAN TRIANGLE is governed by the axioms of
86 Stathis Psillos
Euclidean geometry, while RIEMANNIAN TRIANGLE is governed by the axioms of

Riemannian geometry. It is obvious that ISOSCELES TRIANGLE and SCALENE TRI-
ANGLE are more similar to each other than are EUCLIDEAN TRIANGLE and
RIEMANNIAN TRIANGLE. But there is no reason to think that both EUCLIDEAN
TRIANGLE and RIEMANNIAN TRIANGLE are not TRIANGLE concepts. There is, as
Sellars (1973, p. 90) put it, a generic functioning of TRIANGLE “which abstracts from the
specific differences between Euclidean and Riemannian geometries”. Sellars’ point is that
it is wrong to think of these two concepts as single concepts. Rather, the expressions
“Euclidean” and “Riemannian” qualify the concept TRIANGLE. (So, strictly speaking, we
should write: Euclidean TRIANGLE and Riemannian TRIANGLE.) The Euclidean
TRIANGLE and the Riemannian TRIANGLE are both varieties of TRIANGLE; they
function in a similar way in their respective frameworks and they function in a way that
TRIANGLE (more abstractly specified) functions.
In On Certainty, Wittgenstein (1969) claimed that some propositions play the role of
hinges that hold together a language-game (or a conceptual scheme). He did not think,
however, that hinges share something deep in common in virtue of which they function as
hinges. They are not analytic truths, nor a priori truths. They are rather heterogeneous col-
lections of propositions (including arithmetical truths, straight empirical propositions and
others). What hinges share in common is that they function in a certain way, viz., within
certain language games, they cannot be doubted without revealing some conceptual short-
coming (cf. p. 137). Their functioning this way, according to Wittgenstein, is the result of
an act (or a deed, as he put it): the act that has to do with that
“we just can’t investigate everything, and for that reason we are forced to
rest content with assumption”. (p. 343)
Baltas (this volume) finds in this Wittgensteinian story a way to understand what
happens during a revolution — and in particular a way to understand a conceptual change.
What he calls background “assumptions” are hinges, with the difference that they are hid-
den — they lie in the background of concepts and determine their conditions of their appli-
cation. According to Baltas, background “assumptions” determine the logical grammar of
a concept. His idea is that a revolution (a paradigm change) consists in a leap into the
ungrammatical (from the point of view of the superseded paradigm) while it also consists
in bringing to the foreground and challenging the background assumptions of the concepts
of the old paradigm (from the point of view of the new paradigm). This is an interesting
asymmetrical relation: the new paradigm opens up a new grammatical (hence conceptual)
space that was not possibly available from within the old paradigm.
The emergence of a new paradigm, according to Baltas, turns a background assumption
into an ordinary proposition, and in particular, one that can be subjected to doubt, empiri-
cal examination, revision or rejection. By challenging this assumption, the new paradigm
redefines an old concept by removing a presupposition for its application.
How are the new and the old concepts related to each other? As Baltas himself notes,
there must be some continuity (similarity) between the two concepts; otherwise, it will be
unclear that they relate to the same phenomena and in particular, that the conceptual
change was prompted (at least partly) as a response to an empirical anomaly that the old
Reflections on Conceptual Change 87
paradigm faced. Now, Baltas seems to be saying three things in response to the question
above. First, there is continuity because the non-grammatical character of the new concept
vis-à-vis the old one is repressed in memory — scientists, that is, forget or choose to for-
get that there was a leap into the ungrammatical. Second, in the transition from the old to
the new paradigm, some worldly item (whatever constituted the anomaly that brought the
old paradigm to crisis) remains invariant. Third, the retention of the same name for the old
and the new concepts constitutes “the necessary grammatical reminder of this [conceptual]
continuity”.
However, even if we were to grant these three points, they would locate the continuity
at the wrong place. The locus of conceptual continuity should be similarities among the
concepts, i.e., among their content (their functions, or their conceptual texture, as Sellars
(1973) would put it). Anything else, desirable though it may be, is an extrinsic character-
istic. Baltas’ first point makes conceptual continuity a matter of (fortunate) psychological
contrivance; his second point makes it a matter of referential stability; and his third point,
makes conceptual continuity a matter of naming. It is worth noting that Sellars’ account
sketched above would be a better candidate for capturing Baltas’ insight. If we think of not
just one but of many concepts of X (e.g., classical WAVE; non-classical WAVE etc.),
Baltas’ idea of changes in the background assumptions can be part of the story that
explains how these concepts are similar to, and different from, each other. What more
should we expect from a theory of conceptual change?
Acknowledgements
Many thanks to my student Milena Ivanova, who made me focus my ideas on some key
issues.
References
Sellars, W. (1973). Conceptual change. In: G. Pearce, & P. Maynard (Eds), Conceptual change
(pp. 77–93). Dordrecht, The Netherlands: D. Reidel.
Wittgenstein, L. (1969). On certainty (G. E. M. Anscombe, & D. Paul, Trans.). Oxford: Blackwell.
Chapter 8
Conceptual Change as Structure Change:

Comment on Kuhn’s Legacy
Matti Sintonen
Thomas Kuhn is best known as a chief critic of the once-dominant Received View of
scientific theories and growth of knowledge. The four chapters by Hoddeson, Baltas,
Arabatzis, and Machamer give a clear and philosophically sophisticated account of Kuhn’s
contribution to the new historicist understanding of science that challenged and in part
replaced this Received View. The little I wish to add concerns the way Kuhn conceived
scientific theories, how scientific terms acquired their meaning, how theory change and
conceptual change could be elaborated, as well as the potential implications of these views
for learning and instruction.
The best way to appreciate Kuhn’s contribution is to see it against the problems of the
Received View. In this view scientific theories were conceived as interpreted (and ideally
axiomatized) logical structures, where the theoretical (unobservable) and non-theoretical
(observable) terms and concepts were connected by laws. Theories, therefore, were sets of
sentences or statements, and once so interpreted they were typically considered true
(or often “strictly speaking” false) claims about the world. Although proponents of the
view did not think there could be any logic of discovery, there was a logic of justification:
experiments and observations on one hand and logic and reason on the other hand force
scientists to abandon false theories and hence to pave the way towards Truth.
Kuhn came to doubt the credentials of the Received View already in the 1940s. Together
with a growing number of scientists and philosophers, he felt that it was more of an
exercise in logic than a description of what actually went on in science. One shortcoming,
germane to our topic, was that the view had virtually nothing to say about how theories
were learned and taught, and its account of how the theoretical (non-observational)
concepts acquired their meaning for individuals and the communities was dead wrong.
Reflecting on Hempel’s (1957) canonical work on concept formation and its influence on
his own views, Kuhn (1993) summarized his departure from the Received View as follows.
First, theoretical terms and along with them concepts are not learned the way logicians
interpret so-far uninterpreted symbols. Rather, they are learned in use. Second, these uses

ISBN: 0-08-045355-4
90 Matti Sintonen
are manifested in paradigmatic examples of “nature’s behaviour”. Third, there must be

several such exemplars for the meaning to catch on; and,
finally, when the process is complete, the language or concept learner has
acquired not only meanings but also, inseparably, generalizations about
nature. (Kuhn, 1993, p. 312)1
As Arabatzis (this volume) nicely shows, the idea that language and world, meanings
and beliefs, are learned simultaneously owes to Willard Van Quine’s theory of meaning.
Quine had challenged the dogmas of the Received View but his proposal, adopted by
Kuhn, made a commitment to thorough conceptual contextualism or even holism: when-
ever there is a change in the theoretical assumptions the crucial concepts alter in meaning.
But if theory change always involves conceptual change, it is difficult to escape relativist
and antirealist conclusions concerning the ontology of science. All manners of unpalatable
or absurd philosophical and empirical consequences would follow. Arabatzis argues that
the dynamic view of growth of knowledge championed by Kuhn in particular does not
have these drastic consequences.
Holism in this extreme form flies in the face of scientists’ intuition and common sense,
and I am not sure that even its weaker forms can or need be rescued. The reason is that the
way Kuhn contrasted the Received View and his own relied heavily on the sort of linguis-
tic view that is currently receding. Instead of trying to formulate a theory of reference
suitable for radical conceptual change (some variety of causal theory might work), I
propose to focus on a more mundane description of radical theory changes as a change in
the structure of conceptual systems, that is, paradigms and theories.
Kuhn’s reflections explain some of the ambiguities of “paradigms” and “theory” noted
by Hoddeson and Machamer (this volume). A paradigm narrowly conceived is an exem-
plary problem solution, whereas a broad reading yields a “disciplinary matrix” or “a con-
stellation of group commitments”. As Machamer puts it, paradigms in this broader sense
are mainly non-propositional sets of interconnected concepts, coupled with shared guide-
lines for application. They provide a communal touchstone for what problems are worth
focusing on, and what kinds of conceptual equipment can be used in the pursuit of solu-
tions. Paradigms in this broad sense include not only exemplars as one ingredient, but also
the symbolic generalizations and the (cognitive) values of the group or community.
Also, theories can be taken in two different senses: as more specific, precisely formu-
lated “principles or laws”; and as complex conceptual structures that include precisely for-
mulated laws (and other ingredients). Second, a theory (in either sense) includes as a
crucial element a set of intended applications, what a theory is “about”. To see how Kuhn’s
1
Kuhn is here referring to Carl Hempel’s (1957). It is of some interest to note that this is the positivists’ series
of publications called Encyclopedia of Unified Science, the very same series in which Kuhn’s Structure of
Scientific Theories was first published. Note also that Kuhn’s ties to the positivist views are not accidental; his
manner of phrasing the problems of meaning and concept formation owes to Hempel’s account. Kuhn came to
disagree with it, but his problem was the same. It is now conceivable that the scene in history and philosophy of
science has changed so that the problem is no longer amongst the most pressing ones — and, therefore, also that
our way of seeing the issue has changed.
Conceptual Change as Structure Change: Comment on Kuhn’s Legacy 91
account departs company from the Received View, let us call, following the so-called
structuralist explication, the precisely formulated theories theory-elements and big theories
theory-nets. Theory-nets are sets of interconnected elements, the sort of “larger structures”
or “research programmes” alluded to by Machamer (this volume).2 Now, theory-elements
(and hence nets) have a dual structure, consisting of a set of intended applications and a
theory-core, most notably the laws that specify the models of the theory. In the Received
View the laws were assumed to have a universal scope, whereas in Kuhn they are explic-
itly tied to particular types of structures described in a non-theoretical vocabulary. It is
important for this view that the intended applications are not given as explicit lists but
rather through exemplars and similarity relations: anything that is sufficiently similar to a
paradigmatic exemplar falls into the scope of the theory-element.
Theories in mature scientific fields, therefore, are not isolated individuals but parts of a
larger, even global structure. For our purposes the most interesting type of structure is a
paradigmatic theory-net with one fundamental theory-core, specified in terms of the deep-
est or most entrenched laws and principles. The basic laws give rise to hierarchy-specific
theory-elements that zero in on more limited classes of applications by making more
specific claims. Normal science, Kuhnian puzzle solving, then amounts to adding such
applications (special laws) to the already existing structure. Initially, a net might contain
just a few well-established (sufficiently confirmed) applications, although the claim made
might be extremely strong: for example, that not just gravitational phenomena (falling
bodies) but pendulums, oscillators, and even electromagnetic phenomena can be captured
under the fundamental laws by hitting upon suitable special laws. A theory-evolution
represents such historical development, and as Machamer says, it can span over decades
or even centuries.
In a comment on Sneed’s (1971) account of theory dynamics Kuhn distinguished
between these two types of change of structure, one evolutionary and the other revolu-
tionary:
“Particularly striking to me, of course, is that its manner of doing so appears to demand
the existence of (at last) two quite distinct sorts of alteration of time”, one in which the fun-
damental theory-core remains the same and is extended to cover new applications, the
other one in which a fundamental core is abandoned for another. The allegiance of a
scientist are to the core and not the special laws that govern particular applications. Two
men can “share belief in a core and in certain of its exemplary applications… even though
their beliefs about its possible expansion differ widely.” (Kuhn, 1976, p. 187)
It is this basic core, also, which is immune to refutation during normal scientific evolution.
To get a better handle on revolutionary conceptual change we need to return to the dual-
ity of cores and intended applications. Very roughly, what happens in revolutionary change
is that the structure of the concepts which enable classification of phenomena as similar
2
I shall skip all formal details and only present the fundamental ideas needed to appreciate its virtues in expli-
cating a conceptual change. The view was first formulated by Sneed (1971) and it was developed by Stegmüller
(1973). The resultant collaboration led into what was known as the structuralist or non-statement view. In private
conversation Kuhn confessed that the view had become too technical to his tastes (“I am too old to start learning
category theory”, he said to me in 1986). Nevertheless, he wrote, Stegmüller has understood his work “better than
any other philosopher who has made more than passing reference to it.” (Kuhn, 1976, p. 179)
92 Matti Sintonen
changes. Put metaphorically, scientific theories (and paradigms in the wider sense) are some-
thing that provide a community of scientists with well-defined puzzles, food for thought, as
well as means for solving these puzzles, that is, the cutlery. Kuhn writes of Sneed:
“For him as for me, the adequate specification of a theory must include
specification of some set of exemplary applications”. (Kuhn, 1976, p. 182)
What falls within the jurisdiction of a theory is determined through their similarity with
these exemplars, not by giving a list or by specifying necessary and sufficient conditions.
And if this is how they are specified, and how the concepts entering the theory get their
identities, there can be subtle individual or communal differences, arising from back-
ground and previous training, in the way a theory is understood.
That similarity measures are the way theories are specified is in line with Baltas’s
(this volume) Wittgensteinian suggestion. Behind all explicit claims and theory formula-
tions (theories in the more restricted sense), there always are background assumptions
or hinge propositions in Wittgenstein’s sense (although to what extent they are propositional
is an issue). These implicit assumptions, variously described by different scholars, delineate
the horizon of possible explicit questions and answers. They therefore comprise the bounds
of sense and amount to the grammar of conceptual change. Eventually, when an anomaly
blows up the bounds of sense, questions formerly inconceivable begin to shape. Clearly,
such a revolutionary disclosure corresponds to overthrowing the fundamental theory-
element and, therefore, bringing to end normal-scientific evolution. Baltas suggests that this
Wittgenstein-inspired view by Kuhn does not lead to idealism or relativism, nor does it
jeopardise rationality and scientific progress — and here Arabatzis agrees.
The notion that exemplars, paradigms in the narrow sense, are an aspiring student’s
entering wedge into science, and that paradigms in the narrow and broad sense specify a
grammar of possible questions, has implications for the manner in which would-be mem-
bers of a scientific profession become card-carrying members. To become a Newtonian is
to learn to see the applications through the fundamental Newtonian laws, and hence to
group falling bodies, planets and comets, pendulums, and oscillators as similar. This does
not come out as natural but may need overcoming of initial or “natural” groupings.
Furthermore, this has implications for the way sciences are learnt and taught, and how
what is learnt at some stage directs attention to where to go next: “Need I emphasize”,
Kuhn wrote, “that learning a theory is learning successive applications in some appropri-
ate order and that using it is designing still others?” The failure of the Received View was,
therefore, not just in its purely static account of theories — confinement to snapshots —
but also a failure to address how the crucial concepts of a theory acquire content in and
through use, scientific practice. Note also that the community of card-carrying members
of a paradigm need not be unanimous when concepts are extended to cover new instances.
Despite the communally shared exemplars they allow leeway in judgments on future cases.
Despite seemingly smooth communication within practitioners of a field there may be
underlying differences in the way phenomena are grouped.
I have focused on (mature) scientific theories as hierarchically organized theory-nets that
evolve by budding more specific theories because it gives a visual picture of theory change at
different levels of generality. Theories also are, in this view, historically evolving particulars.
Furthermore, although the picture does not solve the problem of reference, it throws some
light on the problem of meaning and conceptual change, for adding new special laws (e.g.,
force laws in physics) brings about conceptual change. To illustrate this, and to conclude this
comment, I shall point out the relevance of structure change for cognitive science in two cases:
Giere’s (1994) account of the cognitive structure of scientific theories and Thagard’s (1992)
views on conceptual revolutions. The aim is to give an idea of how Kuhn’s account could be
elaborated (for a thorough and insightful account of Kuhn and cognitive science, see
Nersessian, 2003).
Giere (1994) favors the semantic or model-based approach to scientific theories, as has
been briefly outlined above. He is interested in canvassing a naturalistic approach to science
that explicitly incorporates human agents. Consequently, he draws on what cognitive psy-
chologists have had to say about the nature of concepts and categories. Very briefly, Giere
sides with those who object to the classical view of categories as sets of necessary and suf-
ficient conditions. There is, he suggests, a close connection between categories and theories
which can be brought out to the open by following the pioneering work by Jerome Bruner
in the 1950s and Eleanor Rosch and others in the 1970s. In Rosch’s studies on language
acquisition the key hypothesis, confirmed in different experimental setups and across
different cultures, was that concepts are structured or graded. Following work on the essen-
tially culture-independent or near-universal color terms Rosch extended the idea to artifacts
and what philosophers call natural kinds. A prototype of a chair has four legs, but more
peripheral (shall we say, degenerate) instances can be included, with allowance on increased
time of recognition. Within the category “bird” there are prototypic or indeed exemplary
species (robins) but, with some imagination and time, ostriches (no wings) and even bats
(very degenerate birds, on second thought not really birds at all) can be counted in.
In these studies on categorization, there was also another interesting finding, viz., that
grading can vary horizontally (intracategory grading) or vertically (intercategory grading).
Now, if we would have to guess what level, for example, in a vertical category is more fun-
damental, one could easily go for the most general one as the basic category. For instance,
all lassies are collies, collies are dogs, dogs are mammals, mammals are animals and ani-
mals are living things. So the category of “living things” would be the most fundamental
one. However, Rosch (1978) argued that in this hierarchy “dog” is basic and dogs are the
easiest to identify and group together; categories higher than that exhibit less similarity,
and the gain in similarity when moving down to its subcategories is insignificant. The
notion of similarity does heavy-duty work here, and so do those of a prototype or exem-
plar. Although these notions are far from being unproblematic, there is support for the view
that similarity to prototypes or exemplars is fundamental to categorization. Moreover,
exemplars serve as entering wedges into categories and hence conceptual networks.
Giere’s (1994) point in venturing into the realm of cognitive psychology is that just as
natural and artificial kinds exhibit graded structures, so do scientific theories. Theories are
systems of models that have both a vertical and a horizontal structure. His example is, not
surprisingly, the structure of classical mechanics. The most fundamental level of descrip-
tion, from the point of view of physical theory, would be that of the three Newtonian laws,
which employs the notion of force but imposes no particular form in it (this would be the
basic laws in a theory-net in the sense specified above). The next level splits these models
into two groups depending on whether the total energy of the system studied is constant in
94 Matti Sintonen
time or not. On the next level down the hierarchy, different types of forces impose addi-
tional structure. Finally, on levels IV and V we get down to familiar descriptions of classes
of models and their visual representations: free fall and inclined planes as types of recti-
linear motion with constant force, pendulums and springs as types of harmonic motion, as
well as circular and elliptical orbits as types of orbital motion. These categories can then
be given more fine-graded content still. Now, from the point of view of an abstract mind
wedded to viewing all these classes of models as models of classical mechanics the fun-
damental level is level I, that of Newton’s three laws. However, Giere argues that level IV
and the visual models are basic in the Roschian sense.
Physics is taught by giving examples and exercises in terms of this level, but once they
have been utilized as ladders to climb up to the abstract level they can be kicked away. To
become an expert is to learn to see the intended applications of classical mechanics
through and by help of the higher level mathematical equations, indeed Kuhn’s symbolic
generalizations. And Giere indeed surmises that most of Kuhn’s (1962) exemplary prob-
lem solutions do correspond to Roschian basic-level models.
This view also tallies nicely with that advanced by Thagard on conceptual revolutions.
Thagard writes that
an understanding of conceptual revolutions requires much more than a view

of the nature of isolated concepts. We need to see how concepts can fit
together into conceptual systems and what is involved in the replacement of
such systems. (1992, p. 30)
This Thagard seeks to provide. He gives an account of conceptual systems in which

concepts are organized into hierarchies by the help of various types of links, such as kind,
instance, rule, property, and part links. This makes it possible to grade changes with
respect to their degree of radicalness in terms of adding or deleting nodes and links, and
of adding or deleting rules. As Thagard sums up,
changes in kind-relations and part-relations usually involve a restructuring

of conceptual systems that is qualitatively different from mere addition or
deletion of nodes and links. (Thagard, 1992, p. 32)
References
Giere, R. (1994). The cognitive structure of scientific theories. Philosophy of Science, 61, 276–296.
Hempel, C. (1957). Fundamentals of concept formation in empirical science. Chicago, IL:
University of Chicago Press.
Kuhn, T. (1962). The structure of scientific revolutions. Chicago, IL: University of Chicago Press.
Kuhn, T. (1976). Theory-change as structure-change: Comments on the Sneed formalism.
Erkenntnis, 10, 179–199.
Kuhn, T. (1993). Afterwords. In: P. Horwich (Ed.), World changes. Thomas Kuhn and the nature of
science. Cambridge: The MIT Press.
Nersessian, N. (2003). Kuhn, conceptual change and cognitive science. In: T. Nickles (Ed.), Thomas
Kuhn. Cambridge: University of Cambridge Press.
Rosch, E. (1978). Principles of categorization. In: E. Rosch, & B. Lloyd (Eds), Cognition and cate-
gorization. Hillsdale, NJ: Erlbaum.
Sneed, J. D. (1971). The logical structure of mathematical physics. Dordrecht, The Netherlands:
D. Reidel.
Stegmüller, W. (1973). Theorienstrukturen und theoriendynamik. Berlin: Springer.
Thagard, P. (1992). Conceptual revolutions. Princeton, NJ: Princeton University Press.
PART 2:
PERSONAL EPISTEMOLOGY AND

CONCEPTUAL CHANGE
Chapter 9
Personal Epistemology and Conceptual

Change: An Introduction
Stella Vosniadou
Definitional Issues
Epistemology is the branch of philosophy that studies the nature, source, and limits of
knowledge. Psychologists are not interested in epistemology per se but rather in under-
standing individuals’ beliefs about knowledge. The term ‘personal epistemology’ is used
in the psychological literature to characterize the beliefs of individuals about the nature of
knowledge and the processes of knowing. The interest is in finding out how these beliefs
differ from individual to individual, how they change with development and learning, and
how they influence learning and studying in the subject-matter areas.
The terms epistemic and epistemological beliefs are also used in the literature. In this
volume, Murphy et al. and Alexander and Sinatra propose, following Kitchener (2002),
to use the term epistemic to denote individuals’ beliefs about knowledge, and epistemo-
logical to denote beliefs about the study of knowledge. According to this definition, indi-
viduals, such as students and researchers, have epistemic beliefs. Researchers
investigating the beliefs of individuals about knowledge are, however, engaged in
research on epistemological beliefs. These terms are not used in the same way by all the
researchers in the field, including the authors of the chapters in the present volume.
Furthermore, as Stathopoulou and Vosniadou (this volume) note, there is a set of addi-
tional terms, such as epistemic cognition, epistemological perspectives, epistemological
theories, epistemological resources, etc., which different researchers use, reflecting the
diversity of the theoretical approaches to the construct of interest. Most of the researchers
recognize the lack of terminological and sometimes conceptual clarity in this field of
research, which, as Alexander and Sinatra note in their commentary, may nevertheless be
justified on the grounds that it is relatively new and emergent; in fact, hardly more than
about 10 years of age.

ISBN: 0-08-045355-4
100 Stella Vosniadou
Epistemological Beliefs Research
There is considerable disagreement as to how it is best to conceptualize the nature and

development of epistemological beliefs, although practically all researchers agree that per-
sonal epistemologies change and evolve with learning and development. Some researchers
use quantitative instruments assuming that epistemological beliefs can be conceptualized
as consisting of distinct dimensions each of which represents a kind of continuum between
extremes (e.g., Schommer-Aikins, 2002). Others use qualitative measures, such as inter-
views, considering personal epistemologies as more coherent sets of beliefs evolving
through hierarchically organized sequence of positions (Perry, 1998), stages (King &
Kitchener, 1994) or levels (Kuhn, 1991). Regardless of the particular theoretical frame-
work adopted, personal epistemologies are usually seen as changing through development
and learning, following an ‘upward’ movement from dualistic/absolutist and objectivist
views, to relativist and subjectivist views and finally to contextual, constructivist, and eval-
uative perspectives of knowledge and knowing (Hofer, 2002). This issue needs, however,
to be investigated further as there may be important cross-cultural differences in the devel-
opment of personal epistemology.
The Relationship between Personal Epistemology and Conceptual

Change
In the present volume we are not interested in investigating epistemological beliefs per se
but in understanding the role they play in conceptual change. Research has shown that a
constructivist epistemology is positively related to skills and attitudes important for learn-
ing, such as critical thinking, self-regulation, cognitive flexibility, the ability to communi-
cate ideas and to learn from collaboration. Previous studies have also demonstrated the
relationship between a constructivist epistemology and text comprehension. Ryan (1984)
found that students who are relativists in their beliefs about knowledge, i.e., who perceive
knowledge as context dependent, are more successful in comprehension monitoring and
tend to use high-level comprehension strategies. On the contrary, students who are dualists,
i.e., who perceive knowledge as factual, right or wrong, who are more likely to study for
recall of facts from texts. Naïve beliefs concerning the structure of knowledge — beliefs
concerning knowledge as an accumulation of discrete, concrete, knowable facts — have
been found to be related to poor text comprehension in the social and physical sciences, and
to affect the comprehension and related problem solving of statistical text in a negative way
(Schommer, 1990; Schommer, Crouse, & Rhodes, 1992). Naïve beliefs concerning the sta-
bility/certainty of knowledge, i.e., seeing knowledge as unchanging, have a negative effect
on the interpretation of controversial evidence (Kardash & Scholes, 1996) and tentative text.
Students who believe that knowledge is absolute and certain do not change their previous
views easily and tend to draw inappropriate absolute conclusions from tentative text.
(Schommer, 1990).
The theoretical position that we would like to advance is that epistemological beliefs
can have both a direct and an indirect influence on conceptual change. For example, beliefs
Personal Epistemology and Conceptual Change 101
in simple, stable, and certain knowledge can prevent individuals from being open to new
information that questions some of their basic assumptions, while on the contrary, indi-
viduals who believe that knowledge is complex, uncertain, and constantly evolving may be
willing to ‘open up the grammatical space’ and allow new paradigms/theories to be seri-
ously entertained. Epistemological beliefs can influence conceptual change also in indirect
ways, for example, by influencing students’ learning goals, study strategies, and self-
regulation. Some of these issues are addressed in the papers on epistemological beliefs and
conceptual change included in the present volume.
The Chapters in the Present Volume

The chapter by Murphy et al. examines how psychology researchers, implicitly or explic-
itly, convey epistemological beliefs in their conceptual change research. This is a new and
interesting topic of research. Murphy et al. justifiably claim that sometimes the researchers
involved in studying the epistemic beliefs of individuals have not examined their own epis-
temological stances. The attempt to bring some insights and knowledge from the philo-
sophical study of epistemology to bear on the psychological studies of epistemological
beliefs research is also worthwhile.
Nevertheless, a great deal of caution must be exercised with respect to the kinds of con-
clusions that can be drawn from these studies. More specifically, there is a danger here to
attribute to researchers the epistemic beliefs that the researchers attribute to the partici-
pants of their studies, as, for example, attributing to Vosniadou and Brewer (1992) the
foundational epistemic beliefs they attributed to their subjects. Furthermore researchers are
likely to have epistemic beliefs that cut across a number of the epistemic stances described
in the paper, whereas any single empirical study must by necessity concentrate on only a
few variables. Finally there may be important differences between the studies that attempt
to provide information about students’ epistemic beliefs as compared to interventions that
attempt to change these beliefs. In interventions, usually more attention is paid to socio-
cultural factors.
Entwistle (this volume) reviews a range of studies that show that beliefs about knowl-
edge and learning go through similar qualitative changes over time and that they eventu-
ally become integrated and affect the approaches to studying adopted by students. This
chapter takes a perspective on conceptual development and change that derives mainly
from research in teaching and learning at the university level, but it is also based on ideas
about the nature of concepts, epistemological development, and contextual influences on
the emergence and use of conceptions. The author concludes that only through a deep
approach to studying, an approach that links intention with learning processes, can stu-
dents achieve the kinds of conceptual understanding demanded by university degree
courses.
The chapter by Stathopoulou and Vosniadou follows as a nice continuation of the
Entwistle chapter. It is designed to further explore the relationship between personal epis-
temology and conceptual change in physics. The authors present results suggesting that a
constructivist epistemology may guide the adoption of a deep approach to studying, which
may in turn facilitate physics understanding. Previous research by Stathopoulou and
102 Stella Vosniadou
Vosniadou (in press) demonstrated that secondary school students with constructivist
physics epistemologies had better understanding of Newtonian dynamics compared to
their classmates with less constructivist physics epistemologies. In the present study, in-
depth interviews, think-alouds, and observations during problem-solving of 10 students
selected from the Stathopoulou and Vosniadou (2006) study were analyzed in terms of
their approaches to physics learning and studying. Five of these students were found to
hold constructivist physics epistemologies as well as a high degree of physics understand-
ing, whereas the remaining five did not have a constructivist physics epistemology and
were low in physics understanding. The interviews showed that the five students with the
constructivist epistemologies and high conceptual understanding of physics adopted a
deep approach to studying and learning, as described by Entwistle (this volume). In con-
trast, the remaining five students who did not have a constructivist personal epistemology
and were far from having achieved conceptual change in physics, adopted a superficial
approach to learning and studying.
The chapter by Mason and Gava also reports a study that investigated the relationship
between epistemological beliefs and learning from a refutational text. The results showed
that eighth graders with more advanced personal epistemologies (who believed that
knowledge is complex and uncertain) were able to profit more from reading a refutational
text compared to students with less advanced personal epistemologies.
Duit and his colleagues present data from a recent video study on the practice of intro-
ductory physics instruction in Germany. The analysis of these data showed that there is a
large gap between teachers’ beliefs about the nature of knowledge and how to teach it and
the actual use of such ideas in instructional practice. Teachers’ thinking about instruction
was oriented mostly towards the content to be taught. Most of the teachers were not
informed about recent conceptual change ideas or about the influence of epistemological
beliefs on students’ understanding.
In their commentary, Alexander and Sinatra raise a number of important theoretical and
methodological issues regarding epistemological beliefs research and the relationship
between individual’s epistemic beliefs and conceptual change. One important issue is the
lack of theoretical and terminological clarity. The discussants ask for greater explicitness of
definition of the terms used by researchers so that the field can eventually achieve more
solid conceptual and lexical foundations. They also discuss extensively the complexity of
the constructs involved and of their interrelationships, urging researchers to be more explicit
about the theoretical models they propose and to design studies that can further develop and
test these models. Finally, the importance of conducting more long-term developmental and
classroom-intervention studies using carefully designed measures of epistemic beliefs and
appropriate statistical procedures in order to be able to better understand the interrelation-
ship between personal epistemology and conceptual change is discussed.
Acknowledgments
work of the project ‘‘Pythagoras — Support of University Research Groups” with 75%
Personal Epistemology and Conceptual Change 103
References
Hofer, B. K. (2002). Personal epistemology as a psychological and educational construct: An intro-
duction. In: B. K. Hofer, & P. R. Pintrich (Eds), Personal epistemology: The psychology of beliefs
about knowledge and knowing (pp. 3–14). Mahwah, NJ: Erlbaum.
Kardash, C. M., & Scholes, R. J. (1996). Effects of preexisting beliefs, epistemological beliefs and
need for cognition on interpretation of controversial issues. Journal of Educational Psychology,
88, 260–271.
King, P. M., & Kitchener, K. S. (1994). Developing reflective judgement: Understanding and pro-
moting intellectual growth and critical thinking in adolescents and adults. San Francisco,
CA: Jossey-Bass.
Kitchener, R. (2002). Folk epistemology: An introduction. New Ideas in Psychology, 20, 89–105.
Kuhn, D. (1991). The skills of argument. Cambridge: Cambridge University Press.
Perry, W. C. Jr. (1998). Forms of intellectual and ethical development in the college years: A scheme.
San Francisco, CA: Jossey-Bass.
Ryan, M. P. (1984). Monitoring text comprehension: Individual differences in epistemological stan-
dards. Journal of Educational Psychology, 76(2), 248–258.
Schommer, M. (1990). Effects of beliefs about the nature of knowledge on comprehension. Journal
of Educational Psychology, 82, 498–504.
Schommer-Aikins, M. (2002). An evolving theoretical framework for an epistemological belief
system. In: B. K. Hofer, & P. R. Pintrich (Eds), Personal epistemology: The psychology of beliefs
about knowledge and knowing (pp. 103–118). Mahwah, NJ: Erlbaum.
Schommer, M., Crouse, A., & Rhodes, N. (1992). Epistemological beliefs and mathematical text
comprehension: Believing it is simple does not make it so. Journal of Educational Psychology,
84(4), 435–443.
Stathopoulou, C., & Vosniadou, S. (in press). Exploring the relationship between physics-related
epistemological beliefs and physics understanding. Contemporary Educational Psychology.
Chapter 10
Epistemological Threads in the Fabric of

Conceptual Change Research
P. Karen Murphy, Patricia A. Alexander, Jeffrey A. Greene and
Maeghan N. Edwards
Never as much as today have I thought that philosophy was indispensable

to respond to the most urgent questions of society…Never have we had such
a need for the philosophical memory. (Derrida, 2002)
Throughout this volume, the body of theory and research on conceptual change is dis-
cussed and debated by international scholars who have helped to shape this very domain
of study. Among those renowned scholars are individuals invested in understanding the
place of epistemic beliefs in the realization of conceptual change. What we hope to con-
tribute to this dialogue is a preliminary, yet critical, exploration of the philosophical
threads related to epistemology found in the fabric of conceptual change research. For this
initial epistemological foray into conceptual change research, we have chosen to examine
how psychological researchers implicitly or explicitly infuse epistemic tenets into their
conceptual change studies.
This theoretical analysis unfolded in several phases. First, we began with a review of
how philosophers convey the nature of knowledge. Second, based on that review, we
selected several prominent epistemological frameworks (e.g., foundationalism and relia-
bilism), and delineated their defining features (e.g., stances regarding knowledge, sub-
stantive conditions, and justification of knowledge). Next, we constructed prototypic
conceptual change studies representing selected epistemic belief frameworks. Essentially
those prototypes were examples of how the philosophical frameworks might manifest in
conceptual change research. Finally, we searched the literature on conceptual change for
evidence of implicit and explicit epistemic assumptions evident in the design and execu-
tion of research. That evidence is shared here in the form of selected conceptual change
studies that can be distinguished by their differential treatment of knowledge justification.
For clarity we have taken pains to distinguish between epistemic and epistemological
beliefs. The former are beliefs about knowledge, whereas the latter are beliefs about the
study of knowledge. Kitchener (2002) has made a strong argument for maintaining
this distinction. Students most likely have epistemic beliefs, beliefs about knowledge.

ISBN: 0-08-045355-4
106 P. Karen Murphy et al.
Many researchers have epistemic beliefs as well, such as what means of justification are
privileged over others. Finally, those looking at the implicit or explicit assumptions regard-
ing the study of knowledge are engaging in research on epistemological beliefs.
We consider this epistemological delving into the conceptual change literature worth-
while for several reasons. First, there would appear to be an inherent relation between epis-
temic beliefs and conceptual change. That is, if one is to investigate the nature and process
of change in conceptual understanding, it would seem that some judgments as to the nature
or source of knowledge must be assumed by those conducting this research. Second, we
are of the mind that all psychological research, including the conceptual change literature,
would benefit from rediscovering its philosophical roots (Murphy, 2003). Much of con-
temporary educational research seems almost devoid of an awareness of the foundational
literature in philosophy from which it was birthed. As Derrida (2002) described, we need
a philosophical memory, especially when trying to examine some of the more urgent ques-
tions of society, such as conceptual change. A caveat, of course, is that although philo-
sophical memory and even current philosophical thinking can inform theory and
interpretation, it cannot replace empirical evidence. Such empirical evidence is vital in
understanding the nature of the conceptual change process. This is particularly true in the
fields of cognitive and educational psychology due to the epistemic importance of empir-
ical evidence as a form of justification (Southerland, Sinatra, & Matthews, 2001).
Third, and most relevant to this volume, we have found ourselves enticed by the bur-
geoning psychological literature on epistemic beliefs. In many ways, the psychological
literature on epistemic beliefs functions independent of substantive influence from phi-
losophy. Like others, we posit that this independence is due, in part, to variations in the
central questions guiding these disciplines. That is, philosophers are intrigued by what it
means to know, whereas psychologists are more concerned with how one comes to know
(Pollock & Cruz, 1999). Despite the subtle variations in focus, we maintain that there is
much to be learned from the philosophical annuls pertaining to the theoretical study of
epistemology. This sojourn into the conceptual change literature will cast a discerning eye
on the epistemic aspects of studies, including our own, populating the pages of psycho-
logical journals. Reciprocally, by bringing epistemic beliefs into the limelight, we may
furnish conceptual change researchers with a new lens for examining their work.
Framing the Epistemology Literature
Epistemology is a core area of inquiry within philosophy (Moser, 1995). Consequently,

the philosophical literature pertaining to epistemology and its central tenets is both vast
and varied. Nonetheless, there are defining questions that characterize the philosophical
literature on epistemology. To our understanding, those defining questions are: What is
the nature of knowledge; what is the source of knowledge; and what are the limits of
knowledge (Bonjour & Sosa, 2003; Pollock & Cruz, 1999; Moser, 1995)? This charac-
terization of epistemology does not include skepticism, the belief that it is impossible to
“know” anything. While this is an interesting philosophical argument, in terms of our
focus on conceptual change, it is of little merit. An epistemological skeptic would not
only denounce our attempts to understand knowledge, but also question the rationale for
Epistemological Threads in the Fabric of Conceptual Change Research 107
facilitating conceptual change. The skeptic would ask how any ethical human being could
assert that another’s concepts were in need of changing, when knowledge is not possible.
Indeed, many modern philosophers see skepticism as a useful tool for testing the
robustness of epistemological stances, but do not give it serious consideration as an actual
explanation of knowledge or knowing (Pollock & Cruz, 1999; Quinton, 2001; Williams,
2001). Given the skeptics’ refutation of both the content and the analytic lens of this chap-
ter, we set aside the skeptical question as intriguing but tangential to our purposes. Thus,
although the three questions about knowledge have relevance for educational research,
understanding the nature of knowledge is particularly important in the study of conceptual
change. As such, we set ourselves the task of clarifying this defining question.
Our approach to this challenging exercise was multilayered. We first turned to three
well-respected encyclopedias on philosophy: the Routledge Encyclopedia of Philosophy
(Craig, 2004), the Cambridge Dictionary of Philosophy (Moser, 1995), and the Stanford
Encyclopedia of Philosophy (Bradie & Harms, 2004; Feldman, 2001; Fumerton, 2000;
Goldman, 2001; Greco, 2002; Klein, 2001; Kvanvig, 2003; Steup, 2001; Young, 2001).
These volumes are frequently required readings for entry-level courses in philosophy, and
served as clear and concise summaries of the boundless library of classic and contempo-
rary writings on epistemology. These encyclopedias proved useful to us in clarifying and
extending our prior knowledge of epistemology, and in identifying other sources that
functioned as yet another layer of essential discourse to be analyzed and synthesized. For
example, in reading the Stanford Encyclopedia we were directed to Ernest Sosa’s two
edited volumes on Knowledge and Justification. These tiers of sources were invaluable in
conceptualizing eight principal stances in epistemology:
1. Foundationalism, which holds that all knowledge is derived through ascent from basic
beliefs internal to the knower;
2. Coherentism, which sees all of one’s beliefs as mutually reinforcing and thereby justi-
fied as knowledge only in their mutual coherence;
3. Direct Realism, which states that justification cannot be solely a function of beliefs
about perceptions but a function of the perception themselves;
4. Probabilism, which assesses the cognitive mechanisms that produce beliefs in terms of
their ability to produce knowledge with a high likelihood of being true;
5. Reliabilism, which judges the veracity of knowledge based on the cognitive mecha-
nisms’ reliability in producing true beliefs;
6. Social epistemology, which focuses on social practices and their influences on one’s
beliefs about knowledge;
7. Evolutionary epistemology, which emphasizes the process of natural selection in the
formation and maintenance of justified true belief; and
8. Virtue epistemology, which focuses on the character of the knower rather than individ-
uals’ beliefs or collections of beliefs.
Next, we narrowed our focus to four epistemological frames that appear most pertinent
to psychological research, in general, and the conceptual change literature, in particular:
foundationalism; coherentism; reliabilism; and social epistemology. We chose to narrow
our focus for two reasons. First, there are simply too many epistemological frames stud-
ied in philosophy, and a cogent presentation of them would not be possible within one
NORMATIVE
Foundationalism STANCES
DOXASTIC
STANCES (Internalist
Coherentism Theories)
Direct Realism
Virtue Epistemology
NON- Probabilism
NATURALIST
DOXASTIC
Reliabilism STANCES
STANCES
Social Epistemology (Externalist
Theories)
Evolutionary
Skepticism
Figure 10.1: Relational model of epistemological stances.
manuscript. In addition, some of the epistemological frames seemed less relevant to aca-
demic learning or are less well developed in the philosophical literature. In reviewing lit-
eratures on each of these frames as well as literature comparing the frames (e.g., Moser,
1995; Sosa, 1994), it is important to establish a priori that despite the fact that we discuss
each of the frames separately, we do not view the frames as conceptually or theoretically
mutually exclusive. From a conceptual or theoretical perspective, there are overlaps in
nature of justification (e.g., doxastic vs. non-doxastic) and the source of justification
(e.g., internal or external). Such similarities or overlaps are illustrated in Figure 10.1.
From an operational perspective, it would seem that contextual and social factors could
strongly influence the predominant epistemic frame enacted at any particular time.1
For example, an individual might enact a reliabilist frame in regard to their scientific
research while holding to a foundationalist perspective in their personal religious life.
An Epistemological Primer
Prior to exploring the aforementioned epistemological frameworks, we briefly discuss

the philosophical analysis of knowledge. Like psychologists (e.g., Alexander, Schallert,
& Hare, 1991), philosophers distinguish between many kinds of knowledge including
empirical (a posteriori), nonempirical (a priori), knowledge by acquaintance, knowledge
by description, and “how to” knowledge. The kinds of knowledge with which psycho-
logists are most familiar would be “how to” knowledge (i.e., procedural knowledge) and
knowledge by description (i.e., propositional or declarative knowledge). It is propositional
knowledge that is typically the crux of epistemological inquiry; that is, what does it mean
to say that an individual knows something to be the case?
1
The issue of contextual and social influences on the enactment of epistemic beliefs was raised by Stella
Vosniadou during revisions of the chapter.
Formal logic is the means by which epistemological theory is understood and

tested, and those propositions can be expressed in the form of syllogisms of the kind
(DeRose, 2004):
1. P is true,
2. S believes that P, and
3. S is justified in believing that P.
This standard analysis of knowledge claims that truth is
necessary for knowledge, but it is not sufficient. If one is to know that p,

not only must it be true that p: one must also believe that p, at least implic-
itly. (Sosa, 1994, p. xi)
In essence, the nature of knowledge can be broken down into three distinct, yet related,
conditions: the belief condition, the truth condition, and the justification condition.
In terms of the belief condition, most epistemologists agree that knowledge requires
belief but belief does not require knowledge (Moser & vander Nat, 2003). One knows
something only if one believes it, but not all things that are believed are in fact knowledge.
Knowledge also requires truth and justification.
You know that all college professors are brilliant only if it is true that all
college professors are brilliant. If there is one dull college professor, you do
not know that all college professors are brilliant. Knowledge thus has a truth
requirement. (Moser & vander Nat, 2003, p. 6)
Most epistemologists agree that even justified belief is not knowledge unless it is, in
fact, true. One might be epistemically responsible in the means by which one justifies a
belief, but still be incorrect. In this case, the believer is not faulted, but he or she does not
retain the status of having knowledge. It would be expected that the believer would
acknowledge such when shown conclusive contradictory evidence.
The belief and truth conditions are fairly standard aspects of most epistemologies. Most
epistemological stances vary in terms of how they treat the justification condition. The first
differentiation concerns whether or not those stances posit that beliefs about perceptions
are enough to justify knowledge. Doxastic views contend that the only viable and suffi-
cient means of justification derive from the beliefs that one holds; specifically, beliefs
about perceptions (Figure 10.1).
Non-doxastic views agree that perception beliefs are important, but insufficient for jus-
tification. Instead, non-doxastic views require an examination of the cognitive processes
from which knowledge derives. Within the doxastic tradition, there are two fundamental
schools (i.e., foundationalism and coherentism). Foundationalists assert that there are cer-
tain self-evident foundational beliefs that require no further justification. It is from these
beliefs that other knowledge claims are built through the process of epistemic ascent. For
example, sense data about the behavior of clouds might lead to conclusions about the
mechanics of wind. Coherentism agrees that knowledge claims are built from beliefs and
perceptions, but denies that there are certain foundational claims; rather all beliefs must be
judged correct through an evaluation of their coherence with other beliefs. Coherent sys-
tems are knowledge, incoherent systems are not. A coherent system would include a set of
perceptions that are non-contradictory such as seeing, smelling, and tasting the presence
of a piece of cake. Despite their differences, both foundationalism and coherentism rely on
internal sources of justification. As such, one might expect a certain degree of overlap
between these frameworks when evidenced in research and practice. For example, it seems
reasonable that a researcher or teacher could value both the power of students’ experien-
tial beliefs as well as the coherence among those beliefs.
However, doxastic epistemological theories are subject to a serious criticism in the form
of the Gettier problem (Moser, 1995; Pollock & Cruz, 1999; Quinton, 2001). Prior to
Gettier (1963), all epistemology was doxastic, claiming that knowledge was justified true
belief. However, Gettier demonstrated that these conditions were not sufficient. For exam-
ple, we can imagine a woman named Barbara who has good reason to believe a nonethe-
less false proposition that John owns a radio (i.e., she saw John carrying a radio). Barbara
also chooses to guess where another person named Mike is, such as in Florida, knowing
that she has no reason to believe it to be true. Coincidentally, Mike is in Florida. Now,
Barbara is justified in believing the disjunction “either John owns a radio or Mike is in
Florida.” Here, it is obvious that while this statement is true, we would not say that Barbara
knows either that John owns a radio or that Mike is in Florida. This and other challenges
to doxastic theories, while seemingly easy to resolve, have been a persistent and tenacious
problem. Some epistemologists have suggested that the Gettier problem illustrates that the
justification, truth, and belief conditions are not sufficient for knowledge. Their solution
has been to institute various other kinds of additional conditions beyond justified true
belief. Epistemic frames advancing conditions for knowledge beyond justified true belief
are referred to as non-doxastic.
Non-doxastic views divide into two main schools based upon how they choose to bol-
ster the justification condition (i.e., internalism and externalism), while foundationalism
and coherentism rely on beliefs only. Internalists judge the validity of cognitive processes
based upon preestablished epistemic norms that are employed to judge the viability of a
particular cognitive process to serve as justification. For example, reasoning would serve
as justification for knowledge, but guessing would not be an acceptable form of justifica-
tion. Externalists determine the validity of cognitive processes based on external metrics.
Naturalists, for example, believe that cognitive processes should be compared to scientific
theories and data to determine their validity. Among the naturalists, reliabilists assess cog-
nitive processes based upon how effectively they predict events or data in the real world,
thus they employ an external source or criterion for justification of cognitive processing.
Social epistemologists also assess the predictive reliability of cognitive processes, but do
so through the testimony of established authorities and beliefs institutionalized within a
particular culture or community of practice.
Overlaying Epistemological Stances on Conceptual Change Studies
Given this epistemological primer, we turn to four epistemologies prevalent, yet implicit
and unexamined, in the conceptual change literature. As a means of introducing these four,
we have constructed prototypic studies of conceptual change and describe why their fea-
tures coincide with the defining characteristics of each. To make the distinctions between
these studies more transparent, we will hold certain elements consistent. Specifically, our
conceptual problem across the four hypothetical studies will be Galileo’s Principle of
Falling Bodies and our participants will be eighth graders of normal cognitive ability
representing diverse sociocultural backgrounds.
Foundationalism
Because of the importance of students’ self-justifying, core beliefs, and the role that per-
ception plays in foundationalism, we would want to initiate this study by pre-assessing
students’ beliefs relative to the Principle in some detail. Moreover, we would want to cre-
ate an intervention that explicitly confronts students’ unjustified beliefs through sensory
experience. Thus, in the pre-assessment, we might have students make initial predictions
about the outcome of dropping two objects of different sizes and weights, and then
require them to explain those predictions; exposing their self-justifying core beliefs. For
example, students may be asked to watch a simulation that depicts a bowling ball and a
small rubber ball being dropped from a tower. Before the balls hit the ground, the simu-
lation is stopped. Students would then be asked to describe what would happen and why.
This process would be repeated with objects varying along perceptual dimensions. In all
cases, the goal would be for students to render a prediction and to explain that prediction.
Once the pre-assessment is complete and students’ misperceptions are exposed, an
intervention would be initiated that would be multisensory in nature. That is, we would
present a number of predictive events that convey the Principle in a way that draws sepa-
rately on sensory data. For example, as in the pre-assessment task, students might be asked
to watch a simulation similar to that viewed earlier and witness the outcome in slow
motion. Likewise, the students might close their eyes and listen intently to what they hear
when a water-filled balloon and a glass are similarly dropped from a set height. As before,
the outcomes would be discussed in terms of any initial self-justifying beliefs. Following
a delay, students would again be shown the pre-assessment simulation and predict the out-
come. The explanations for their predictions would provide evidence of conceptual change
relative to the target Principle.
Coherentism
What differentiates coherentism from foundationalism is the lack of foundational beliefs;

thus, any intervention designed to change a concept or mental model must instead
address multiple beliefs in an attempt to invalidate the coherence of the system.
Consequently, rather than rely solely on the previous simulation to alter students’ con-
ceptions of the Principle, we might employ a structured inquiry during the simulation to
uncover multiple misperceptions within the network and then provide evidence contrary
to those misperceptions. For example, because mass, velocity, and acceleration are vari-
ables intertwined within the understanding of the Principle, we might design interven-
tions to correct misperceptions in any of these areas. Thus, in the simulation we might
first attempt to isolate the effect of velocity. Students would then collect data relative to
velocity during the interventions. Similarly, we could run specific simulations for which
mass or acceleration is the key variable. Moreover, because we are concerned about the
relations among the beliefs that are part of this system, we would also seek to address the
seeming inconsistencies across these key concepts and their relations to each other by
requiring students to explain the effect of mass in relation to velocity, or acceleration in
association with mass and velocity. The coherentist assumption would be that a “tipping
point” would be reached where the contrary evidence would outweigh the coherence of
the misperceived system, thus allowing for instantiation of new beliefs.
Reliabilism
A reliabilist would approach conceptual change differently. Since reliabilism is an exter-

nalist theory, one necessary criterion for justification would be the quality of the cognitive
processes used in substantiating belief, where the validity of the cognitive processes would
be based on external metrics (e.g., comparisons to scientific data). Therefore, a reliabilist
intervention could depend upon experimental data that are beyond human perception
(e.g., reasoning) to illuminate students’ misperceptions so long as the cognitive processing
effectively predicts events or data in the real world. For example, scientific theories could
be gleaned from the subject’s misperceptions and empirically tested. When those theories
fail to reliably produce accurate predictions, the experimenter could facilitate conceptual
change by presenting principles or accurate scientific theories to the students. The modified
or newly formed theories could then be experimentally tested. When the results demonstrate
the reliability of the newly presented theory, students could accrete that theory through cog-
nitive processing. This cognitive processing of the new theory could be enough to foster
conceptual change in the reliabilists’ view. Thus, the reliabilist has the ability to address
conceptual change through perceptual data and scientific hypothesis testing that is not
directly observable. Whether a student would find this convincing is a separate matter, but
the externalist epistemology allows for greater variety in the ways conceptual change could
be invoked. In essence, while doxastic theories are “I’ll believe it when I see it” theories,
non-doxastic theories, such as reliabilism, allow for the possibility that more abstract evi-
dence may be convincing.
Social Epistemology
Social epistemologists not only acknowledge the importance of the natural sciences in the
justification of beliefs, but also consider the influence of social and cultural forces as
equally privileged mechanisms of justification. Therefore, in planning a prototypic study
of conceptual change from this perspective, several features would likely become impor-
tant. For one, in the process of exposing students’ pre-existing beliefs, we would not only
want to determine what students’ hold to be true about the Principle of Falling Bodies, but
we would also need to ascertain the source of those beliefs. Moreover, in the pre-assess-
ment, we would want to learn what sources of evidence (e.g., known experts or written
text) students deem more convincing or authoritative. It might be that the experimenter is
not seen as a valid source of new information, and therefore, another method should be
used to promote conceptual change.
In addition, while we could use the types of simulation described under reliabilism,
we would incorporate more cooperative or collaborative group discussions to introduce
alternative evidential sources that students might perceive as more convincing. Further,
when it comes to the post-test, we would be interested in the changes in understanding
that students manifest about the Principle, and concerned with any demonstrated
changes in their sources of evidence or shifts in the perceived credibility of sources.
Prototypic Conceptual Change Investigations
Given the extensiveness of the conceptual change literature and in light of our stated
intention to forward a preliminary analysis, we set out to narrow the research base to
those articles likely to display epistemological threads. Thus, we first conducted a broad
search employing several electronic databases including PsycINFO, ERIC,
Philosopher’s Index, and Education Abstracts. Our initial parameters were reference to
conceptual change in title, abstract, or in key concepts, and a publication or release date
of 1999 or later. We also conducted a search by author for the past 10 years, including
the names of six researchers known for their work both in conceptual change and epis-
temology. Those authors were Stella Vosniadou, Clark Chinn, William Brewer, Lucia
Mason, Gale Sinatra, and Janice Dole. This process resulted in an initial pool of almost
400 works. To narrow this pool, we established six key inclusion criteria. Specifically,
we decided to focus on those works that: (a) mentioned conceptual change in title or
in abstract; (b) were empirical in nature (i.e., no research summaries or theoretical
analyses); (c) could be fully retrieved from the archived source (i.e., no summary of con-
ference proceedings or dissertation abstracts); (d) were not solely about teacher concep-
tual change; (e) incorporated some learning outcome; and (f) were published in English
(our linguistic limitations at cause).
By applying these criteria, we succeeded in narrowing the pool to 90 works. Of these,
we selected 40 of the most promising pieces for subsequent analysis. In particular, we
chose those works that were written by well-known conceptual change researchers and
were published in leading general educational research journals (e.g., Cognition and
Instruction, Journal of Educational Psychology, or Contemporary Educational
Psychology). Our goal was to locate studies seemingly well matched to the epistemologi-
cal frameworks we identified. Ultimately, for this chapter, we engaged in detailed analy-
ses of six articles that afforded us an opportunity to view the conceptual change research
through an epistemological lens. Those six studies are: Vosniadou and Brewer (1994),
Chinn and Brewer (2001), Nussbaum and Sinatra (2003), Mason (1996), Qian and
Alvermann (1995), and Hofer (2004).
We acknowledge that the categorizations posed herein are highly inferential, since none
of the authors freely exposed their epistemic leanings for our benefit. In essence, our goal
was to look for evidence of similarities between the aforementioned philosophical, episte-
mological stances and the views of selected researchers in a particular study. Also, in all
the studies perused, the researchers’ lens was focused on the manifestations of students’
epistemic cognition and not on their own epistemic beliefs. Still, we considered the goals,
design, methodology, and data sources of these studies to be vestiges of the researchers’
epistemic views. As such, there is the possibility that a given researcher might emphasize
or expose different epistemic beliefs in one study than another. Work by William Brewer
is a good case in point. Specifically, in the pages that follow, we categorized his work with
Vosniadou (i.e., Vosniadou & Brewer, 1994) as evidencing threads of foundationalism,
whereas we interpret his work with Chinn as more closely aligned with coherentism
(Chinn & Brewer, 2001). In both cases, however, we see evidence of a focus on beliefs
internal to the knower (i.e., doxastic).
Vosniadou and Brewer (1994)
Stella Vosniadou and William Brewer (1994) offer an analysis of first-, third-, and fifth-
grade children’s understandings of the day/night cycle (e.g., disappearance of the sun
during the night, disappearance of the stars during the day, or apparent movement of the
moon). Their results revealed that children used a collection of well-defined mental
models to explain those various phenomena. The mental models were empirically accu-
rate, consistent, and parsimoniously explained. Younger children constructed intuitive
mental models based on beliefs about prior experiences and situational factors, whereas
older children constructed mental models that synthesized aspects of culturally accepted
scientific information with prior knowledge, and a few students constructed culturally
accepted scientific mental models as evidenced in their verbal protocols and pictorial
representations.
Traces of foundationalism can be seen in Vosniadou and Brewer’s theoretical frame-
work. For example, the researchers assumed that children possessed beliefs on the basis of
their interpretations of their everyday experiences.
Children are not blank slates when they are first exposed to the culturally
accepted, scientific views, but bring to the acquisition task some initial
knowledge about the physical world that appears based on interpretations
of everyday life. (p. 124)
Moreover, they posited that children’s beliefs have a hierarchical structure in which
some beliefs are foundational (i.e., presuppositions), while other beliefs can be conceptu-
alized as second-order.
Presuppositions may be innate or empirically acquired constraints which

are present from early infancy and which guide the way children interpret
their observations and the information they receive from the culture to con-
struct knowledge structures. (p. 124)
Within Vosniadou and Brewer’s frame those presuppositions and second-order

beliefs served as constraints that must be exposed, and potentially lifted, during the con-
ceptual change process. Given the strength of some presuppositions and second-order
beliefs, students will sometimes form alternative models that incorporate aspects of
their presuppositions. Indeed, Vosniadou and Brewer have termed such models synthetic
because they are formed precisely because presuppositions cannot be removed or
replaced through exposure to cultural information during instruction about scientific

phenomena.
Conceptual change is seen as the product of the gradual lifting of con-

straints, as presuppositions, beliefs, and mental models are added, elimi-
nated, or revised during the knowledge acquisition process. (p. 124)
The purpose of this particular study was to document mental models of the day/night
cycle of various aged children, and to compare the variations in those models to the
hypothesized transition from intuitive models to scientific models.
We also see evidence of coherence theory in this study. For instance, the researchers
were interested in the coherence between children’s explanations of various individual
phenomena (e.g., sun, moon, stars) and their relations to the overarching process of the
day/night cycle. Whether foundationalism or coherence theory, it is clear that Vosniadou
and Brewer’s mental models framework is doxastic in nature. This understanding is con-
firmed by the presence of the Gettier problem in the results. That is, the children seemed
to offer what they perceived as justified beliefs. But the beliefs were false because they
were based on inaccurate perceptions gleaned from everyday experiences. Children
believed
…from their observations that the day/night cycle is causally related to the
appearance and disappearance of both the sun and the moon. The erroneous
belief that the moon is causally implicated in the day/night cycle appeared
to be present in the synthetic models of the oldest children in our sample.
(p. 171)
The researchers go on to state that they do not understand why these beliefs persisted.
The Gettier problem seems to shed light on this situation. These students were justifying
their beliefs based on false or coincidentally true beliefs, and philosophically speaking,
such instances should not be considered knowledge.
Chinn and Brewer (2001)
The purpose of this research was to test a theory (i.e., models-of-data) of how students eval-
uate data. The models-of-data theory is intended to provide an account of both how students
evaluate data and of how they design research studies such as experiments. Chinn and
Brewer (2001) hoped to categorize the types of reasons given to explain why students
accept or reject aspects of data. To investigate students’ reasons, these researchers asked 168
undergraduate students to read one of the two theoretical texts on either mass extinction or
dinosaur metabolism. After reading, students rated the extent to which they believed the the-
ory. After rating the initial theory, students read about data either supporting or refuting the
original theory. Then, students rated their beliefs in the data and explained their beliefs.
Finally, students offered ratings of the consistency between the theory and the data, as well
as explanations of their data. Chinn and Brewer found that students’ ratings and explana-
tions generally upheld the models-of-data theory. That is, these researchers found that when
evaluating data, individuals created a cognitive model that integrated features of the data
with a theoretical interpretation of the data.
In essence, the strength of Chinn and Brewer’s present work is its modeling of how
people interpret data and evidence. What is important for our purposes are Chinn and
Brewer’s theoretical assumptions in their models-of-data, as well as their interpretations
of the findings. What we saw in both their theoretical framework and their interpretation
of the outcomes was evidence of a coherence approach to epistemic cognition. Like
coherence theory, Chinn and Brewer’s model focuses on the relation between beliefs and
their place in a comprehensive system of knowledge.
[W]e propose that when people evaluate data, they construct a cognitive
model of the data according to the perspective of the person who is report-
ing the data....The individual seeks to undermine one or more of the links
in the model, often by seeking alternative causes that individuate particular
links. If the individual succeeds in identifying a plausible alternative cause
for an event, the data can be discounted. (p. 337)
This notion of examining the links between various beliefs is at the heart of coherence
theory and Chinn and Brewer’s research findings.
A model of data is evaluated by examining the plausibility of the links in

the model. (p. 334)
Nussbaum and Sinatra (2003)
Argumentation was used as a means to encourage cognitive engagement in Nussbaum and

Sinatra’s (2003) study of conceptual change. Participants were asked to guess the path of
an object dropped from various moving contexts such as a plane or a conveyor belt.
Participants who did not predict a Newtonian-path for the object were prompted to argue
for “another option” that was in fact the scientifically correct one. The authors found that
subjects who argued for the scientific option were more likely to revise their initial guess
to one more scientific as compared to a control group. This change in conception was
hypothesized to be due to the degree of cognitive engagement involved in the argumenta-
tion intervention.
The Nussbaum and Sinatra (2003) article demonstrates a reliance on not only natura-
listic empiricism, but also the power of individuals’ ability to critique their own cognitive
processes. The authors discussed a commitment to fostering scientific thinking, conduct-
ing experiments, and resolving discrepancies between points of view and data. The
researchers’ focus on science places them within a naturalistic epistemology. Additional
evidence of a naturalistic approach to epistemic cognition can be found in the fact that dur-
ing the multiple iterations of the falling object simulation, participants were asked to eval-
uate the reliability of their underlying cognitive activities. Thus, we believe the Nussbaum
and Sinatra research was derived from a reliabilist epistemology. What is significant in this
categorization is that it appeared that the authors believed conceptions change through
introspection and examination of the internal logic of one’s conceptions. Yet, Nussbaum
and Sinatra attributed conceptual change to argumentation, as opposed to the kinds of

cognitive dissonance that such argumentation most likely invokes in the subject who ini-
tially guessed incorrectly. In essence, they are describing the manner in which analysis of
cognitive processing is encouraged, but they do not mention the actual analysis itself.
Mason (1996)
In the context of a larger two-year study, Mason (1996) attempted to explicate students’
processes in making justified scientific claims in a group discussion. Students in five fifth-
grade classrooms discussed different environmental education units in both whole-class
and small-group settings in order to construct new scientific knowledge. Mason identified
students’ scientific claims and justifications for those claims in their group discourse. After
categorizing students’ appeals to personal experiences, accepted scientific principles,
analogies, generalizations, and other relevant data, Mason concluded that the students in
the discussions were able to learn from each other by negotiating and sharing meanings.
We hold that this study and its underlying theoretical framework offer threads of social
epistemology. This framework differs from reliabilism in that the justifications formed by
the students do not come solely from the reliability of their own cognitive processes.
Instead, Mason posited that knowledge could be constructed with and through others, with
or without an appeal to one’s own cognitive processes. Mason specifically expressed this
social epistemic framework by stating that not only can students build knowledge with
each other, but that they can also use the same social interactions with their peers as justi-
fications for their new conceptions.
When a shared idea passes from mind to mind, co-constructed knowledge

becomes stronger and deeper than the knowledge transmitted in traditional
science classrooms. (p. 427)
Hofer (2004) and Qian and Alvermann (1995)
Both Hofer’s (2004) and Qian and Alvermann’s (1995) studies highlight differences
between philosophical and psychological definitions of epistemology. Hofer (2004) per-
formed exploratory qualitative analyses with students in two different chemistry courses.
In each course, student interviews and class observations provided data used to analyze the
underlying epistemic assumptions fostered by instruction. Hofer’s goal was to “examine
how beliefs about knowledge and knowing are communicated in the classroom and how
they are situated in classroom interaction” (Hofer, 2004, p. 135). Ultimately, she hoped to
describe the evolution of students’ epistemic understanding in a contextualized learning
environment. Hofer categorized the epistemic beliefs into dimensions including the cer-
tainty of knowledge, the simplicity of knowledge, the source of knowledge, and the justi-
fication for knowing, based upon Schommer-Aikins’ work (e.g., 2002). Thus, Hofer
asserted that she was able to substantiate the existence of dimensions and to explicate how
they were instantiated in the classroom and internalized by students. One of Hofer’s main
points was that students inferred epistemic assumptions of instructors and milieus, and in
so doing, acquired learning in ways that were contextualized and situated.
Qian and Alvermann (1995) examined the relation between learned helplessness, epis-
temic beliefs, and conceptual change within the domain of science. A group of 212 high
school students were classified according to their level of prior knowledge about Newton’s
Theory of Motion. The authors examined how learned helplessness and epistemic beliefs
differentially impacted students who varied in their level of prior knowledge. Refutational
text on Newton’s Theory was employed as the intervention, and students were given a sub-
sequent post-test to measure conceptual change.
Qian and Alvermann (1995) found three epistemological (as they termed them) belief
factors: quick learning, the simplicity and certainty of knowledge, and the innate nature
of learning ability. The researchers found little evidence of any relation between learned
helplessness and epistemic beliefs, and the only factors significantly associated with con-
ceptual change were simple/certain knowledge and quick learning factors, both of which
were negatively related. In prior knowledge groups, significant gains in conceptual
change were realized but there was no interaction with epistemological beliefs or learned
helplessness. Overall, Qian and Alvermann found support for the three latent factors that
they termed epistemological beliefs, and they found that refutational text facilitated con-
ceptual change among learners.
We would submit that both of these studies have more psychological than philosophical
views of epistemic cognition. Pollock and Cruz (1999) discussed a fundamental difference
between philosophical and psychological analyses of epistemology. While philosophers are
concerned primarily with the question “What is it to know?” psychologists are interested in
the range of ways in which humans acquire and justify their knowing. Therefore, psycho-
logical analyses of epistemic cognition should capture the variability people display. For
example, while the psychological literature on personal epistemology discusses student
assumptions about knowledge, such as the limits and simplicity of knowledge (Hofer, 2004;
Qian & Alvermann, 1995), the philosophical literature seems unconcerned with such issues
(Pollock & Cruz, 1999; Sosa, 1994).
Concluding Thoughts and Future Directions
As we stated at the outset, we engaged in this process of tracing the threads of episte-
mology in conceptual change research with an enthusiasm that can only be born from
naïveté. Had we realized the truly daunting nature of the task we had set for ourselves,
we may not have been willing to initiate the pursuit. Nonetheless, we have personally
been well served by the endeavor, learning much about the ideas and issues that reside at
the juncture of epistemology and conceptual change. We also believe that our analysis can
prove informative to those invested in understanding conceptual change or sparking con-
ceptual growth in others.
First, among the insights we derived from this venture is the realization that within the
conceptual change literature, the epistemic threads are often so buried as to be unidentifiable.
It might be argued that conceptual change researchers are not required to make their epis-
temic views public or even to acknowledge the vast literature in epistemology or some may
maintain that interpretations of their views are likely dependent on the nature of the selected
study. Yet, we maintain our initial argument that those seeking to change knowledge or
beliefs in others must themselves possess a tacit, if not explicit, conception as to what it
means to know or some grounded notions of what constitutes knowledge.
Second, while we retain our belief that conceptual change researchers would benefit
from rediscovering their epistemological roots, we acknowledge that this wading into the
philosophical waters might be tantamount to an arctic dip. For those who are unfamiliar
with contemporary philosophical writings or who have not revisited classic works, dealing
with the terminology and discourse of this literature could prove a shock to the system. As
with any complex field, philosophers have their own lexicons and discursive patterns that
seem strange and uninviting to the casual reader. We would welcome those rare souls who
are equally at home in both the psychological and philosophical literatures, and who are
able to weave the theories and research of one community into the fabric of another.
Perhaps in this way, our desire to see deeper and richer threads of epistemological thought
in the conceptual change research might be realized.
Finally, as suggested, there appears to be two rather distinct streams of epistemic
inquiry that only sporadically intertwine. On the one hand, there is philosophical episte-
mology that utilizes formal logic and that coheres around the three basic questions of what
is knowledge, particularly the justification condition: what are its sources; and what are its
limits; which are primarily questions about the nature of knowledge? On the other hand,
there is psychological epistemology that seems more invested in the acquisition of knowl-
edge or, in the case of conceptual change research, in changing knowledge. The inherent
nature of knowledge is either taken as a given or considered tangential to the agenda at
hand. Thus, we focused here on the justification condition as it is the most relevant to the
conceptual change literature. However, researchers would do well to survey other condi-
tions and explore how their assumptions about knowledge influence their work.
Even given these conclusions, we are left with the question as to whether or not most
psychologists are aware of their underlying epistemic stances. If they are not aware, then
the authors themselves are unsure how their stance might manifest in their conceptual
change work. It could be that researchers posit a different epistemic stance for themselves
than they do for those they study. For example, most conceptual change researchers, in our
view, espouse a naturalist epistemology as the most correct. This implicit assumption
could manifest in two different but equally questionable ways.
First, it could be the case that a naturalist researcher sees naïve student conceptualiza-
tions (e.g., Vosniadou and Brewer, 1994’s findings regarding children’s belief that the sun
goes to sleep behind a mountain at night) as an instance of a “less-developed” epistemic
stance, such as foundationalism. Such beliefs about the sophistication of epistemic stances
and implicit epistemologies have been raised by philosophers such as Pollock and Cruz
(1999) in their suggestions regarding the critiques of doxastic theories using the Gettier
problem, and non-doxastics’ superiority in their ability to deal with such challenges. Thus,
a psychological researcher with this set of implicit epistemic assumptions would see con-
ceptual change development as an analogue to development in the philosophical episte-
mology literature itself, moving from doxastic to non-doxastic theories.
However, a second way in which a psychologist’s implicit epistemological stance
could manifest would be in interpreting all data from that point of view. In this example,
a naturalist researcher, using a reliabilist method of establishing justification, would char-
acterize the Vosniadou and Brewer (1994) example not as an instance of foundationalist
thinking but rather as an improper cognitive mechanism lacking sufficient reliability in

testing. There are two empirical questions here. First, do researchers assume that people
can hold different epistemic stances, and second, how do students justify their knowledge,
via simple perception, cognitive mechanisms, or another way? Obviously, the answers to
these questions would have clear ramifications for the manner in which conceptual
change was studied and what kinds of interventions would be appropriate to foster
change.
There is an obvious danger in attempting to categorize the epistemological underpin-
nings of researchers’ work. A more valid and collaborative next step would be to interview
selected researchers to expose qualitatively their epistemic and epistemological beliefs.
Not only would this help us understand the topology of the conceptual change landscape,
but it might also afford us insights into the kinds of stances held by experts in the domain
of psychology. Likewise, a more comprehensive review of the 90 articles we initially iden-
tified would give us further evidence of the variability of epistemic and epistemological
assumptions within the psychological literature. It could be that individual studies have
varying degrees of each epistemology discussed herein. To these future studies, we would
add a review of the major models of conceptual change as a means of juxtaposing these
researchers’ theoretical premises with the conceptualizations and operationalizations that
emerge in their empirical studies. Our goal would be to subject the conceptual change
research to a test of internal consistency.
There may be those who would appreciate our efforts to delve into the philosophical lit-
erature and seek to weave its elements into the fabric of conceptual change research, but
others might perceive this as a futile undertaking with little to contribute to the real prob-
lems of learning and teaching. To the contrary, the eloquent words of Derrida (2002) that
opened this chapter are a poignant reminder that we cannot afford to lose our philosophical
memory. Rather than be esoteric or divorced from significant educational questions of the
day, philosophy is a voice that needs to be heeded. When it comes to conceptual change that
voice reminds us that questions of belief, truth, and justification cannot be set aside when
we elect to expose the understanding of others and clearly not when we feel empowered to
alter those understandings.
Without question, there is much more to learn about the place of epistemology in con-
ceptual change theory and research. We have done little more here than open the door on
a complex and controversial realm of investigation, and we welcome the questions, con-
cerns, and insights of those who have long been invested in this program of research or
who share our passion for the building a philosophical memory that can inform and guide.
References
Alexander, P. A., Schallert, D. L., & Hare, V. C. (1991). Coming to terms: How researchers in learn-
ing and literacy talk about knowledge. Review of Educational Research, 61, 315–343.
Bonjour, L., & Sosa, E. (2003). Epistemic justification: Internalism vs. externalism, foundations vs.
virtues. Malden, MA: Blackwell Publishing.
Bradie, M., & Harms, W. (2004). Evolutionary epistemology. In: E. N. Zalta (Ed.), Stanford encyclo-
pedia of philosophy. Retrieved April 2, 2004, from http://plato.stanford.edu/archives/win2003/bradie
Chinn, C. A., & Brewer, W. F. (2001). Models of data: A theory of how people evaluate data.
Cognition and Instruction, 19(3), 323–393.
Craig, E. (Ed.). (2004). Routledge encyclopedia of philosophy. London: Routledge.
DeRose, K. (2004). Sosa, safety, sensitivity, and skeptical hypotheses. In: J. Greco (Ed.), Ernest sosa
and his critics (pp. 22–41). Malden, MA: Blackwell Publishing.
Derrida, J. (2002). Culture et dépendences—spécial Jacques Derrida, presented by F.-O. Giesbert,
with the participation of E. Levy, C. Pépin, D. Schnick, & S. Werba (France 3 Television,
May 2002). D. Egéa-Kuehne (Trans.).
Feldman, R. (2001). Naturalized epistemology. In: E. N. Zalta (Ed.), Stanford encyclopedia of phi-
losophy. Retrieved April 2, 2004, from http://plato.stanford.edu/archives/win2003/feldman
Fumerton, R. (2000). Foundationalist theories of epistemic justification. In: E. N. Zalta (Ed.),
Stanford encyclopedia of philosophy. Retrieved April 2, 2004, from http://plato.stanford.edu/
archives/win2003/fumerton
Gettier, E. (1963). Is justified true belief knowledge? Analysis, 23, 121–123.
Goldman, A. (2001). Social epistemology. In: E. N. Zalta (Ed.), Stanford encyclopedia of philoso-
phy. Retrieved April 2, 2004, from http://plato.stanford.edu/archives/win2003/goldman
Greco, J. (2002). Virtue epistemology. In: E. N. Zalta (Ed.), Stanford encyclopedia of philosophy.
Retrieved April 2, 2004, from http://plato.stanford.edu/archives/win2003/greco
Hofer, B. K. (2004). Exploring the dimensions of personal epistemology in differing classroom con-
texts: Student interpretations during the first year of college. Contemporary Educational
Kitchener, R. (2002). Folk epistemology: An introduction. New Ideas in Psychology, 20, 89–105.
Klein, P. (2001). Skepticism. In: E. N. Zalta (Ed.), Stanford encyclopedia of philosophy. Retrieved
April 2, 2004, from http://plato.stanford.edu/archives/win2003/klein.
Kvanvig, J. (2003). Coherentist theories of justification. In: E. N. Zalta (Ed.), Stanford encyclopedia
of philosophy. Retrieved April 2, 2004, from http://plato.stanford.edu/archives/win2003/kvanvig
Mason, L. (1996). An analysis of children’s construction of new knowledge through their use of
reasoning and arguing in classroom discussions. International Journal of Qualitative Studies in
Education, 9(3), 411–433.
Moser, P. K. (1995). Epistemology. In: R. Audi (Ed.), Cambridge dictionary of philosophy
(pp. 233–238). New York: Cambridge University Press.
Moser, P. K., & vander Nat, A. (2003). Human knowledge: Classical and contemporary approaches
(3rd ed.). New York: Oxford University Press.
Murphy, P. K. (2003). The philosophy in thee: Tracing philosophical influences in educational psy-
chology. Educational Psychologist, 38, 137–146.
Nussbaum, E. M., & Sinatra, G. M. (2003). Argument and conceptual engagement. Contemporary
Educational Psychology, 28, 384–395.
Pollock, J. L., & Cruz, J. (1999). Contemporary Theories of Knowledge (2nd ed.). New York:
Rowman & Littlefield.
Qian, G., & Alvermann, D. (1995). Role of epistemological beliefs and learned helplessness in
secondary school students’ learning science concepts from text. Journal of Educational
Psychology, 87(2), 282–292.
Quinton, A. (2001). The rise, fall, and rise of epistemology. In: A. O’Hear (Ed.), Philosophy at the
New Millennium (pp. 61–72). New York: Cambridge University Press.
Schommer-Aikins, M. (2002). An evolving theoretical framework for an epistemological belief
system. In: B. K. Hofer, & P. R. Pintrich (Eds), Personal epistemology: The psychology of
beliefs about knowledge and knowing (pp. 103–118). Mahwah, NJ: Erlbaum.
Sosa, E. (1994). Knowledge and justification (Vol. 1). Brookfield, VT: Dartmouth Publishing
Company Ltd.
Southerland, S. A., Sinatra, G. M., & Matthews, M. R. (2001). Belief, knowledge, and science
education. Educational Psychology Review, 13(4), 325–351.
Steup, M. (2001). The analysis of knowledge. In: E. N. Zalta (Ed.), Stanford encyclopedia of
philosophy. Retrieved April 2, 2004, from http://plato.stanford.edu/archives/win2003/steup
Vosniadou, S., & Brewer, W. F. (1994). Mental models of the day/night cycle. Cognitive Science,
18, 123–183.
Williams, M. (2001). Problems of knowledge: A critical introduction to epistemology. New York:
Oxford University Press.
Young, J. O. (2001). The coherence theory of truth. In: E. N. Zalta (Ed.), Stanford encyclopedia of
philosophy. Retrieved April 2, 2004, from http://plato.stanford.edu/archives/win2003/young
Chapter 11
Conceptions of Learning and the Experience

of Understanding: Thresholds, Contextual
Influences, and Knowledge Objects
Noel Entwistle
Introduction
This chapter takes a perspective on conceptual development and change that derives
mainly from research into teaching and learning at university level, but is also based on
ideas about the nature of concepts, epistemological development, and contextual
influences on the emergence and use of conceptions. An important distinction has to be
drawn between concepts that are used to describe the physical world and those which
describe human behaviour. The latter not only carry with them descriptions of a particular
way of thinking about a construct, but also have implications for action.
The starting point of this chapter is a set of overlapping ideas about conceptions of
knowledge and learning. In both, there is a transition in the development where a marked
qualitative change takes place. At this stage, ways of thinking about aspects of knowledge
and processes of learning become integrated to create a conception that can affect the
approaches to studying adopted by students and the level of understanding they reach in
their courses. Whether changes in behaviour actually do flow from conceptual develop-
ment, however, depends on the context within which students are operating and the par-
ticular circumstances facing them.
Crucial for students is their intention in approaching learning tasks. Only through a
deep approach, which links intention with learning processes, can the kinds of conceptual
understanding that degree courses demand be reached. This chapter explores students’
experiences in developing academic understanding to provide a counterpoint to the
research into conceptual development. The various elements within the chapter are finally
brought together to show parallels in the development of conceptions and of understand-
ing, and in the ways in which context influences the expression of conceptions in
approaches and actions.

ISBN: 0-08-045355-4
124 Noel Entwistle
Concepts and Conceptions
A necessary starting point is to look at the sense in which the terms concept and con-
ception are being used here. “Concept” is most frequently used to describe a grouping
of objects or behaviours with the same defining features that has become recognised
through research or widespread usage. The same term has also been used in the litera-
ture on conceptual change to indicate individuals’ different ways of thinking about a
particular grouping. Here, for clarity, these individual variations are referred to as “con-
ceptions”.
The early work on conceptual development, based on memory research, described
the importance of building up an extensive set of clearly defined concepts that are well
differentiated from similar ones (Ausubel, Novak, & Hanesian, 1978). And it also sug-
gested the existence of hierarchical concept “trees” in which narrower concepts are
subsumed within broader ones, rather like a zoological classification system (Baddeley,
1976). While there was good evidence for some such logical mental organisation of con-
cepts, the effect of everyday experience and individual differences in conceptualisation
make the structures much less straightforward. Conceptions evolve through increasing
knowledge and experience as they bring together additional aspects of the concept, mov-
ing closer, in most cases, to commonly agreed definitions and also becoming more
inclusive.
Physical concepts help us understand the world around us and also to use those under-
standings to our advantage in controlling and utilising its materials and properties. Social
science concepts help us to make sense of our personal and social worlds, and the con-
ceptions that people develop also have a use — to provide guidelines for action. Kelly
(1955) was one of the first psychologists to draw attention to the idiosyncratic nature of
such conceptions. He argued that people develop their own personal constructs as a way
of making sense of their own perceptions of the world and of social behaviour. In trying to
understand initially puzzling phenomena, Kelly suggested that individuals build up their
own personal constructs from their experiences. A personal construct was seen as a device
for construing or interpreting perceptions that helps not only in understanding other peo-
ple’s behaviour but also in controlling the person’s own behaviour. Kelly saw these per-
sonal constructs as being continuously modified through subsequent experiences and, as
they related to specific contexts, they were seen as explaining only a limited set of occur-
rences. These ideas paved the way for a more contextualised view of the way conceptions
develop.
Contextualisation of Conceptions
Educational research in the development of concepts in the “hard” sciences initially devel-
oped from memory research and therefore concentrated on helping students to replace
naïve concepts with scientifically accurate ones which have greater explanatory potential.
The more advanced conceptions were expected to replace those derived from everyday
experience, with the change sometimes occurring in response to the “correct” concept
being offered. But increasingly it is being recognised that conceptual change can be a slow
Conceptions of Learning and the Experience of Understanding 125
process, as in the change from novice to expert that involves both the restructuring of exist-
ing conceptions and an awareness of a broader view.
In trying to understand what is going on, it is important to make a distinc-

tion between the process of conceptual change and the end result of con-
ceptual change … With increases in age and with expertise, we not only
have a restructured system, in the sense of developing radically different
representations of reality that were not available before; … [we also
acquire] a more flexible system, a system that makes it easier to take dif-
ferent perspectives and different points of view … [Of course], it is difficult
to understand other points of view if you do not even recognise what your
own point of view is … Increased awareness of one’s own beliefs and pre-
suppositions … [or metaconceptual awareness] … is a necessary step in the
process of conceptual change. (Vosniadou, 1999, pp. 11–12)
Recent research has also been showing that conceptions are developed in relation to cer-
tain situations and demands, and for distinct purposes, and so have to be seen as contex-
tualised (Halldén, 1999; Halldén et al., 2002). Halldén (1999) has argued that conceptions
can be contextualised in a variety of ways. Some have their roots in everyday experience
— a situational context. Such conceptions, developed in everyday interactions with, say,
parents or peers, will often appear naïve or incomplete when compared with those required
for sophisticated academic understandings of phenomena. Halldén has also pointed to con-
textualisation within the “more embracing ideas” (1999, p. 64) of a cognitive context, with
educational institutions providing such contexts in terms of generally agreed concepts, but
this context also includes the conceptions built up by the individual learner. In academic
debate, linkages between concepts are generally made explicit, but in the learning experi-
ences of students, influenced by their own personal cognitive context, the connections
often remain tacit or implicit, and may not even be recognised. In developing their con-
ceptions, students experience the limitations suggested by Meno’s paradox: an adequate
conception often requires a student to create links with a broader conceptualisation of the
topic, which has to be built from the very ideas that the student has yet to understand. In
practice, the initial grasp of abstract concepts will often be incomplete or inaccurate, but it
is improved by revisiting the ideas several times and conceptualising them in increasingly
complete and complex ways.
The final form of contextualisation identified by Halldén is the cultural context or the
form of discourse in which the concept is discussed (whether everyday or academic), with
development being an inevitably gradual and recursive process through multiple and vary-
ing experiences.
Learning is not to be looked upon as a linear process in which we first learn

‘facts’ … and then try to understand these facts … Rather learning is to be
regarded as a simultaneous processing of these levels where the learner is
continuously oscillating between [them] … In the beginning … both the
understanding of the meaning of facts and the theoretical understanding are
vague … When we are trying to learn something entirely new, our point of
126 Noel Entwistle
departure can perhaps only be constituted by common speech genre.

However, when knowledge grows, the theoretical understanding as well as
the ability to interpret empirical evidence become more articulated … If
instruction is linearly organised, or if the theoretical context is not made
explicit …, the learner has to invent higher order structures … [from] a
common-sense view of the world. (Halldén, 1999, pp. 64–65)
Halldén goes on to distinguish three distinct processes within conceptual change. Early
research into the development of scientific conceptions focused on the first of these —
replacing naïve conceptions with “approved” versions through direct teaching or reading.
The second process involves providing an entirely new conception or conceptual frame-
work, but in a way that enables students to reinterpret their existing knowledge through the
lens of this new way of thinking. In this way, the new conception may have transformative
properties, leading to more inclusive, and more complex, representations. The final
process involves the independent development of a new conceptualisation from people’s
own experiences and knowledge, which enrich their “repertoire of conceptualisations”
(Halldén 1999, p. 56). These repertoires will include naïve conceptions acquired early in
life, others developed to make sense of subsequent everyday experience, and those which
are personal variants of the formally defined concepts introduced in educational contexts.
Particular events and circumstances will trigger conceptions acquired in equivalent con-
texts, as an automatic response. Subsequent reflection, or the reactions of others, may
cause an alternative conception to be brought to the fore.
Such reflection can be seen in Halldén’s research into young children’s conceptions of
the earth, which suggests that an initial flat-earth conception changes towards a more accu-
rate representation, in response to subsequent questions from the interviewer (Halldén
et al., 2002). The sudden switches in conception observed in this study indicate that several
conceptions had been previously established that could be triggered by specific questions.
If we accept that people acquire and retain repertoires of conceptions that have been devel-
oped out of particular sets of experiences within specific contexts, and that these are
brought into play by similar experiences, contexts and situations, and for specific purposes,
then we should arrive at a more complete picture of conceptual development.
In higher education, all three of Halldén’s processes of conceptual change play a part.
Students are given concepts, some of which can be understood immediately through prior
knowledge and experience, although the more abstract or counter-intuitive concepts often
cause difficulty before a satisfactory individual conception can be achieved — satisfactory,
that is, to both the learner and the teacher. Eventually, students are expected to develop
their own distinctive conceptions of the discipline and to describe and justify their under-
standings within a well-validated academic discourse. In a recent project investigating the
influence of teaching–learning environments on student learning (ETL project, 2005), we
have identified a category of concepts that appear to have a clear transformative property.
Meyer and Land (2002, 2006) have labelled these as threshold concepts, with the notion
of “opportunity cost” in economics being a particularly clear example.
A threshold concept can be considered as akin to a portal, opening up a new

and previously inaccessible way of thinking about something. It represents
a transformed way of understanding, or interpreting, or viewing something

without which the learner cannot progress. As a consequence of compre-
hending a threshold concept, there may thus be a transformed internal view
of subject matter, subject landscape, or even world view. (Meyer & Land,
2003, p. 1)
This notion of threshold concepts that open up a way of thinking is valuable in making
sense of students’ difficulties in many subject areas, but it is particularly important in con-
sidering certain concepts in the social sciences. Conceptions developed out of experience
from social situations lead to the formation of personal constructs that are used both to
make sense of other people’s behaviour and to guide one’s own. Thus, if there are thresh-
old concepts in relation to academic learning, they are likely to affect the way in which stu-
dents go about their everyday studying. We shall come to this shortly but, first, some
explanation is needed of the distinctive research approaches that have been used to iden-
tify conceptions relating to student learning.
Research Approaches to Investigating Student Learning
The earliest systematic work on students’ ways of studying used Likert-type attitude scales
to assess differences in study methods and attitudes to teaching (Brown & Holtzman,
1966). More recent research in Europe and Australasia has used a combination of inter-
views and inventories to identify concepts and categories that provide clear ways of
describing variations in the ways in which students learn and study. This had its origins in
the work of a research group in Gothenburg in the 1970s led by Ference Marton, which
introduced a distinctive qualitative research approach that was subsequently developed
into phenomenography (Marton & Booth, 1997).
The main method involved interviewing students about their ways of learning, but a
distinctive aspect of the interviewing took students, stage by stage, into deeper and deeper
reflections on the processes of learning and studying involved. The subsequent analysis of
the transcripts involved not just establishing categories describing the main similarities and
differences in students’ responses, but also systematically exploring the relationships
between those categories, and clarifying their meaning. Subsequent work on phenom-
enography identified the main conceptions held by students of a variety of key scientific
concepts, showing how incomplete conceptions contained elements of the whole, while
others represented a historical way of conceiving the idea (Marton & Booth, 1997). This
research displays a cross-section of the differing stages in conceptual development reached
by the students interviewed and draws attention to the importance of looking critically at
the meaning of the categories, a process which often identifies several types of relation-
ship between them.
One problem with the way in which phenomenography developed was the tendency to
focus on conceptions without reference to the context in which they were formed, and this
has been a particular focus of criticism (Säljö, 1997). Other work using the same general
approach has sought to retain the link with context, using a variety of approaches that
involve what Svensson (1997a) has described as contextual analysis, while a different
128 Noel Entwistle
technique — intentional analysis developed by Halldén — has also been used in research
in student learning (Scheja, 2002).
Interview studies, in their varying forms, have played a major part in establishing the
concepts described in this chapter, with their meaning and validity having been explored
further through many surveys in several countries. These surveys have used inventories to
refine the definition of the categories and to determine the statistical relationships that exist
between them (see, for example, Entwistle & Ramsden, 1983; Biggs, 1987; Vermunt,
1998; Meyer, 2002). The interplay between qualitative and quantitative methods has
proved a powerful way of developing a better understanding of student learning and how
it is influenced by teaching.
Concepts Describing Students’ Ways of Thinking About Knowledge

and Learning
Conceptions of Knowledge
In research into the conceptual development in the “hard” sciences, the “target” concepts
are always clearly defined and students’ conceptions can be seen as moving towards those
targets. The qualitative differences found in conceptions within the social sciences are less
easy to interpret, as the meaning of the concept may well be contested by theorists offer-
ing differing definitions. The notion of “knowledge” is certainly affected by such contro-
versy with post-modernists confronting positivists over relativism. Such varying
interpretations can also be seen in the variation of conceptions among university students,
but there is also evidence of progression towards more complete conceptualisation.
Perry (1970) conducted a seminal investigation into changes in the ways students at
Harvard and Ratcliffe Colleges thought about knowledge, which seems to link mainly with
the third of Halldén’s processes of conceptual development. Perry identified a series of
positions or stages through which students progressed as they gradually acquired the ways
of thinking characteristic of academic discourse. Although staff provide the “raw material”
to encourage changes in conceptions of knowledge, few courses examine the nature of sub-
ject knowledge in a way that allows students to progress epistemologically through direct
teaching. Rather, a realisation of the existence of alternative conceptions of knowledge
dawns only slowly as the ways of thinking within academic discourse are added to their
conceptual repertoires.
Perry’s interviews enabled him to distinguish dualistic thinking — expecting a single
“right” answer — from relativistic reasoning in which students are able to accept uncer-
tainty in knowledge and use evidence and logic to test alternative explanations. Perry
traced the initially hesitant move from dualistic to relativistic thinking which, once estab-
lished, could lead on to a firm, evidentially based commitment to a particular interpreta-
tion or perspective. The commitment among the students Perry interviewed was associated
with a tolerant view of competing interpretations, and also moral feelings about taking up
justifiable stances.
The top half of Figure 11.1 summarises the steps in this epistemological progress and
draws attention to the transformative characteristics of the emergence of relativistic think-
ing. For Perry (1988), this stage represented a pivotal position that allows students to see
Conceptions of knowledge
Dualism Relativism
Knowledge Multiple Evidence Commitment

as absolute, perspectives - Awareness of
knowledge used to to a personal,
provided by opinions of reason among reasoned
authorities equal value as provisional
alternatives perspective
Pivotal
position
Changing
Recognising differing as a person -
Expanding awareness through a broader, integrative conception forms of knowledge sense of identity
and learning processes
Threshold
Acquiring Memorising Applying Understanding Seeing things
factual what has and using what has been in a
information to be learned knowledge learned different way
Reproducing Seeking meaning
Conceptions of learning
Figure 11.1: Categories used in describing conceptions of knowledge and learning.
knowledge in a crucially different way, and so becomes a threshold which opens up the
subject in important ways.
(Relativism) has taken us over a watershed, a critical traverse in our

Pilgrim’s Progress … In crossing the ridge of the divide, … (students) see
before (them) a perspective in which the relation of learner to knowledge is
radically transformed. In this new context, ‘Authority,’ formerly a source
and dispenser of all knowing, (becomes) … a resource, a mentor, a model
… (Students) are no longer receptacles but the primary agents responsible
for their own learning … As students speak from this new perspective, they
speak more reflectively. And yet the underlying theme continues: the
learner’s evolution of what it means to know. (Perry, 1988, p. 156)
The recognition of both knowledge and values as provisional can bring with it a feeling of
insecurity, which may inhibit further development for a while, or lead to retreat into the “safer”
territory provided by the apparent certainty of dualism. If students manage to move on, they
develop not just an understanding of the academic discourse and ways of reasoning used in
the discipline they are studying, but also the greater tolerance of other interpretations of evi-
dence that Vosniadou (1999) sees as a more general characteristic of conceptual growth.
Subsequent research has largely confirmed the developmental trend identified by Perry
(Hofer & Pintrich, 2002), but with some important qualifications suggesting gender dif-
ferences (Clinchy, 2002), as well as suggestions that the trend is not unidimensional.
Schommer-Aikins (2002) has found several factors within her measures of epistemologi-
cal development that follow rather different trajectories, although some of these seem to
overlap with the descriptions of students’ conceptions of learning (Pintrich, 2002).
130 Noel Entwistle
There are certain aspects of Perry’s scheme that deserve further consideration. First, the
recognition that students can regress in their epistemology suggests that the less developed
conceptions of knowledge are not entirely replaced by relativistic ways of thinking, and so
we have further evidence of co-existence of conceptions and the alternation between con-
ceptions due to contextual variations. Second, if we examine the nature of the categories
there is some confusion. Perry found good evidence of progression towards more sophis-
ticated views of knowledge, but his descriptions of relativism bring together two rather dif-
ferent aspects — philosophical views on the nature of knowledge and students’ gradual
recognition of how evidence is used in their discipline to justify conclusions or theories.
The extent to which students will perceive relativism in its philosophical sense will depend
markedly on the subject area: in the sciences, competing theories can be tested against
some generally agreed measurements, while in the humanities the contested nature of
knowledge and its relativism becomes self-evident, and so is more likely to be recognised
by the students. It seems likely, therefore, that students in different subject areas will be
seeing relativism in importantly different ways.
Conceptions of Learning
Säljö (1979) investigated different conceptions of learning by interviewing adults with dif-
fering levels of educational experience. He found a series of categories that differentiated
broadly between learning as “reproducing knowledge” and learning as “seeking meaning”,
a distinction which parallels the changes in students’ ways of thinking about knowledge.
The earlier categories describe learning simply in terms of acquiring factual information,
with learning being equated with memorising what has to be learned, and reproducing it
as required. A further step recognises the way learning depends on using knowledge in
future situations, before reaching a crucial transformation in ways of thinking about learn-
ing. The emphasis shifts from seeing learning as the accumulation and use of unrelated
pieces of information at the behest of others, to viewing it as a process in which under-
standing is developed and knowledge takes on a personal meaning. Here, a threshold is
again reached and the landscape of learning changes, as relativism in relation to learning
processes is appreciated. The final step in Säljö’s progression describes learning as seeing
things in a different way (as with the threshold concepts mentioned previously) and
implies the use of a more sophisticated conceptualisation of the material, with existing rep-
resentations being reorganised into forms of understanding that fit reality more closely and
are more satisfying to the learner.
This set of categories, shown in the bottom half of Figure 11.1, describes qualitative dif-
ferences that were empirically related to the length and type of education received, and so
appear to be developmental. In a subsequent study, Säljö (1982) found that adults who
think about learning in a more sophisticated way recognise a variety of different learning
processes, and understand that effective learning depends on finding the most appropriate
ways of tackling a particular task for a specific purpose, in a given context. This awareness
can also lead to a more active engagement with learning as students monitor their progress
towards understanding. Until students have grasped the importance and value of develop-
ing their own individual conceptions and understandings of academic material, they are
unlikely to appreciate the implications of contrasting learning processes. The recognition
of learning as developing personal meaning, rather than just acquiring other people’s
knowledge, thus creates what can be seen as an emergent property of the more advanced
conceptions — sometimes referred to as metalearning — which parallels the notion of
“metaconceptual awareness”, mentioned in the quote from Vosniadou, earlier.
Here we have not just a potential repertoire of conceptions, but also an advanced con-
ception that includes a recognition of how best to marshal the different learning processes
to achieve academic goals. We also have two processes — rote learning and meaningful
learning (Ausubel et al., 1978), which are different in kind, thus creating, as with Perry’s
categories, a progression which is not straightforward, being influenced by experience and
circumstances. Students initially rely on rote learning in a routine way, and only gradually
recognise that other kinds of learning are more appropriate for the tasks they meet in
higher education. Although staff will generally expect students to be already well aware of
the kinds of learning they should be using, that is often not the case. If we take the paral-
lels with Perry’s students, the metacognitive awareness of multiple ways of learning can
be expected to emerge, along with the recognition of the academic discourse and use of
evidence appropriate to their specific discipline, only in the latter stages of a degree course.
Other research on conceptions of learning (Marton, Dall’Alba, & Beaty, 1993) sug-
gested a sixth category — changing as a person — which indicates a distinct form of learn-
ing that goes beyond the cognitive aspects reported by Säljö. It links the development of
conceptions of learning to a changing sense of identity and so introduces the broader sense
of learning also seen in Perry’s final category.
Approaches to Learning and Studying
As we have seen, the more sophisticated conceptions of both knowledge and learning are
transformative, leading to greater metacognitive awareness. It would be strange, then, if
such changes in ways of thinking about knowledge and learning did not have marked
effects on how students study, or on the quality of learning achieved. Later on we shall see
that this does happen, although not in a straightforward way. First, however, we need to
explore how students go about their everyday study activities. In the literature, there are
differing sets of terms and constructs describing these differences from various theoretical
perspectives, but with a good deal in common (Entwistle & McCune, 2004). For simplic-
ity, we will use just one set here, one that is based on the work of Marton and his col-
leagues.
The construct of approach to learning was derived from interviews with students who
had been asked to read a short academic article that contained evidence used to build up
an argument and reach a conclusion (Marton & Säljö, 1976). The students were told that
they would be asked questions on the article afterwards. First, they were asked to sum-
marise the general meaning of the article and then to explain how they had gone about their
reading. Analysis of the first question indicated substantial differences in the level of
understanding of the author’s meaning. The second question was analysed to try to provide
reasons for these different levels of outcome and the researchers found a major distinction
in the focus of attention students reported. Some students had adopted what came to be
called a surface approach: they concentrated on the words and information, trying to spot
132 Noel Entwistle
questions likely to be asked and then memorising the answers. Their focus was thus on the
information itself, rather than on its meaning or the relationship between evidence and
conclusions. Other students had been trying to understand the meaning of the article for
themselves, which involved paying careful attention to how well the evidence supported
the conclusions — a deep approach. This contrast can be seen in comments of two stu-
dents in later interviews at Lancaster.
In reading the article, I was looking out mainly for facts and examples. I
read the article more carefully than I usually would, taking notes, knowing
that I was to answer questions about it. I thought the questions would be
about the facts in the article … This did influence the way I read; I tried to
memorise names and figures quoted, etc. I tried to concentrate — too hard
— therefore my attention seemed to be on ‘concentration’ rather than on
reading, thinking, interpreting and remembering — something I find hap-
pening all the time I’m reading text-books.
I read more slowly than usual, knowing I’d have to answer questions, but I
didn’t speculate on what sort of questions they’d be. I was looking for the
argument and whatever points were used to illustrate it. I could not avoid
relating the article to other things I’d read, past experience, and associa-
tions, etc. My feelings about the issues raised made me hope [the author]
would present a more convincing argument than he did, so that I could for-
mulate and adapt my ideas more closely, according to the reaction I felt to
his argument. (Entwistle, 1988, pp. 77–78)
Svensson (1977) analysed the Gothenburg interviews independently and saw an equiva-
lent distinction, but with a different emphasis. He described the difference as atomistic and
holistic, contrasting students who were merely noting and ordering parts of the text with
those who were “integrating parts by use of some organising principle” (Svensson, 1997b,
p. 65). The atomistic focus brings with it the use of rote learning, while a deep, holistic
approach depends on making connections between ideas. Developing understanding often
involves an interplay between inductive and deductive thinking, considering the patterns of
relationships in relation to the whole topic and examining the detailed evidence and logic.
Here we see parallels with relativistic reasoning and the more integrative conceptions of
learning. The metacognitive awareness of the uses of contrasting learning processes is also
a feature of deep approaches; students intending to understand may choose to supplement
inductive and deductive thinking with rote-learning, where that is a prerequisite for under-
standing. The deep approach thus can include the memorising of detail or terminology,
although its focus will still be on the development of personal understanding.
Subsequent studies developed the contrast between deep and surface approaches by
asking students about their experiences of everyday studying (Entwistle & Ramsden,
1983). These investigations confirmed the validity and importance of the distinction across
a wide range of educational contexts and cultures (Richardson, 2000), but there has been
less agreement about the extent to which approaches to learning are consistent across con-
texts. The evidence on the contextualisation of conceptions, and Marton’s own ideas about
the nature of approaches to learning, indicate that approaches must be variable. As every-
one has the basic mental processes used in deep and surface approaches, there has to be an
element of choice in deciding how learning tasks are tackled. Previous success in using one
or other approach will, however, lead to its subsequent use, and can build towards its more
habitual use. To the extent that the learning required in different courses, or by different
academic staff, is similar, a relatively consistent approach can be expected, but where the
contexts are perceived to be different, then the approaches will vary accordingly
(Ramsden, 1997). We thus have to accept elements of both consistency and variability in
approaches to learning (Thomas & Bain, 1984).
The element of consistency means that students can be asked about their typical
approaches using inventories, with the deep/surface dichotomy generally emerging from
the factor analysis of students’ responses (Biggs, 1993; Entwistle & McCune, 2004;
Richardson, 2000). Questionnaire surveys have repeatedly shown overall relationships
between approaches and students’ experiences of instructional methods, which imply an
important element of stability (ETL project, 2005), but interview studies have drawn atten-
tion to individual reactions to the same experiences of teaching, and also the effect of this
specific context in determining whether an established approach is changed.
In a study by McCune (2000), for example, a student described how a tutor’s encour-
agement of a critical approach in reading had affected her approach.
That’s really made me think a lot more about what I’m actually reading …
I think that’s helped me with my psychology as well, because I have to keep
remembering … that they’re only ideas and psychologists’ views on some-
thing, and it’s not actually dead set … It makes me less trusting of what I’m
reading, but in a way it makes me feel more independent in my work, …
less like I’m being taken in by what they’re writing, if I actually think,
‘Why are they saying (that)?’ (Entwistle, McCune, & Walker, 2001, p. 121)
From other comments in this study, it appears that students have to be both alert and
“ready” if they are to take advantage of a tutor’s encouragement to use a deep approach
(McCune, 2000, 2004). There also has to be sufficient incentive to change established
habits. Having a conception of learning that embraces personal understanding is usually
associated with an intention to adopt a deep approach, but the necessary learning processes
may not always flow from that intention. There must be adequate prior knowledge, but the
approach adopted also depends on the reasons for taking the course and feelings about the
tutor. Context, circumstance and feelings all seem to affect whether a conception is fully
expressed in action, or whether the intention to use a specific approach actually leads to
the actions necessary to carry it through successfully (Calder, 1989).
A similar conclusion can also be drawn from research into the conceptions of teaching
held by academic staff. Over the last decade, a series of interview studies have described
differences in the conceptions that academic staff have of teaching, with a main dichotomy
emerging that contrasts teaching seen as the process of transmitting knowledge with teach-
ing envisaged as the encouragement of conceptual development (Kember, 1998). At one
pole, the focus is on the knowledge that the student is expected to learn, viewed entirely
from the perspective of the member of staff. The opposite conception sees teaching as the
134 Noel Entwistle
process of encouraging the student to acquire a new way of looking at the world and
includes descriptions of how best to explain the ideas to students in ways that open out the
subject for them. This contrast can be illustrated in extracts from two lecturers interviewed
as part of a study by Prosser, Trigwell, & Taylor (1994).
I put great emphasis on behavioural objectives and making sure that I cover
the syllabus thoroughly. I also think it is important that the students take
away a good set of notes: it gives them confidence if they have those notes
to study from and also gives them a clear idea of what they have to know in
order to pass the exam. And I’ll help them by essentially writing the notes
for them … In my teaching, I also try to make the students realise the impor-
tance of being accurate, for example in the way they write up lab reports …
(and) I also give a lot of (tests) to make sure they know their stuff.
Lectures provide a presence that a book doesn’t, and you can utilise that in
the lecture by a directly engaging question. The lecturer can actually
engage (the student) in that question, in a much more interactive mode …
I suppose I’m saying that the function of the lecture is to bring inquiry to
life … [In my teaching], I’m constantly challenging the students … to think
something through for themselves. I think that generates a certain sense of
intrigue and understanding is related to that. (And) in my preparation I
actually have to create this (situation) every time, rather than just remem-
ber (the content). (The lecture) is a conversation in which there’s active lis-
tening involved. (Prosser, personal communication)
Van Driel, Verloop, Van Werven, and Dekkers (1997) also found these two extremes in
the views of the lecturers they interviewed, with staff who emphasised the transmission of
knowledge also blaming students for poor performance and low levels of motivation. But
they also described an intermediate category in which staff saw the importance of making
the students actively involved in their own learning, while the staff retained responsibility
for introducing and controlling those opportunities. There is some indication that an
advanced conception integrates different forms of knowledge about the subject, teaching,
and student learning (Entwistle & Walker, 2002) and also brings together the earlier
approaches and actions into a flexible whole which can be varied according to circum-
stances, as explained in the following reflective comment from a physics lecturer.
In the early days, it was a matter of being prepared, presenting content con-
fidently and accurately, and being in control. Some days it worked well and
others it was a struggle, and I never knew quite why — it must have been
the students or perhaps the weather that day. Later, as I developed more
mastery, I could teach in a more conversational style, maintaining a sense
of theatre, creating and taking opportunities to engage students’ interest and
thinking. Examples, demonstrations and questions can be chosen to max-
imise such engagement and wherever possible to elucidate and challenge
students’ preconceptions. The experience of teaching now, from my point
of view, is more akin to a masterful jazz musician improvising and inter-

acting with partners, allowing the instrument itself to speak, to express and
inspire — rather than having to clumsily pluck or blow to force a pre-
dictable outcome. (Entwistle & Walker, 2002, pp. 29–30)
In interviews about teaching, staff often move interchangeably in their comments from
expressions of general conceptions to descriptions of specific approaches and actions
(Samuelowicz & Bain, 2001). Unless the questions are phrased in ways that focus clearly
on one aspect, the intermixture of conceptions, approaches, and actions creates confusion
in the analysis. Beyond the immediate confusion, however, a pattern emerges. Staff who
have reached a sophisticated conception of teaching are more likely to show metacogni-
tive awareness of the links between teaching and learning, and to develop flexible ways of
teaching that are adopted to the needs of both the subject and the students. But an advanced
conception does not also lead to the equivalent actions. There is accumulating evidence
that lecturers are more likely to make explicit use of their conceptions when planning
courses and thinking broadly about teaching (McAlpine, Weston, Berthiaume, & Fairbank-
Roch , 2006) than when preparing their day-to-day teaching (Eley, 2006).
It thus appears that conceptions of either learning or teaching become conscious and
explicitly used only in certain circumstances and for particular purposes. In a recent study
across five disciplines, we have found many instances in which staff have been prevented
from teaching in what they would see as better ways by lack of time or resources, or by
the competing pressures from research and administration (ETL, 2005). An undeveloped
conception strongly limits the approaches and actions that follow from it, but a more
sophisticated conception will not always lead to equivalent approaches and actions.
It depends on context and circumstances.
As we have seen, correlational studies of conceptions and approaches emphasise con-
sistent relationships between conceptions and approaches, but interview studies bring out
individual variations in reactions, indicating how relative stability in approaches can dis-
guise important elements of variability and change. Because the evidence contains both
these elements, either one can be highlighted, depending on the expectations and theoret-
ical orientation of the researcher, but both remain part of the phenomenon being investi-
gated (Entwistle, McCune & Walker 2001). Bringing together all these findings with those
from other studies that have linked conceptions of learning with study strategies (Lonka &
Lindblom-Ylanne, 1996; Vermetten, 1999), a chain of influence in terms of student char-
acteristics can be traced from the more stable individual differences and prior knowledge,
through conceptions of learning, to approaches to studying and levels of understanding
(Entwistle, 2007).
We come now to variations in the development of academic understanding, which seem
to parallel the processes involved in conceptual growth, but on a broader canvas.
The Nature of Academic Understanding
Although understanding is generally cited as being one of the main aims of higher educa-
tion, with conceptual understanding being specifically mentioned, rather little research has
136 Noel Entwistle
been conducted into its nature or the way in which it is experienced by students. In psy-
chology, understanding is often described in relation to studies of problem solving
(Nickerson, 1985), but that is a special case. Academic understanding itself has only
recently become a major focus in educational research, and then mainly at school level
(Wiske, 1998; Newton, 2000; Entwistle & Smith, 2002).
Much of the research on student learning has taken the meaning of understanding
rather for granted, but there has been a series of small-scale studies investigating how
understanding is experienced by final-year students as they prepare for final examina-
tions (Entwistle & Entwistle, 1992, 2003). In the interviews, students were taken
deeply into their experiences through successive invitations either to describe more
fully or to explain their approaches. They were also asked specifically how they had
experienced understanding. At its heart was a feeling of satisfaction, but its expression
ranged from the sudden “aha” (as confusion on a particular topic was replaced by
insight) to a less dramatic feeling coming from the ability to follow a lecture or from
an emerging appreciation of the nature of the discipline itself. This feeling involved a
recognition of the meaning and significance of the material learned and, on occasions,
was explicitly linked to previous personal or professional experience. The feeling of
understanding also included a sense of coherence and connectedness — “things click-
ing into place” or “locking into a pattern” — conveying an implication of completeness
and closure. Nevertheless, students recognised that their understanding might well
develop further, that understanding involved no more than provisional wholeness. Their
current understanding was felt to be adequate for the present, but might well be modi-
fied and extended in the future. This almost paradoxical combination of completeness
and potential for further development does seem to be central to the concept of under-
standing.
Interview comments suggested that, even close to the completion of their degree, some
of the students had not actively sought conceptual understanding, while several were so
concerned about “passing” that they talked mainly about how they would meet the per-
ceived exam requirements. These marked differences suggested that there was a hierarchy
of forms of understanding (Table 11.1), which could be seen from students’ comments on
their revision strategies. The initial analysis produced five categories of description that
Table 11.1: Contrasting forms of understanding in revising for final examinations.
Students’ understandings were found to differ in three main ways
1. Breadth of understanding
2. Depth or level of understanding
3. Structure used to organise the material being learned
a. imposing little or no structure on the facts learned
b. relying exclusively on the lecturer’s structures
c. producing prepared answers to previous years’ questions
d. adapting own understanding to expected question types
e. using own well-justified conception of the topic and subject area
indicated the different ways in which students had prepared for the exam, but looking at
the relationships between the categories suggested underlying differences that were more
salient. The form of understanding depended on its breadth — the amount of material that
the students sought to integrate within their understanding. It also reflected its depth — the
amount of effort put into deepening understanding, leading to different levels of under-
standing. Perhaps most importantly, the form of understanding reflected the extent to
which there was a coherent structure shaping the understanding, and whether this structure
had been provided by the teacher or created through the student’s own active consideration
of the topic.
As the categories shown in Table 11.1 were derived from students’ descriptions of their
understanding, and processes of reaching and using it, these seem to represent conceptions
of the type of understanding students believed they would have to demonstrate. The appar-
ent progression is towards greater self-reliance in constructing a personal understanding,
rather than relying on the content and structure provided, and it also represents a concep-
tion broadening from a focus on individual topics to be revised prior to the exam, towards
a conception of the subject as a whole and its underlying discourse. There are echoes here
of both Halldén’s third process within conceptual development, showing independent
development of a new conceptualisation, and Perry’s relativism in the sense of recognis-
ing and using the academic discourse.
This development towards a coherent, idiosyncratic structure developed by students to
encapsulate their understanding led to a further exploration of the data, focusing on what
had been said about constructing the personal structures. It soon became clear that students
who had put substantial effort into developing their own structure as part of the revision
process, often experienced their understandings almost as entities that could be brought to
mind, “viewed” in some sense, reconsidered and reshaped to fit new requirements, without
fundamentally changing what had been understood. These entities were described as
knowledge objects (Entwistle & Marton, 1994; Entwistle, 1998). (It should be noted that
the term has been used in a rather different sense by some other researchers.) The main
characteristics of knowledge objects came from their being seen “virtually as a picture”,
and with components within them which could be used to pull in related information, as
required. The details were not in the picture but could be accessed from it, while the struc-
ture offered the students a logical path through which to develop an argument in the essays
they were required to write in demonstrating their understanding. Given the abstract nature
of the idea of a “knowledge object”, more extensive quotes have been included here to
clarify its meaning. The comments relate mainly to the use of summary notes in preparing
for the exams and how the understandings were used during them.
I can see that (part of my summary notes) virtually as a picture, and I can
review it and bring in more facts about each part … Looking at a particular
part of the diagram sort of triggers off other thoughts. I find schematics, in
flow-diagrams and the like, very useful because a schematic acts a bit like
a syllabus; it tells you what you should know, without actually telling you
what it is … The basic diagram is on paper, but details about the diagram
are added on later by myself in my head … The facts are stored separately,
and the schematic is like an index, I suppose ….
138 Noel Entwistle
(As I wrote), it was almost as though I could see it all fitting into an over-
all picture. I think you’re almost developing what you know, and are play-
ing it in a slightly different way (to fit the question) …. Following that logic
through, it pulls in pictures and facts as it needs them … Each time I
describe (a particular topic), it’s likely to be different … Well, you start with
evolution, say, … and suddenly you know where you’re going next. Then,
you might have a choice … to go in that direction or that direction … and
follow it through various options it’s offering … Hopefully, you’ll make the
right choice, and so this goes to this, goes to this — and you’ve explained
it to the level you’ve got to. Then, it says ‘Okay, you can go on to talk about
further criticisms in the time you’ve got left’. (Entwistle & Entwistle, 2003,
pp. 32, 34)
Students’ structured summary notes were often recalled as images in the exams, with the
summary notes often structured to form a visual mnemonic, but the feeling of understand-
ing did not always relate directly to imagery. Some students only visualised the structure of
their notes during the examinations when they were in difficulty. For them, understanding
was tapped directly, with visualisation being “available”, but not necessarily used.
(In exams), I just clear my mind and something comes … You know it’s
visual in some ways, but it’s also just there without necessarily being visual
… (It’s not as if) you remember a page, and the page is locked in your
memory. What I’m saying is that the ideas are locked in your memory and
they display as a page when you’re thinking about it, but not necessarily
when you’re putting it down … You can sort of by-pass the conscious per-
ception of your memory: it may not be a visual memory, but (sometimes) it
may have to be perceived as a visual memory … I think, in a stress situa-
tion like an examination, you don’t actually (have to) reach for it, it comes
out automatically. That may show that it’s not actually a visual memory, as
such, but a visual expression of ‘central memory’. (Entwistle & Entwistle,
2003, p. 38)
Although knowledge objects have been described here in only one context — prepara-
tion for examinations — they have also been detected in students’ descriptions of writing
coursework essays (term papers), but only when the students engaged actively and per-
sonally with the topic (Entwistle, 1995). Evidence of the potential longevity of knowledge
objects has also been found — in one instance, over a period of several years, with visual
imagery playing an important part in the retained structure. There is also anecdotal evi-
dence to suggest that staff regularly make use of knowledge objects in preparing papers
(or a chapter like this). Staff also have an awareness of knowledge objects as they are
giving lectures, although evidence for this remains impressionistic (Entwistle, 1998).
There is apparent similarity between a “knowledge object” and what has previously
been called a “schema” (see Westen, 1999), but the knowledge object involves a sense of
personal attachment to the way the understanding has been constructed, and also describes
a particular way of bringing together ideas and evidence to justify an understanding within
an academic discourse. It is thus a form of contextualised schema that can also be seen in
terms of neurological descriptions of mental processes.
Mind is a stream of conscious and subconscious experience … It is at root

the coded representation of sensory impressions and the memory and imag-
ination of sensory impressions … Long-term memory recalls specific
events… It also re-creates not just moving images and sound but meaning
in the form of linked concepts simultaneously experienced … By spreading
activation, the conscious mind summons information from the store of
long-term memory … and holds it for a brief period in short-term memory.
During this time it processes the information, … while scenarios arising
from the information compete for dominance … As the scenarios of con-
sciousness fly by, driven by stimuli and drawing upon memories of prior
scenarios, they are weighted and modified by emotion … which animates
and focuses mental activity … What we call meaning is the linkage among
neural networks created by spreading activation that enlarges imagery and
engages emotion. (Wilson, 1998, pp. 119, 121, 122, 123, 126)
Conclusion
This series of examples of conceptual development and experiences of understanding has

helped to clarify the nature of advanced conceptions of psychological constructs, and how
conceptions relate to approaches and actions. There is evidence of stability in the use of a
particular conception, approach, or understanding, but also indications that people have
repertoires that are built up within different contexts, triggered by specific contexts, and
used for differing purposes. It is not at all clear just how many conceptions are likely to
co-exist. Phenomenographic analyses have generally produced between three and seven
distinguishable categories, although with no suggestion that these can co-exist within any
individual. The tension between the idea of stable conceptions stored in long-term mem-
ory, and the evidence for contextualisation of conceptions, leads to intriguing questions
about the relationships between differing conceptions within the memory, and how each is
triggered.
The descriptions of students using knowledge objects provides a window into the
processes of concept formation, as the development of understanding becomes a more con-
scious process that can be reported. In the interviews at university level, conceptions and
understandings were described as if they had become firmly established in memory, ready to
be brought into play confidently and quickly in exams. But the knowledge objects, although
well constructed, were not in a fixed form. They were perceived to be flexible and students
talked about adapting them to differing demands and purposes. It appears that they provided
a generic logical path to explain a topic within an exam, but one that could be modified to
take account of the demands of the question and the anticipated audience of examiners.
We suggested earlier that some social science concepts guide actions, and from the
evidence presented it seems that the best-developed conceptions create emergent proper-
ties, seen for example in threshold concepts that open up the subject, and in the way
140 Noel Entwistle
metacognitive awareness of studying arises from an advanced conception of learning. But

the path from advanced conceptions of knowledge and learning, to intentions to under-
stand, and on to the actions that fulfil that intention within approaches to studying, is
strongly affected by the context. It is influenced by the situation the student experiences
in the classroom, the relationship with the teacher, the perception of the task, and the
reward system in terms of grade criteria (Entwistle, 2000, 2007). Research into the con-
ceptions of teaching held by staff also points to an advanced conception that develops
over time into a metacognitive view of the links between subject knowledge, pedagogical
knowledge and the needs of students. Whether or not the conception is expressed in
equivalent actions, however, again depends on circumstances and context.
It appears that an advanced conception of conceptual development must now be seen in
a more complex way than was envisaged in the earlier studies based on cognitive psy-
chology, while the ideas from social constructivism on their own also fail to capture the
whole process. We need a theory that brings together cognitive constructivism with current
ideas about the influence of context and evidence for the co-existence of rather different
conceptions related to varying situational contexts. From a review of the current literature,
such a theory is still no more than a gleam in the eye of the author.
Acknowledgements
I am grateful to Ola Halldén for lengthy discussions about the nature of conceptions on a
number of occasions, but in particular on the journeys between Athens and Delphi. I also
found comments made by Patricia Alexander at the symposium very helpful in improving
Figure 11.1. Above all, I am grateful to colleagues who have collaborated in the research
on student learning over the years and, in particular, Velda McCune, Hilary Tait, and Paul
Walker who were involved in the work on conceptions and approaches, and Dorothy and
Abigail Entwistle who were involved with developing the ideas on knowledge objects.
References
Ausubel, D. P., Novak, J. S., & Hanesian, H. (1978). Educational psychology: A cognitive view. New
York: Holt, Rinehart & Winston.
Baddeley, A. D. (1976). The psychology of memory. New York: Harper & Row.
Biggs, J. B. (1987). Student approaches to learning and studying. Melbourne: Australian Council for
Educational Research.
Biggs, J. B. (1993). What do inventories of students’ learning processes really measure? A theoreti-
cal review and clarification. British Journal of Educational Psychology, 63, 3–19.
Brown, W. F., & Holtzman, W. H. (1966). Manual of the ‘survey of study habits and attitudes’. New
York: Psychological Corporation.
Calder, I. (1989). The study and learning strategies of students in a New Zealand tertiary institution.
Unpublished Ph.D. Thesis. University of Waikato, New Zealand.
Clinchy, B. McV. (2002). Revisiting women’s ways of knowing. In: B. K. Hofer, & P. R. Pintrich
(Eds), Personal epistemologies: The psychology of beliefs about knowledge and knowing
(pp. 63–88). Mahwah, NJ: Lawrence Erlbaum.
Eley, M. (2006). Teachers’ conceptions of teaching, and the making of specific decisions in planning
to teach. Higher Education, 51, 191–214.
Entwistle, A. C., & Entwistle, N. J. (1992). Experiences of understanding in revising for degree
examinations. Learning and Instruction, 2, 1–22.
Entwistle, N. J. (1988). Styles of learning and teaching. London: David Fulton.
Entwistle, N. J. (1995). Frameworks for understanding as experienced in essay writing and in prepar-
ing for examinations. Educational Psychologist, 30, 47–54.
Entwistle, N. J. (1998). Approaches to learning and forms of understanding. In: B. Dart, &
G. Boulton-Lewis (Eds), Teaching and learning in higher education (pp. 72–101). Melbourne:
Australian Council for Educational Research.
Entwistle, N. J. (2000). Approaches to studying and levels of understanding: The influences of teach-
ing and assessment. In: J. C. Smart (Ed.), Higher education: Handbook of theory and research
(Vol. 15, pp. 156–218). New York: Agathon Press.
Entwistle, N. J. (2007). University teaching and student understanding: Research guiding good
practice. London: Palgrave-Macmillan (in preparation).
Entwistle, N. J., & Entwistle, D. M. (2003). Preparing for examinations: The interplay of memoris-
ing and understanding, and the development of knowledge objects. Higher Education Research
and Development, 22, 19–42.
Entwistle, N. J., Marton, F. (1994). Knowledge objects: understandings constituted through intensive
academic study. British Journal of Educational Psychology, 64, 161–178.
Entwistle, N. J., McCune, V., & Walker, P. (2001). Conceptions, styles and approaches within higher
education: Analytic abstractions and everyday experience. In: R. J. Sternberg, & L.-F. Zhang
(Eds), Perspectives on thinking, learning and cognitive styles (pp. 103–136). Mahwah,
NJ: Lawrence Erlbaum.
Entwistle, N. J., & McCune, V. (2004). The conceptual bases of inventories measuring study strate-
gies in higher education. Educational Psychology Review, 16, 325–346.
Entwistle, N. J., & Ramsden, P. (1983). Understanding student learning. London: Croom Helm.
Entwistle, N. J., & Smith, C. A. (2002). Personal understanding and target understanding: Mapping
influences on the outcomes of learning. British Journal of Educational Psychology, 72, 321–342.
Entwistle, N. J., & Walker, P. (2002). Strategic alertness and expanded awareness in sophisticated
conceptions of teaching. In: N. Hativa, & P. Goodyear (Eds), Teacher thinking, beliefs and knowl-
edge in higher education (pp. 15–40). Dordrecht, The Netherlands: Kluwer.
ETL Project. (2005). Enhancing teaching-learning environments in undergraduate courses. Various
working papers from this ESRC-funded project within the Teaching and Learning Development
Programme will be found at http://www.tla.ed.ac.uk/etl/publications.html or http://www.tlrp.org.uk
Halldén, O. (1999). Conceptual change and contextualisation. In: W. Schnotz, S. Vosniadou, &
M. Carretero (Eds), New perspectives on conceptual change (pp. 53–66). Oxford: Pergamon.
Halldén, O., Petersson, G., Scheja, M., Ehrlén, K., Haglund, L., Österlind, K., & Stenlund, A. (2002).
Situating the question of conceptual change. In: M. Limon, & L. Mason (Eds), Reconsidering con-
ceptual change: Issues in theory and practice (pp. 137–148). Dordrecht, The Netherlands: Kluwer.
Hofer, B. K., & Pintrich, P. R. (Eds) (2002). Personal epistemologies: The psychology of beliefs
about knowledge and knowing. Mahwah, NJ: Lawrence Erlbaum.
Kelly, G. A. (1955). The psychology of personal constructs. New York: Norton.
Kember, D. (1998). Teaching beliefs and their impact on students’ approach to learning. In: B. Dart,
& G. Boulton-Lewis (Eds), Teaching and learning in higher education: From theory to practice
(pp. 1–25). Melbourne: Australian Council for Educational Research.
Lonka, K., & Lindblom-Ylanne, S. (1996). Epistemologies, conceptions of learning, and study prac-
tices in medicine and psychology. Higher Education, 31, 5–24.
Marton, F., & Booth, S. (1997). Learning and awareness. Mahwah, NJ: Lawrence Erlbaum.
142 Noel Entwistle
Marton, F., & Säljö, R. (1976). On qualitative differences in learning – I: Outcome and process.
British Journal of Educational Psychology, 46, 4–11.
Marton, F., Dall’Alba, G., & Beaty, E. (1993). Conceptions of learning. International Journal of
Educational Research, 19, 277–300.
McAlpine, L., Weston, C., Berthiaume, D., & Fairbank-Roch, G. (2006). How do professors explain
their thinking when planning and teaching? Higher Education, 51, 125–155.
McCune, V. (2000). The development of first-year students’ approaches to studying. Unpublished
Ph.D. thesis. University of Edinburgh, Edinburgh.
McCune, V. (2004). Development of first-year students’ conceptions of essay writing. Higher
Education, 47, 257–282.
Meyer, J. H. F. (2002). An overview of the development and application of the Reflections on
Learning Inventory (RoLI). Durham, NC: University of Durham, Centre for Learning, Teaching
and Research in Higher Education.
Meyer, J. H. F., & Land, R. (2002). Threshold concepts and troublesome knowledge: Linkages to
ways of thinking and practising within the disciplines. In: C. Rust (Ed.), Improving student learn-
ing: Improving student learning theory and practice – 10 years on (pp. 412–424). Oxford: Oxford-
Brookes University, Centre for Staff and Learning Development.
Meyer, J. H. F., & Land, R. (2006). (Eds.) Overcoming barriers to student understanding: Threshold
concepts and troublesome knowledge. London: Routledge.
Newton, D. P. (2000). Teaching for understanding. London: Routledge Falmer.
Nickerson, R. S. (1985). Understanding understanding. American Journal of Education, 93, 201–239.
Perry, W. G. (1970). Forms of intellectual and ethical development in the college years: A scheme.
New York: Holt, Rinehart and Winston.
Perry, W. G. (1988). Different worlds in the same classroom. In: P. Ramsden (Ed.), Improving learn-
ing: New perspectives (pp. 145–161). London: Kogan Page.
Pintrich, P. R. (2002). Future challenges and directions for theory and research on personal episte-
mology. In: B. K. Hofer, & P. R. Pintrich (Eds), Personal epistemologies: The psychology of
beliefs about knowledge and knowing (pp. 389–414). Mahwah, NJ: Lawrence Erlbaum.
Prosser, M., Trigwell, K., & Taylor, P. (1994). A phenomenographic study of academics’ conceptions
of science learning and teaching. Learning and Instruction, 4, 217–232.
Ramsden, P. (1997). The context of learning in academic departments. In: F. Marton, D. J. Hounsell,
& N. J. Entwistle (Eds), The experience of learning (2nd ed.) (pp. 198–216). Edinburgh: Scottish
Academic Press (now available at http://www.tla.ed.ac.uk/resources/EoL.html).
Richardson, J. T. E. (2000). Researching student learning: Approaches to studying in campus-based
and distance education. Buckingham: SRHE/Open University Press.
Säljö, R. (1979). Learning in the learner’s perspective – I: Some Common-Sense conceptions.
(Report 76). Gothenburg: University of Gothenburg, Department of Education.
Säljö, R. (1982). Learning and understanding: A study of differences in constructing meaning from
a text. Gothenburg: Acta Universitatis Gothoburgensis.
Säljö, R. (1997). Talk as data and practice: A critical look at phenomenographic inquiry and the
appeal to experience. Higher Education Research and Development, 16, 173–190.
Samuelowicz, K., & Bain, J. D. (2001). Revisiting academics’ beliefs about teaching and learning.
Higher Education, 41, 299–325.
Scheja, M. (2002). Contextualising studies in higher education: First-year experiences of studying
and learning in engineering. Ph.D. thesis. Department of Education, Stockholm University,
Stockholm.
Schommer-Aikins, M. (2002). An evolving theoretical framework for an ‘epistemological belief sys-
tem’. In: B. K. Hofer, & P. R. Pintrich (Eds), Personal epistemologies: The psychology of beliefs
about knowledge and knowing (pp. 103–118). Mahwah, NJ: Lawrence Erlbaum.
Svensson, L. (1977). On qualitative differences in learning – III: Study skill and learning. British
Journal of Educational Psychology, 47, 233–243.
Svensson, L. (1997a). Theoretical foundations of phenomenography. Higher Education Research
and Development, 16, 159–171.
Svensson, L. (1997b). Skill in learning and organising knowledge. In: F. Marton, D. J. Hounsell, &
N. J. Entwistle (Eds), The experience of learning (2nd ed.) (pp. 59–71). Edinburgh: Scottish
Academic Press (now available at http://www.tla.ed.ac.uk/resources/EoL.html).
Thomas, P. R., & Bain, J. D. (1984). Contextual dependence of learning approaches: The effects of
assessments. Human Learning, 3, 227–240.
Van Driel, J. H., Verloop, N., Van Werven, H. I., & Dekkers, H. (1997). Teachers’ craft knowledge
and curriculum innovation in higher engineering education. Higher Education, 34, 105–122.
Vermetten, Y. (1999) Consistency and variability of student learning in higher education. Doctoral
dissertation. Brabant: Catholic University of Brabant.
Vermunt, J. D. (1998). The regulation of constructive learning processes. British Journal of
Vosniadou, S. (1999). Conceptual change research: The state of the art and future directions. In:
(pp. 3–13). Oxford: Pergamon.
Westen, D. (1999). Psychology: Mind, brain and culture (2nd ed.). Chichester: Wiley.
Wilson, E. O. (1998). Consilience. London: Little, Brown & Co.
Wiske, M. S. (Ed.). (1998). Teaching for understanding: Linking research with practice. San
Francisco, CA: Jossey-Bass.
Chapter 12
Conceptual Change in Physics and

Physics-Related Epistemological Beliefs:
A Relationship under Scrutiny
Christina Stathopoulou and Stella Vosniadou
Introduction
There is a diversity of theoretical and conceptual approaches to the construct of ‘episte-

mological beliefs.’ This is reflected by the fact that the construct is assigned different terms
in the literature, such as personal epistemology, epistemic beliefs, ways of knowing, epis-
temological perspectives, epistemological reflection, epistemological thinking, epistemo-
logical theories, epistemological resources, etc. This diversity indicates that what we call
‘epistemological beliefs’1 may not be the same in all relevant studies, or at least, that the
boundaries of the construct may differ (Hofer & Pintrich, 1997; Pintrich, 2002). As
Pintrich points out,
the key issue concerns what should be considered as the core or essence of
personal epistemology and what should be left out of the definition or con-
sidered as related but distinct constructs. (Pintrich, 2002, p. 390)
The present chapter draws on the more conventional definition of epistemology, as the
study concerning “the nature and limits of claims to know” (Harre, 2002), and suggests
that personal epistemological beliefs cluster in two general areas: the nature of knowledge,
including structure and stability of knowledge, and the nature of the process of knowing,
including source and justification of knowing (see also Hofer & Pintrich, 1997). We con-
sider personal epistemological beliefs as individually held theory-like structures, namely,
systems of beliefs that are nonetheless interconnected (see also Hofer & Pintrich, 1997).
1
We have decided to keep the term epistemological beliefs and, alternatively, personal epistemology since this is
widely used by researchers in the field, until there is some consensus on a new terminology.

ISBN: 0-08-045355-4
146 Christina Stathopoulou and Stella Vosniadou
This theoretical position is in line with the conceptual change approach adopted here, and
suggests that personal epistemology forms initially a narrow but relatively coherent set of
beliefs regarding the nature of knowledge and the process of knowing, which is based on
the limited range of children’s initial experiences and information they receive from lay
culture. This set of beliefs becomes gradually more differentiated as experience and/or
cultural information accumulates, gradually changing some beliefs, but not others, and
connecting them to different contexts of use. It should be mentioned that the term theory-
like is used to denote an explanatory structure that can generate explanations and predic-
tions, but which, unlike a scientific theory, is not explicit, well formed or socially shared.
It is assumed that most individuals are not metaconceptually aware of their epistemologi-
cal beliefs.
The theory approach to epistemological beliefs adopted in the present study can be seen
as a bridge between the developmental approach (according to which epistemological
beliefs form a rather coherent, developmental structure that does not allow for within-stage
of development variation (e.g., Baxter Magolda, 1992; Belenky, Clinchy, Golberger, &
Tarule, 1986; King & Kitchener, 1994; Kuhn, 1991; Perry, 1998)), and the multidimen-
sional approach (according to which personal epistemology is a system of rather orthog-
onal, uncoordinated dimensions, that are more or less independent, developing not
necessarily in synchrony (e.g., Schommer, 1990, 1994; Schommer, Crouse, & Rhodes,
1992)). Conceptualizing epistemological beliefs as theory-like structures helps us under-
stand better how they can be acquired and changed, how they can influence individuals’
learning in areas such as physics, and how it is possible to have different epistemological
beliefs in different disciplines (Buehl, Alexander, & Murphy, 2002), since it allows for
general and domain-specific beliefs to co-exist in an interconnected network (see also
Hofer, 2000; Hofer & Pintrich, 1997), which is nonetheless contextually bound.
Regardless of the conceptual and theoretical approach to the construct, changes in epis-
temological thinking have always been conceptualized as involving an ‘upward’ movement
from dualistic/absolutist and objectivist views to more and more relativist, subjectivist, con-
textual, constructivist, and evaluative perspectives of knowledge and knowing (Hofer, 2002;
Hofer & Pintrich, 1997; Pintrich, 2002). In this context, physics-related epistemological
beliefs have been seen to change in a variety of specific ways. Songer and Linn (1991) focus
on what they call ‘productive science views.’ According to their suggestion, scientific
knowledge is a dynamic, socially constructed set of ideas, that progresses through either
evolutionary or even revolutionary changes in perspective and, therefore, it is controversial,
particularly in periods preceding discoveries, and is relevant to the lives of individuals and
societies. Smith, Maclin, Houghton, and Hennessey (2000), following Carey and Smith
(1993), describe the preferred, constructivist, epistemology of science as
an epistemology in which students are aware of the central role of ideas in

the knowledge acquisition process and of how ideas are developed and
revised through a process of conjecture, argument and test. (Smith et al.,
2000, p. 350)
Driver, Leach, Millar, and Scott (1996) think that the more complex and presumably
more mature epistemological views are characterized by ‘model-based reasoning.’
Conceptual Change in Epistemological Beliefs about Physics 147
According to this view, students understand that scientific inquiry involves evaluation of
theories in light of new evidence, and that there may be multiple explanatory
models, involving theoretical entities that cannot be observed. Also, description is clearly
differentiated from explanation. Finally, other researchers emphasize that students’ views
about the nature of scientific knowledge and knowing are strongly influenced by context
(e.g., Elby & Hammer, 2001; Leach, Millar, Ryder, & Sere, 2000). We will use the term
‘constructivist epistemology’ to refer to a set of epistemological beliefs that have more or
less the contextual, constructivist, and evaluative characteristics described above, having,
however, in mind, that the context-dependence of epistemological beliefs challenges the
across-contexts generalizations about the nature of knowledge and knowing.
A constructivist personal epistemology has been found positively related to skills and
attitudes important for learning. Previous studies have demonstrated the relationship of the
construct with comprehension, learning, academic performance, and conceptual change.
For example Ryan (1984), following and extending the work of Perry (1998), investigated
the effect of epistemological development on comprehension and metacomprehension. He
found that students who are relativists in their beliefs about knowledge, i.e., who perceive
knowledge as context dependent, are more successful in comprehension monitoring and
tend to use high-level comprehension strategies, as opposed to dualists (who perceive
knowledge as factual, right or wrong). The latter are more likely to study for recall of facts
from texts. Beliefs concerning the structure of knowledge, that is, considering knowledge
as an accumulation of discrete, concrete, knowable facts, have been found to be related to
poor text comprehension in such areas as the social and physical sciences. They have also
been found to affect the comprehension and related problem solving of statistical text in a
negative way (Schommer, 1990; Schommer et al., 1992). Beliefs concerning the stability/
certainty of knowledge, that is, viewing knowledge as unchanging, attaining/approximat-
ing absolute truth, have been found to negatively affect the interpretation of controversial
evidence (Kardash & Scholes, 1996) and tentative text (Schommer, 1990).
Epistemological Beliefs and Physics Understanding
There is some empirical evidence that supports the position that a constructivist physics epis-
temology facilitates physics understanding. Songer and Linn (1991) found that secondary
students who viewed science as a dynamic process of developing and changing ideas, and
also considered interpretation and integration of ideas as strategies that facilitate learning,
were more likely to understand concepts in thermodynamics and to integrate them around
scientific principles, than students who viewed science as an accumulation of true and
unchanging facts. Qian and Alverman (1995) investigated the influence of secondary
students’ epistemological beliefs on learning counter-intuitive science concepts from a refu-
tational text (Newton’s first law vs. the ‘impetus’ theory). Their results showed that epistemo-
logical beliefs were strongly related to what they call ‘conceptual change learning’ regarding
projectile motion. More specifically, the students who believed in simple-certain knowledge
and quick learning were less likely to change their positions on projectile motion after read-
ing an expository refutational text, compared to the students ‘who viewed knowledge as
complex and evolving’ and learning as a time-consuming process.
In our previous work we investigated the relationship between Greek secondary school
students’ physics-related epistemological beliefs and physics understanding (Stathopoulou
& Vosniadou, 2006). The beliefs of 394 Greek 10th-grade students were measured through
a specially designed paper and pencil questionnaire, the Greek Epistemological Belief
Evaluation instrument for Physics (GEBEP). The results revealed four dimensions under-
lying students’ beliefs: structure of knowledge, construction & stability of knowledge,
attainability of truth, and source of knowledge. In a subsequent study we selected
38 students with the highest scores in all the dimensions underlying the GEBEP (the
constructivist epistemological beliefs group) and 38 students with the lowest scores (the less-
constructivist epistemological beliefs group). These 76 students were administered a reli-
able instrument for measuring their conceptual understanding of Newtonian dynamics, the
Force and Motion Conceptual Evaluation instrument (FMCE) (Thornton & Sokoloff,
1998). The design of the FMCE is based on results received by the extant science educa-
tion research on students’ ‘misconceptions.’ FMCE has been administered to many thou-
sands of secondary and university students in the US. It has also been used extensively in
our research lab with secondary and university physics students in Greece. Thus, substan-
tial information is available that allows us to know, on the basis of the students’ perfor-
mance on this test, when they can be said to have understood Newton’s three laws, in other
words, to have achieved conceptual change regarding force and motion.
The results showed that beliefs regarding the nature of physics knowledge and the
process of knowing are related to conceptual change in physics. More specifically, students’
epistemological beliefs concerning the structure as well as the construction and
stability of physics knowledge were found to predict high scores in the FMCE, suggesting
a deep understanding of Newtonian dynamics. Furthermore, the results showed that only the
students in the constructivist epistemological beliefs group achieved a deep conceptual
understanding of Newtonian dynamics. As shown in Figure 12.1, only 11 students out of
70
27
60
Number of students
50
40 E.B. Group
38
30 Less-constructivist
Epistemological Beliefs
20 (N=38)
10 Constructivist
11
0 (N=38)
0-21 22-43
Physics Conceptual Understanding
on the basis of the FMCE score
Figure 12.1: Only 11 students, all of whom were found to hold constructivist
physics-related epistemological beliefs, were found to achieve high scores in the Force
and Motion Conceptual Evaluation (FMCE) instrument.
the 76 were found to have high scores in the FMCE and all of these students had construc-
tivist physics-related epistemological beliefs. We have interpreted these results to suggest
that a constructivist physics epistemology may be a necessary, although not sufficient, con-
dition for conceptual change in Newtonian dynamics (Stathopoulou & Vosniadou, 2006).
In a third study, we studied three 10th-grade students in the context of a Computer
Supported Collaborative Learning (CSCL) environment2 implemented in the instruction of
Newtonian dynamics (Mol, Stathopoulou, Kollias, & Vosniadou, 2003). These three
students had strong non-constructivist beliefs regarding the source of knowing in physics,
which limited their ability to a more effective use of the CSCL environment. For example,
they believed that they should rely on the authority of the teacher to tell them which was
the correct answer and, as a result, they had great difficulty resolving inconsistencies
and disagreements by themselves through argumentation. Their epistemological beliefs
prevented them from relying on themselves and developing the more reflective approach
necessary in order to resolve, in a principled way, differences of opinion regarding physics
knowledge.
Why may Epistemological Beliefs Facilitate Conceptual Change?
In their influential model of conceptual change as a rational process, Posner, Strike,

Hewson, and Gertzog (1982), suggested four conditions3 for a successful conceptual
change to take place in the learner’s conceptual ecology. The term conceptual ecology was
used to describe the learner’s existing interrelated networks of concepts that influence the
selection of a new concept playing a central and organizing role in thought. Among these,
personally held ‘epistemological commitments,’ namely, assumptions or views concerning
the nature of knowledge and knowing were considered as playing an important role. In a
‘revisionist’ approach of the initial overtly rational model of conceptual change, Strike and
Posner suggested that “motives and goals and their institutional and social sources need
to be considered” (1992, p. 162) as well, in attempting to describe a learner’s evolving
conceptual ecology and understand the construct’s impact on conceptual understanding.
The need to incorporate variables of motivational and affective character, such as per-
sonal beliefs and attitudes,4 into models of conceptual change is stressed by
cognitive/developmental psychologists who also go beyond an approach that emphasizes
the overtly rational nature of conceptual change (e.g., Pintrich, Marx, & Boyle, 1993;
Pintrich, 1999; Dole & Sinatra, 1998; Gregoire, 2003; Sinatra, 2005). Hofer and Pintrich
(1997) suggested that epistemological beliefs functioning as implicit theories interacting
with the educational context can influence academic achievement indirectly by affecting
goal orientation. In other words, epistemological beliefs can give rise to certain types of
learning goals, such as mastery, performance, and completion goals, which in turn, can
function as guides for cognitive and metacognitive strategy use.
2
The research was performed as part of the European Project ITCOLE (http://www.euro-cscl.org/site/itcole )
3
That is, dissatisfaction with the current concept and also the ineligibility, plausibility, and fruitfulness of the new
concept.
4
For example, beliefs about the nature of knowledge, and knowing, beliefs about learning, about the role of self
as learner, goal orientation, motivation to engage in academic tasks, interest/values.
The conceptual change approach adopted in this study considers that, in many cases,
the construction of a scientific concept, such as the concept of force, requires students to
radically reorganize prior knowledge (Carey, 1985, 1992; Vosniadou, 1999, 2002, 2003;
Vosniadou, in press). This is the case, because by the time they are exposed to systematic
science instruction, students have already constructed an initial naïve physics based on
their everyday observations and cultural experiences. This naïve physics is very different
from the scientific theories to which they are exposed at school and can stand in the way
of learning physics. We explain the phenomenon of ‘misconceptions’ observed in science
classrooms, at all levels of education, as resulting from students’ attempts to add the new,
to-be-acquired information to an incompatible knowledge base, thus forming synthetic
models (Ioannides & Vosniadou, 2002; Vosniadou, 1999, 2002). We believe that concep-
tual change is a slow and gradual process that not only involves cognitive factors but is also
influenced by motivational and affective variables, such as personal beliefs and attitudes,
as well as by the physical and social/cultural environment (Dole & Sinatra, 1998; Pintrich,
1999; Sinatra, 2005; Vosniadou, 2002, 2003).
According to our theoretical position, physics-related epistemological beliefs can influ-
ence the knowledge acquisition process directly or indirectly, just like ontological presup-
positions and other beliefs of a motivational and affective character can do. They can
influence both the kind of new information that is picked up from the physical and socio-
cultural context and the way in which this information is interpreted (Vosniadou, 1994,
2002, 2003; Vosniadou & Brewer, 1994). For example, beliefs in simple and/or certain
knowledge may affect the learning process directly by focusing students’ attention on
factual information, while beliefs in complex and/or evolving knowledge may cause stu-
dents to focus more on patterns of relationships and their change over time. Figure 12.2
presents a skeletal theoretical framework for conceptualizing the relationship between
physics-related epistemological beliefs and physics conceptual understanding.
Physical and Socio-Cultural Context
Presuppositions- New Cultural

Commitments-Beliefs & Observational
Information
Ontological
Epistemological Cognitive&
Metaconceptual
Motivational Processes
Strategy use
Metaconceptual Conceptual
awareness Understanding
in Physics
Self-Regulation
Figure 12.2: A skeletal theoretical framework for conceptualizing the relationship

between (physics-related) epistemological beliefs and physics conceptual understanding.
Epistemological, ontological, and motivational beliefs may also affect students’

achievement indirectly, by influencing students’ learning goals, study strategies, and self-
regulation, as discussed earlier. For example, the belief in simple knowledge may lead to
the selection of rehearsal strategies to strengthen memorization and recall of piecemeal
factual information. The skeletal framework presented in Figure 12.2 is based on prior
work by Vosniadou regarding conceptual change in physical science (e.g., Vosniadou,
1994, 2003), and also takes into consideration some current suggestions about the role of
motivational and affective as well as social/cultural variables (Pintrich, 1999; Pintrich
et al., 1993; Strike & Posner, 1992).
A Case Study: How do Study Strategies Intervene?
In the pages that follow we present preliminary results from a case study which attempted,
through in-depth interviews, think-alouds, and observations, to understand the indirect
effect of physics-related epistemological beliefs on physics understanding. The in-depth
interviews were administered to 10 students, who were selected out of the 76 students in
the Stathopoulou and Vosniadou (2006) study that investigated the relationship between
physics-related epistemological beliefs and physics understanding. These 10 students were
selected on the basis of their scores in the GEBEP, as well as in the FMCE. Five of the stu-
dents had high scores in all dimensions underlying the GEBEP and also high scores in the
FMCE, thus comprising the group of students who were considered to hold constructivist
physics epistemologies and to have achieved an in-depth understanding of physics. In con-
trast, the remaining five had low scores in all dimensions underlying the GEBEP and also
low scores in the FMCE, thus comprising the group of students considered to hold less-
constructivist physics epistemologies and to have poor/superficial physics understanding.
It was hypothesized that the approach to learning and studying adopted by the students,
that is, deep vs. superficial (e.g., Entwistle, this volume; Entwistle, Tait, & McCune, 2000),
and the related selection of study strategies, may intervene in the relationship between per-
sonal physics-related epistemological beliefs and conceptual change in physics. More
specifically, we hypothesized that a constructivist physics personal epistemology is more
likely to guide students to the adoption of a deep approach to learning and studying, and
therefore, to facilitate physics understanding, than a less-constructivist epistemology. Of
course, the relationship between personal epistemology and physics understanding is
likely to be a reciprocal one. As students develop a deeper understanding in physics their
personal physics-related epistemologies would be likely to change. On the basis of these
hypotheses, the five students who were found to score high on both the GEBEP and the
FMCE were expected to adopt a deep approach to studying, as opposed to the remaining
five students, with low scores on both the GEBEP and the FMCE, who were expected to
adopt a superficial approach to studying.
Following Entwistle (e.g., Entwistle, this volume; Entwistle et al., 2000), a deep
approach to learning and studying involves goals of personal making of meaning, and
accordingly, deep strategy use, such as integration of ideas, looking for patterns and under-
lying principles, examining in detail evidence and logic, and monitoring of understanding.
It also involves metaconceptual awareness, that is, awareness of one’s own beliefs. A deep
approach to learning and studying may also involve some performance goals, such as time
management and organized studying that do not contradict the goal of personal making of
meaning. In contrast, a superficial approach to learning is characterized by performance
orientation, lack of purpose, and superficial strategy use, such as memorization of facts,
and syllabus boundness. It is also characterized by lack of metaconceptual awareness.
Method
Participants
Ten 10th-grade Greek students participated in this study, six of whom were boys and four
were girls. As mentioned earlier, these students were selected from a pool of students that
participated in the Stathopoulou and Vosniadou (in press) study, on the basis of their scores
in the GEBEP and the FMCE. Five of the students were selected because they had
achieved high scores in all dimensions underlying the GEBEP and also high scores in the
FMCE, whereas the remaining five were selected because they had achieved low scores in
all dimensions underlying the GEBEP and also low scores in the FMCE.
Materials and Procedure
The interview started with a discussion with each participating student about the nature of
physics knowledge, the role of teachers, textbooks, peers and the self in learning physics,
as well as about the role of experience and experiment in the justification of knowing. This
was done in order to re-examine the participants’ physics-related epistemologies and to
reveal aspects of students’ approaches to physics learning and studying. The discussion
also concerned the interviewees’ attitudes towards physics and their study goals and strate-
gies. The second part of the interviews involved discussion, think-alouds, and observation
as each student was faced with tasks, such as answering questions regarding particular
dynamics-related situations, and solving problems in the area of Newtonian dynamics. We
wanted in this way to re-examine the depth of students’ conceptual understanding of
dynamics, but most importantly, to investigate in a context-sensitive way students’
approaches to learning and studying and particularly their strategies in the context of a
problem-solving task. Some examples of the physics questions and problems that were
used during the interviews are shown in Figure 12.3. We selected the particular questions
and problems because they targeted some well-known ‘misconceptions’ of students, such
as the ‘impetus misconception.’
Each student was interviewed for about two hours. Data regarding students’ grades in
school physics were also collected.
Results
The analysis of the interviews gave results consistent with those received by the instruments
that were earlier used to measure epistemological beliefs and physics understanding
Identify the forces acting on the ball.
i. Identify the forces acting on the sliding

down girl.
ii. In case that the boy on the top of the
slide had pushed the girl instantly, do
you believe that the same, as previously,
forces would act on her, as she slides
down the slide? Explain.
iii. Consider the slide as a frictionless
inclined plane of given length and angle.
If you also know the girl’s mass and
weight can you find her speed at the
bottom of the slide? Explain.
An object with a mass of 2 Kg moves with a speed of 3 Km / h. Can you find the net
external force on the object so that it keeps moving in the same direction with the
speed of 3 km / h? If yes, find it. If not, what other information should you be given?
Figure 12.3: Some examples of physics questions and problems used in the interviews.
(i.e., the GEBEP and the FMCE respectively). More specifically, all five students who were
selected because they scored high in both the GEBEP and the FMCE, were indeed found to
hold constructivist physics epistemologies and to have achieved conceptual change in
Newtonian dynamics, in contrast to the remaining five students who were selected for their
low scores in the GEBEP and the FMCE. Further analysis of the interview-data was done
for the purpose of identifying patterns of responses pointing to what may be described as
deep approach to studying, as opposed to superficial approach to studying.
Preliminary results showed, as expected, that all five students who were found to hold
constructivist physics epistemologies, and also to have achieved conceptual change in
physics (in the area of Newtonian dynamics), had adopted what may be considered a deep
approach to learning and studying. The remaining five students showed evidence of adopting
a superficial approach to learning and studying. The criteria for identifying students’
approaches were defined, as mentioned earlier, on the basis of Entwistle’s descriptions
of a deep as opposed to a superficial approach to learning and studying (Entwistle, this
volume; also, Entwistle et al., 2000) and are presented in Table 12.1. They were categorized
around three dimensions of learning and studying: goals, strategies, and metaconceptual aware-
ness. A deep approach to learning and studying involves goals of personal making of meaning,
and deep strategy use, such as integration of ideas. It also involves awareness of one’s own
beliefs and thus, a good sense of understanding or failure to understand. A superficial approach
involves performance goals, lack of purpose, and superficial strategy use, such as rote learning.
We also assume that this approach is accompanied by low awareness of one’s own beliefs.
To illustrate the differences in approaches to learning and studying adopted by students
with qualitatively different physics-related epistemological beliefs, two students, John and
Michael, were selected as examples. Below we present ‘in greater details’ results from the
interviews of these two students.
A closer look at John and Michael
The two students who served as examples in the present study, John and Michael, were
very different in terms of their personal physics epistemologies and their conceptual under-
standing of dynamics, but both had very high grades in school physics. Comparable per-
formance in school physics was important for selecting the two students because
controlling for this variable has the potential to help us understand better the influence of
physics-related epistemological beliefs on physics understanding. John had scored high on
all the factors underlying the GEBEP and also impressively high (40 out of 43) on the
FMCE, while Michael had scored low on all the factors underlying the GEBEP and also
very low (6 out of 43) on the FMCE. In what follows, we present excerpts from the tran-
scribed interviews with the two students, in order to clarify the differences in their
approaches to learning and studying. We will focus on the three criteria used to identify
students’ approach to learning and studying, described in Table 12.1.
Table 12.1: Criteria for identifying students’ approach to learning and studying.
Approaches to learning Criteria
and studying
Goals Strategy use Metaconceptual
awareness
Deep approach Meaning-making Integrating ideas Being aware of

one’s own
beliefs
Superficial approach Performance Rote learning Not being
orientation or aware of
lack of purpose one’s own
beliefs
Goals: Meaning Making vs. Performance Orientation John appears to seek personal
making of meaning, by trying to work things out for himself, to relate them to what he
already knows.
John: Physics is relatively difficult for the average student because (…) you
have to learn things, not by rote, but to learn, to put them in your mind. You
need time to consolidate things…
Interviewer: What do you mean by saying “to put them in your mind” and
“to consolidate things”?
John: I mean to relate them, that is, you don’t need to have [separate] laws
for force and motion, and later other for energy and other for momentum.
If you don’t relate them you can’t learn them really good.
His meaning seeking is demonstrated by the fact that he monitors his performance and
takes corrective steps, when he realizes failures in understanding:
John: [Sometimes] you realize that something is missing, or that you

haven’t really understood [something]. You have to go back, I mean, I don’t
know what most students do, but I will go back, because if I haven’t under-
stood, I can’t go on.
Michael, on the other hand, appears to be performance-oriented in the sense that he wor-
ries not really about understanding for himself, but rather, about doing what, he thinks, the
physics teacher expects of him. This seems to be his major concern and having the
teacher’s approval is what he is after. This is rather clear in the following excerpt:
Michael: When the physics teacher asks something, regardless of whom he

asks, I say the answer to myself and if I was quicker in answering [than the
other student], and the answer was correct, I believe that it is OK. I’m sure
that I know that.
Interviewer: Is this enough to be sure that everything is OK? Is there any-
thing else [about your learning] to worry about?
Michael: (laughing) No, no, no, and if the teacher tells you, which is not
usual, “you’re doing well, keep working” then, I believe, that you are at a
good level.
Strategies: Integrating Ideas vs. Rote Learning
John appears to adopt the deep strategy use such as integration of ideas on the basis of
some organizing principles. He also admits that he finds it rather intriguing to try to relate
ideas and to play around with ideas.
John: Our mind can remember a lot of things, but you cannot overload
it, so, you can keep in your mind some basic formulas that you can eas-
ily store in your memory (…) You can relate different parts of a theory.
Thus you can relate formulas with other things that may help you to
reproduce any formula needed (…) Formulas derive from theory. It’s
easier to remember the formulas if you understand the theory. (…) [You
know that linear velocity in uniform circular motion is equal to 2!r/ T
rather than 2!T/r] because you remember from theory that linear velo-
city equals the arc length divided by the [corresponding] time. You know
that the length of a circle is 2!r and the corresponding time equals the
period T, thus…
And later, when he was asked to identify the forces exerted on a girl sliding down a
slide, and to find her speed at the bottom of the slide, he said:
John: (having drawn the forces, in the scientifically accepted way, without
difficulty) Right… her speed. Let me see, I can work either with kinematic
equations or with energy (He starts working with kinematics, suggesting
that there is friction between the slide and the girl).
Interviewer: Suggest the friction is negligible.
John: Then I would work with energy.
Interviewer: I see, [in that case] you would prefer a different way.
John: I don’t mind actually, both ways are good.
Also, in another instance:
John: [You may engage yourself in solving a problem that is not an assign-
ment and looks strange or intriguing, because] it is the curiosity to deal
with something that may not come to an end but you want simply to play
around with this, to try this out.
In contrast, Michael appears to adopt superficial strategy use, such as rehearsal and
memorization:
Michael: I have my own way, I mean with some formulas that are difficult
[to remember] I find some key words and I keep them in mind so that when
is needed I can remember exactly what it was about. Or by writing formu-
las, if you write them many times, I believe that your hand is getting used
[to writing them] and at the moment you must write them it is easier, they
come back in memory.
Later, answering a question concerning what he does when he simply cannot remem-
ber a formula; for example, how he can decide whether the linear velocity in uniform cir-
cular motion is equal to 2!r/T or to 2!T/r he said:
Michael: I believe that if you have learned a formula well, from my per-
sonal experience, because I’m good with formulas and I can remember
them, hum I’m familiarized…, but I believe that if my mind gets stuck,
the student will try to recall it; if he doesn’t, he will try to solve the
problem and (…) he will get such a result that he will understand that
something’s wrong.
Interviewer: This is what you do?
Michael: I am sure that I will write the formulas [right], because I know
them, I mean, it is so easy to write them if you have learnt them (…) I have
no trouble with formulas. I write the formula and everything goes right, the
problem is easily solved (…). The problem goes itself, for me it is something
that you can learn, what to do first, second, third, and the problem is solved
like that. I mean it’s so easy.
In another instance, Michael was given the mass and the speed of a moving object and
was asked to find the net force on the object in case it kept moving in the same direction
with the same speed.
Michael: I usually have more data given. I believe that I should have been
given a force F so that I could find the resultant force in the x direction, to
find in the vertical direction that the weight equals the [normal force] N
and then to find the friction, classical methodology. Something is missing.
I believe that I should be given the F.
Being Metaconceptually Aware vs. Not Being Metaconceptually Aware Another impor-
tant aspect of a deep approach to learning and studying is awareness of one’s own beliefs.
Being aware of one’s own beliefs, means being also able to recognize different points of
view. John is aware of his own beliefs and of the change in his beliefs. For example, he
knows that he has abandoned the ‘impetus misconception,’ that is, that there must be
always a force in the direction of the movement (McCloskey, 1983; Gunstone & Watts,
1985; Vosniadou, 2002; Ioannides & Vosniadou, 2002). Thus, when John is asked to iden-
tify the forces on a ball that was hit by a football player he says:
John: Now there is no force from the football player, it is away from him,
the only force is its weight. Yes there is no force [from the football player],
there is a velocity that was acquired when it was in contact with him, when
he exerted a force on the ball.
Interviewer: Is this a spontaneous response? I mean, are there any cases
when you are about to say that there is also an acquired force and then you
say “no there isn’t [such a force]”?
John: Yes, I haven’t got away [from this belief] completely, but when I think
about it I understand that there isn’t such a force, since there is no contact
[with the football player]. I mean with a first thought, I may be confused,
but with a second, no I don’t.
Michael, on the other hand, is not aware of his own beliefs and their development. The
following dialogue, when he was also asked to identify the forces on a ball that was
headed by a football player is revealing, as it shows that he is not aware of his failure to
understand:
Michael: This body has definitely a weight and we can see its direction
and put another force F (he draws a force F in the direction of the ball’s
movement)
Interviewer: I can see why you say “there is definitely its weight” it is like
saying that the ball is in the field of gravity, but why do you think there is
also this force F?
Michael: We can see that he hits the ball with his head and now the ball is
free [to move]
Interviewer: Yes, the ball is not in contact with his head isn’t it?
Michael: (crossing out the force F) Yes it has left his head.
Interviewer: I see that you cross out the force F.
Michael: Yes, because I made a mistake, there is not such a force; there is
its weight and a backward force. We can say that it is the reaction from the
air that slows down the ball.
Interviewer: If we consider this force negligible Michael, how does it move?
Michael: It slows down, ah we said [it is] negligible, hum it falls freely?
Interviewer: It moves downwards, vertically?
Michael: Let’s say this is the ground, right? It will move…I forgot, I don’t
remember how it moves.
Later, when Michael was faced with the problem concerning the girl sliding down a
slide, he drew a force in the direction of the movement and when he came to a dead end
due to the lack of information about such a force, he wondered whether he should cross
out that force.
In short, John, the student with a constructivist physics-related epistemology, and
the one showing evidence of in-depth physics understanding, adopted a deep approach
to learning and studying as exemplified by his tendency to seek personal meaning-
making, to actively monitor his understanding, to integrate ideas and also, to be aware of
his own beliefs and of changes in his beliefs. The opposite, that is, the adoption of a
superficial approach to learning and studying, was found for Michael, the student with a
less-constructivist physics epistemology who showed evidence of poor conceptual
physics understanding. Michael appeared to be performance-oriented, to rely on memo-
rization of what has to be learned and rote learning, not to be concerned about integrating
ideas and finally, not to be aware of his own beliefs.
Discussion
The study presented in this chapter provided preliminary evidence that the adopted
approach to learning and the consequent selection of study strategies may intervene in the
relationship between epistemological beliefs and conceptual change. Of the 10 students
investigated, the five who held constructivist physics epistemologies and had achieved
conceptual change in dynamics, adopted a deep approach to learning and studying. They
were all oriented to personal meaning-making through the selection of deep study strate-
gies. This process paralleled an increasing metaconceptual awareness. In contrast, the
remaining five students who had less-constructivist physics epistemologies and were far
from having achieved deep conceptual understanding of dynamics, showed evidence of a
superficial approach to learning and studying. They were performance-oriented and pre-
ferred the selection of superficial study strategies, while no evidence of substantial meta-
conceptual awareness was found. The examples of John and Michael helped us to better
understand these differences.
The findings of the present study are in line with Entwistle’s (this volume) suggestion
that the development of conceptions of knowledge (dualistic vs. relativistic) parallels the
development of conceptions of learning (reproducing vs. seeking meaning), and that this
is a process of increasing metacognitive awareness. They also agree with the hypotheses
emerging from the suggested theoretical framework according to which epistemological
beliefs may either facilitate or constrain the knowledge acquisition process directly,
through guiding attention to certain information and through influencing intentions regard-
ing knowledge construction and revision, as well as indirectly through certain mediating
cognitive, metacognitive, and/or motivational factors, such as goal orientation and study
strategies (e.g., Vosniadou, 1994, 2003, in press; Sinatra & Pintrich, 2003; Mason, 2003;
Pintrich, 1999; Dole & Sinatra, 1998; Sinatra, 2005).
Students’ belief that physics knowledge is a piecemeal collection of factual information,
rather than a complex system of interrelated concepts, may make them more likely to
‘early foreclose’ (Kruglanski, 1989) their critical thinking in the process of learning
(Pintrich et al., 1993; Pintrich, 1999; Qian & Alverman, 1995). It is also possible that
students with such a predisposition may be precluded from deep information processing
and from developing useful metacognitive skills. They would thus be expected to use
superficial strategies (e.g., memorization of facts and formulas) instead of deeper strate-
gies (e.g., organization and integration on the basis of principles). Such strategies lead to
inert knowledge and prevent successful transfer. At the metacognitive level, they would be
expected not to be aware of the need for information management and evaluation of learn-
ing outcomes. In any case, a significant impact on conceptual change in physics would be
expected. Schommer et al. (1992) also suggested that belief in simple, as opposed to com-
plex, knowledge may influence both students’ selection of cognitive strategies during the
process of learning and their standards for judging learning outcomes.
With regard to the finding that students’ beliefs in unchanging physics knowledge are
also related to a superficial physics understanding, it could be argued that this may be the
case because such a belief may constrain the evaluation and filter the interpretation of tenta-
tive and controversial information that does not concur with existing knowledge
(e.g., Schommer–Aikins, 2002; Qian & Alverman, 1995). Students, who believe that physics
knowledge does not change, may prefer to avoid ‘threatening’ new information, rather than
examining and changing their existing conceptions. Similar explanations have been provided
in the literature to account for the way students respond to anomalous data in conditions of
cognitive conflict (e.g., Chinn & Brewer, 1993; Chinn & Malhotra, 2002; Mason, 2000).
Finally, the results of our studies also suggest that epistemological beliefs maybe better
predictors of conceptual change in physics than grades in physics (Stathopoulou &
Vosniadou, in press). As we saw, the two students who served as examples in the present
study, John and Michael, were very different in terms of their personal physics episte-
mologies and their conceptual understanding of Newtonian dynamics, but both had very
high grades in school physics. Previous studies with Greek 1st-year university physics stu-
dents have also shown that about half of them could not answer in the scientifically
accepted way all the FMCE questions concerning the three Newton’s laws of motion,
despite the fact that they took demanding entrance examinations and were selected on the
basis of their high grades in school physics (Mol, Stathopoulou, & Vosniadou, 2004). This
may be the case, because high grades can result not only from in-depth physics under-
standing but also from such factors as efficient use of rules, formalistic and algorithmic
approach to problem-solving, adaptation to the teacher’s preferred techniques, rote learn-
ing, or what could be called a ‘strategic approach’ to learning and studying (Entwistle
et al., 2000). Thus, it seems reasonable to suggest that a student with a deep conceptual
understanding in physics would be expected to have high grades in school physics but the
opposite may not necessarily be the case.
To conclude, it appears that epistemological beliefs influence conceptual change in a
variety of different ways. Understanding these ways involves more than an approach to
conceptual change as an overtly rational process. The conflict between what is already
known and the new, to-be-acquired information creates a learning situation in which affec-
tive and motivational variables can play an important role (Dole & Sinatra, 1998; Gregoire,
2003; Pintrich et al., 1993; Pintrich, 1999; Sinatra, 2005; Vosniadou, 2003). The relation-
ship between epistemological beliefs and conceptual change is likely to be, to some extent,
a reciprocal one (Pintrich, 2002). Since personally held beliefs about knowledge and
knowing (in physics) are themselves subject to change, it is rather reasonable to suggest
that deeper understanding (in physics) may provide feedback that influences epistemolog-
ical beliefs.
The exact processes through which epistemological beliefs change is not addressed in
this study, but it is definitely an issue that needs careful investigation. A number of
researchers emphasize the importance of constructivist instruction in facilitating the devel-
opment of personal epistemologies (e.g., Bell & Linn, 2002; Carey & Smith, 1993, Roth &
Roychoudhury, 1994; Smith et al., 2000). As noted earlier, conceptualizing epistemological
beliefs as theory-like structures can help us understand better the mechanisms of their
change by drawing on various cognitive mechanisms, as well as on motivational and affec-
tive variables that are involved in conceptual change models (e.g., Dole & Sinatra, 1998;
Gregoire, 2003; Pintrich, 1999; Pintrich et al., 1993; Sinatra, 2005; Vosniadou, 1994, 2003).
References
Baxter Magolda, M. B. (1992). Knowing and reasoning in college: Gender-related patterns in stu-
dents’ intellectual development. San Francisco, CA: Jossey-Bass.
Belenky, M. F., Clinchy, B. M., Golberger, N. R., & Tarule, J. M. (1986). Women’s ways of know-
ing: The development of self, voice, and mind. New York: Basic Books.
Bell, P., & Linn, M. C. (2002). Beliefs about science: How does science instruction contribute?
In: B. K. Hofer, & P. R. Pintrich (Eds), Personal epistemology: The psychology of beliefs about
knowledge and knowing (pp. 347–363). Mahwah, NJ: Lawrence Erlbaum Associates.
Buehl, M. M., Alexander, P. A., & Murphy, P. K. (2002). Beliefs about schooled knowledge: Domain
specific or domain general? Contemporary Educational Psychology, 27, 415–449.
Carey, S. (1992). The origin and evolution of everyday concepts. In: R. N. Giere (Ed.), Cognitive
models of science (pp. 89–128). Minneapolis, MN: University of Minnesota Press.
Carey, S., & Smith, C. (1993). On understanding the nature of scientific knowledge. Educational
Psychologist, 28(3), 235–251.
Chinn, C. A., & Brewer, W. F. (1993). The role of anomalous data acquisition: A theoretical
framework and implications for science instruction. Review of Educational Research, 63(1),
1–49.
Chinn, C. A., & Malhotra, B. A. (2002). Children’s responses to anomalous scientific data: How is
conceptual change impeded? Journal of Educational Psychology, 94(2), 327–343.
Dole, J., & Sinatra, G. (1998). Reconceptualizing change in the cognitive construction of knowledge.
Educational Psychologist, 33, 109–128.
Driver, R., Leach, J., Millar, R., & Scott, P. (1996). Young people’s images of science. Buckingham:
Open University Press.
Elby, A., & Hammer, D. (2001). On the substance of a sophisticated epistemology. Science
Education, 85(5), 554–567.
Entwistle, N., Tait, H. & McCune, V (2000). Patterns of response to approaches to studying inven-
tory across contrasting groups and contexts. European Journal of Psychology of Education, 15(1),
33–38.
Gregoire, M. (2003). Is it a challenge or a threat? A dual-process model of teachers’ cognition and
appraisal processes during conceptual change. Educational Psychology Review, 15(2), 147–179.
Gunstone, R., & Watts, M. (1985). Force and motion. In: R. Driver, E. Guesne, & A. Tibergien (Eds),
Children’s ideas in science. Milton Keynes, England: Open University Press.
Harre, R. (2002). Cognitive science. A philosophical introduction. London: Sage.
Hofer, B. K. (2000). Dimensionality and disciplinary differences in personal epistemology.
Contemporary Educational Psychology, 25, 378–405.
Hofer, B. K. (2002). Personal epistemology as a psychological and educational construct: An intro-
duction. In: B. K. Hofer, & P. R. Pintrich (Eds), Personal epistemology: The psychology of beliefs
about knowledge and knowing (pp. 3–14). Mahwah, NJ: Lawrence Erlbaum Associates.
Hofer, B. K., & Pintrich, P. R. (1997). The development of epistemological theories: Beliefs about
knowledge and knowing and their relation to learning. Review of Educational Research, 67(1),
88–140.
Ioannides, C., & Vosniadou, S. (2002). The changing meanings of force. Cognitive Science
Quarterly, 2(1), 5–62.
Kardash, C. M., & Scholes, R. J. (1996). Effects of pre-existing beliefs, epistemological beliefs and
need for cognition on interpretation of controversial issues. Journal of Educational Psychology,
88, 260–271.
King, P. M., & Kitchener, K. S. (1994). Developing reflective judgement: Understanding and pro-
moting intellectual growth and critical thinking in adolescents and adults. San Francisco, CA:
Jossey-Bass.
Kruglanski, A. W. (1989). Lay epistemics and human knowledge: Cognitive and motivational bases.
New York: Plenum Press.
Kuhn, D. (1991). The skills of argument. Cambridge: Cambridge University Press.
Leach, J., Millar, R., Ryder, J., & Sere, M. G. (2000). Epistemological understanding in science
learning: The consistency of representations across contexts. Learning and Instruction, 10,
497–527.
Mason, L. (2000). Role of anomalous data and epistemological beliefs in middle students’ theory
change on two controversial topics. European Journal of Psychology of Education, 15, 329–346.
Mason, L. (2003). Personal epistemologies and intentional conceptual change. In: G. M. Sinatra, &
P. R. Pintrich (Eds), Intentional conceptual change (pp. 199–236). Mahwah, NJ: Lawrence
Erlbaum Associates.
McCloskey, M. (1983). Naive theories of motion. In: D. Gentner, & A. L. Stevens (Eds), Mental
models. Hillsdale, NJ: Erlbaum.
Mol, A., Stathopoulou, C., Kollias, V. P., Vosniadou, S. (2003, July). Gradual learning of science in
a CSCL environment and the quest of epistemologically sophisticated learners. Paper presented at
the 3rd IEEE International Conference on Advanced Learning Technologies (ICALT’03), Athens,
Greece.
Mol, A., Stathopoulou, C., & Vosniadou, S. (2004). Consistency versus fragmentation in Mechanics.
In: S. Vosniadou, C. Stathopoulou, X. Vamvakoussi, & N. Mamalougos (Eds), Abstracts of the 4th
European symposium of the European association for research on learning and instruction on
conceptual change: Philosophical, historical, psychological and educational approaches
(pp. 133–135). Athens: Gutenberg Press.
Perry, W. C., Jr. (1998). Forms of intellectual and ethical development in the college years: A
scheme. San Francisco, CA: Jossey-Bass.
Pintrich, P. R. (1999). Motivational beliefs as resources for and constrains on conceptual change. In:
(pp. 33–50). Oxford: Elsevier.
mologies. In: B. K. Hofer, & P. R. Pintrich (Eds), Personal epistemology: The psychology of
beliefs about knowledge and knowing (pp. 103–118). Mahwah, NJ: Lawrence Erlbaum
Associates.
Pintrich, P. R., Marx, R. W., & Boyle, R. A. (1993). Beyond cold conceptual change: The role of
motivational beliefs and classroom contextual factors in the process of conceptual change. Review
of Educational Research, 63(2), 167–199.
conception. Toward a theory of conceptual change. Science Education, 66(2), 211–227.
Qian, G., & Alvermann, D. (1995). Role of epistemological beliefs and learned helplessness in
secondary school students’ learning science concepts from text. Journal of Educational
Psychology, 87(2) 282–292.
Roth, W. M., & Roychoudhury, A. (1994). Physics students’ epistemologies and views about know-
ing and learning. Journal of Research in Science Teaching, 31 (1), 5–30.
Ryan, M. P. (1984). Monitoring text comprehension: Individual differences in epistemological stan-
dards. Journal of Educational Psychology, 76(2), 248–258.
Schommer, M. (1994). An emerging conceptualization of epistemological beliefs and their role in
learning. In: R. Garner, & P. Alexander (Eds), Beliefs about text and instruction with text
(pp. 25–40). Hillsdale, NJ: Erlbaum.
Schommer–Aikins, M. (2002). An evolving theoretical framework for an epistemological belief
system. In: B. K. Hofer, & P. R. Pintrich (Eds), Personal epistemology: The psychology of beliefs
Schommer, M., Crouse, A., & Rhodes, N. (1992). Epistemological beliefs and mathematical text
comprehension: Believing it is simple does not make it so. Journal of Educational Psychology,
84(4), 435–443.
Sinatra, G. M. (2005). The “warming trend” in conceptual change research: The legacy of Paul
Pintrich. Educational Psychologist, 40(2), 107–115.
Sinatra, G. M., & Pintrich, P. R. (2003). The role of intentions in conceptual change learning. In:
G. M. Sinatra, & P. R. Pintrich (Eds), Intentional conceptual change (pp. 1–18). Mahwah, NJ:
Lawrence Erlbaum Associates, Inc.
Smith, C. L., Maclin, D., Houghton, C., & Hennessey, M. G. (2000). Sixth-grade students’ episte-
mologies of science: The impact of school science experiences on epistemological development.
Cognition and Instruction, 18(3), 349–422.
Songer, N. B., & Linn, M. C. (1991). How do students’ views of science influence knowledge inte-
gration? Journal of Research in Science Teaching, 28 (9), 761–784.
Stathopoulou, C., & Vosniadou, S. (in press). Exploring the relationship between physics-related
epistemological beliefs and physics understanding. Contemporary Educational Psychology.
Strike, K. A., & Posner, G. J. (1992). A revisionist theory of conceptual change. In: R. A. Duschl, &
R. J. Hamilton (Eds), Philosophy of science, cognitive psychology and educational theory and
practise (pp 147–176). New York: State University of New York Press.
Thornton, R. K., & Sokoloff, D. R. (1998). Assessing student learning of Newton’s laws: The force
and motion conceptual evaluation and the evaluation of active learning laboratory and lecture
curricula. American Journal of Physics, 66(4), 338–352.
Vosniadou, S. (1994). Capturing and modeling the process of conceptual change. Learning and
W. Schnotz, S. Vosniadou, & M. Carretero (Eds), New perspectives on conceptual change.
Oxford: Elsevier.
Vosniadou, S. (2002). On the nature of naive physics. In: M. Limon, & L. Mason (Eds),
Reconsidering conceptual change: Issues in theory and practice (pp. 61–76). Dordrecht, The
Netherlands: Kluwer.
Vosniadou, S. (2003). Is intentional learning necessary for conceptual change? In: G. Sinatra, &
P. Pintrich (Eds), Intentional conceptual change. Hillsdale, NJ: Erlbaum.
Vosniadou, S. (in press). Conceptual change and education. Human Development.
Vosniadou, S., & Brewer, W. F. (1994). Mental models of the day/night cycle. Cognitive Science, 18,
123–183
Chapter 13
Effects of Epistemological Beliefs and

Learning Text Structure on Conceptual
Change
Introduction
Classroom learning often implies the re-organization of students’ existing knowledge

structures, that is, conceptual change. Research in this field is carried out by cognitive,
developmental, and educational psychologists, as well as by educators, especially in
science domains, who are interested in the learning processes that are activated when
disciplinary contents are taught in the classroom. Different models have been elaborated
to explain why students hold alternative conceptions and have difficulty revising them
(e.g., Chi, Slotta, & de Leeuw, 1994; Dole & Sinatra, 1998; Vosniadou, 1994). An impor-
tant issue that arises from conceptual change research is that several factors are involved
in the complex and dynamic process of knowledge revision. These include not only
learners’ cognitive, motivational, and affective characteristics, but also social, instruc-
tional, and contextual aspects of learning situations (Limón & Mason, 2002; Schnotz,
Vosniadou, & Carretero, 1999; Sinatra, 2005; Sinatra & Pintrich, 2003a). This study aimed
at extending current understanding about the role of two of these variables, one related to
the learners, that is their epistemological beliefs, and the other related to the material to be
understood, that is the structure of the text to be learned. These two factors have received
attention only relatively recently and their effects have been examined separately (Qian &
Pain, 2002). On the one hand, it has emerged that beliefs about the nature of knowledge
and knowing can constrain or facilitate revision of existing knowledge (e.g., Qian &
Alvermann, 2000). On the other hand, it has been shown that the use of innovative learn-
ing texts, such as “refutational” ones, can be effective in knowledge restructuring
processes (e.g., Guzzetti, Snyder, Glass, & Gamas, 1993). Since most school learning is
the outcome of studying texts, the design of effective learning texts is a crucial question.
Issues from the two different lines of investigation have been taken into account in the
present study by focusing on the effects on conceptual change of epistemological beliefs
and the design of refutational texts.

ISBN: 0-08-045355-4
166 Lucia Mason and Monica Gava
Epistemological Beliefs and Conceptual Change
When confronted with new knowledge to be learned, students do not just activate their pre-
existing knowledge about the topic but also their beliefs about knowledge itself. Beliefs
about knowledge and knowing, namely epistemological beliefs, are individuals’ convic-
tions about the nature, organization, and source of knowledge, its truth value and justifi-
cation criteria of assertions (Hofer & Pintrich, 1997, 2002). Schommer (1990) proposed a
conceptualization of personal epistemology as a set of more or less independent belief
dimensions. Two dimensions regard the nature of knowledge: belief in simple (isolated and
unambiguous items of information) or complex (set of interrelated concepts) knowledge
and belief in certain (absolute and stable) or uncertain (continuously evolving) knowledge.
Two dimensions regard the nature of learning: belief in quick (fast process) or gradual
(slow process) learning and belief in a fixed (unmodifiable) or malleable (improvable)
ability to learn. The inclusion of beliefs about learning was questioned by researchers
(Hofer & Pintrich, 1997) who propose limiting personal epistemology to the nature of
knowledge only. However, they recognize that beliefs about knowledge are strictly con-
nected to beliefs about learning.
A systematic research program on the relationship between epistemological beliefs and
learning has been carried out by Schommer (e.g., 1990, 1993) through her 63-item
Epistemological Questionnaire. Among other findings, for instance, she showed that the
more students believe in certain knowledge and quick learning, the more they write inap-
propriate absolute conclusions to a text involving hypothetical material.
Of particular interest to the present study are investigations about the effects of students’
epistemological beliefs on conceptual change, which have all been conducted using revised
versions of Schommer’s instrument. In the study by Qian and Alvermann (1995), students
with less sophisticated beliefs were less likely to abandon their naïve conceptions of motion
after reading a text on the Newtonian theory, while students with more advanced beliefs
generated greater change in their conceptual structures. Windschitl and Andre (1998) inves-
tigated the relationship between epistemological beliefs and knowledge revision with
respect to an innovative, constructivist or a traditional, objectivist learning environment.
Students in the former produced significantly greater conceptual change regarding the topic
of the human cardiovascular system than students in the latter. In addition, students with
more sophisticated beliefs about the nature and acquisition of knowledge changed their own
conceptions more in the constructivist, rather than the traditional learning environment.
Mason (2000, 2001) also investigated the effects of beliefs about the nature of knowledge
in relation to the acceptance of anomalous data on theory change involving two controver-
sial topics, one scientific and the other historical. For the first topic in particular, students
who believed in the changing nature of knowledge were more likely to accept evidence con-
flicting with their prior conceptions and, consequently, to change them. In the study by
Southerland and Sinatra (2003), not only epistemological beliefs but also cognitive dispo-
sitions (e.g., the disposition to engage in effortful thinking, open-minded thinking and not
to identify with one’s beliefs but rather to weigh up new evidence) were examined in rela-
tion to the acceptance of three scientific topics varying in degree of controversy, that is,
human evolution, animal evolution, and photosynthesis. It was found that epistemological
beliefs and cognitive dispositions predicted acceptance of scientific knowledge only for the
Effects of Epistemological Beliefs and Learning Text Structure 167
most controversial topic, human evolution. Students with more sophisticated views about
knowledge and a greater disposition toward critical and open-minded thinking were more
likely to accept human evolution than students who have less advanced epistemological
beliefs and are less disposed toward critical and flexible thinking.
To give an account of the influence of personal epistemology on processes of concep-
tual change, Mason (2003) appealed to the key notion of intentionality in learning (Sinatra
& Pintrich, 2003b). She proposed that beliefs about knowledge and knowing may or may
not guide students toward the goal of learning through knowledge revision, starting from
recognizing a problem of knowledge, such as lack of knowledge or contradictory infor-
mation, in their conceptual structures. Only advanced epistemological beliefs, that is,
beliefs in complex, hypothetical, and evolving knowledge, are conducive to that recogni-
tion, which requires that students be engaged in learning as problem solving to be able to
intentionally produce changes in knowledge. The result of this process, sustained by a
sophisticated epistemological view, is a knowledge goal that implies the revision of prior
conceptions and higher levels of understanding.
Refutational Texts and Conceptual Change
A number of studies in reading education research have investigated the effects of text
characteristics, such as coherence and the relationship between coherence and prior knowl-
edge, on learning complex disciplinary content (e.g., Beck, McKeown, Sinatra, &
Loxterman, 1991; Boscolo & Mason, 2003; McNamara, Kintsch, Songer, & Kintsch,
1996). Of particular interest to the present study is the design of refutational texts, that is,
texts that directly state students’ alternative conceptions about a topic, refute them, and
present the scientific conceptions as viable alternatives (e.g., Alvermann & Hynd, 1989;
Hynd, 2003). The design of refutational texts to promote conceptual change has taken into
account Kintsch’s (1988) model of text comprehension, which distinguishes two levels of
comprehension: the textbase and situation model. The textbase is a propositional repre-
sentation of the information expressed in a text, which is formed by extracting semantic
information from a text at both local (microstructure) and global (macrostructure) levels,
and coding it into a coherent network of propositions. The situation model is the result of
integrating the reader’s prior knowledge with the information provided by the text.
Learning from text is more than forming a good textbase. It implies a situation model,
which is a mental representation of the text topic linked to the reader’s knowledge base.
Refutational texts have been proposed as powerful means of helping students construct
understanding at the situational model level. Not only do they activate the readers’ alterna-
tive prior knowledge but they also make explicit the inconsistencies between this knowledge
and the new knowledge to be learned. Refutational texts were found to be more effective
than traditional, straightforward informational texts in engendering conceptual change
about physics contents. This was especially so for less-skilled readers and also when vari-
ous other tools are used to facilitate knowledge revision (Alvermann & Hynd, 1989;
Alvermann, Hynd, & Quin, 1997; Guzzetti et al., 1993; Hynd, McWhorter, Phares, &
Suttles, 1994). The greater effectiveness of a refutational text, compared with a traditional
expository text about energy was also found in a study with elementary school students,
although it did not explicitly address pre- and post-reading conceptions (Diakidoy,
Kendeou, & Ioannides, 2003). Mikkilä-Erdman (2002) has also documented the advan-
tages, in terms of conceptual change, of a refutational text about photosynthesis over a
typical textbook text. Finally, it has emerged that a refutational text promotes greater change
in epistemological beliefs about mathematics in preservice teachers than a traditional text
(Gregoire-Gill, Ashton, & Algina, 2004).
Hynd (2003) has interpreted the benefit of refutational texts by reference to the four
conditions for conceptual change, highlighted in the well-known article by Posner and
associates (Posner, Strike, Hewson, & Gertzog, 1982). These conditions are satisfied by
refutational texts because they elicit dissatisfaction with the reader’s current conceptions,
explain the scientific concept clearly and, in depth, make it plausible through believable
examples, and, finally, show the usefulness of the explanatory value of the new concept.
The power of refutational texts has also been explained by reference to the characteristics
that anomalous data should have, as underlined by Chinn and Brewer (1993, 1998). They
should be credible, unambiguous, and in multiple forms. Furthermore, Hynd (2003)
referred to social psychology literature and explained the power of refutational texts also
in terms of central, not peripheral, text information processing because readers are actively
engaged in processing the text ideas they perceive as personally relevant.
Research Questions and Hypotheses
Relying on the findings from the two lines of research mentioned above, this study aimed
at examining how students’ epistemological beliefs, and the structure of a learning text,
affect revision of their alternative conceptions about natural selection and biological evo-
lution. The variable of students’ reading comprehension skill was also considered as
covariate in the statistical analyses. Although participants with reading difficulties were
not considered in the analysis, the reading skills of the other students, who made up regu-
lar heterogeneous classes, varied.
Three research questions guided the study:
1. Is students’ text-based comprehension at the immediate and delayed posttests affected
by epistemological beliefs, text structure, and/or their interaction?
2. Is students’ conceptual change about natural selection and biological evolution, as
reflected in changes in their explanations from pre to immediate and delayed tests,
affected by epistemological beliefs, text structure, and/or their interaction?
3. Is students’ metaconceptual awareness of the changes in their own conceptual struc-
tures at the immediate posttest affected by epistemological beliefs, text structure, and/
or their interaction?
For research question 1, we hypothesized that no significant differences would
emerge between participants for text comprehension at the textbase level in relation to
epistemological beliefs and text structure. According to the model of text comprehension
proposed by Kintsch (1988), retention-comprehension questions should ascertain partic-
ipants’ comprehension at the level of textbase, that is, the propositional representation of
the information expressed in the text. Since the refutational and traditional texts did not
differentiate in terms of information, participants would construct the same textbase rep-
resentation, that is, a coherent network of propositions, in the two conditions.
For research question 2, we hypothesized that beliefs about the nature of knowledge and
text structure would make a difference immediately after reading the text and at a delayed
posttest. Students with more advanced epistemological beliefs would change their alternative
conceptions more than students with less advanced epistemological beliefs, and these changes
would be more persistent. Conceptual change would imply text comprehension at the situa-
tion model level in Kintsch’s terms (1988), that is, the reader’s prior knowledge being inte-
grated with information given in the text. Participants with more advanced beliefs about the
nature of knowledge would be helped in perceiving a knowledge problem in their conceptual
structures and in working toward solving it. We hypothesized that the text structure would also
affect students’ revision of knowledge. By directly stating and challenging their alternative
conceptions, the refutational text, more than a traditional text, would create or refine students’
metaconceptual awareness about their own representations as well as the scientific represen-
tations. This is an essential condition for conceptual change (Vosniadou, 2003; Wiser & Amin,
2001). The former can be perceived as limited while the latter as having more explanatory
value. In addition, we wondered whether an interaction between the two variables would
emerge and, if so, whether the refutational text could contribute to compensating less mature
beliefs about the nature of knowledge in the change of explanations about biological evolu-
tion. Compared with a traditional text, a text that activates and challenges students’ alternative
conceptions could create awareness of a conflict between their own conceptions and scientific
ones in students who believe less in the complex and uncertain nature of knowledge.
For research question 3, we hypothesized that both epistemological beliefs and text
structure would affect metaconceptual awareness about changes in students’ conceptual
structures, as both factors would help students to be aware of the differences between their
prior conceptions and the scientific ones. In this case too we wondered whether the inter-
action between the two variables could also affect this performance by producing a com-
pensation effect between reading a refutational text and less advanced epistemological
beliefs. The refutational text may facilitate students who believe less in the complex and
uncertain nature of knowledge to be metaconceptually aware that their initial conceptions
changed as they were considered inappropriate after the text reading.
Method
Participants
First phase. In the first phase, the epistemological beliefs of 559 eighth graders, attending
public middle schools in a province of northeastern Italy, were measured. There were 282
girls and 277 boys. All were native speakers of Italian, Caucasian, and shared a homoge-
neous middle-class social background.
Second phase. In the second phase, 147 of the 559 students, who made up seven classes,
were involved in the study. Of these, 10 participants were found to have reading difficulties
(see below) on administration of a standardized reading comprehension text. In agreement
with their teachers, they took part in the study, but their production was not considered for
statistical analyses. In addition, 12 students were not present at the posttest. There were
therefore 125 participants who provided data for both the pre and immediate posttests
(65 girls and 60 boys). Of these 125 students, 15 were absent on the day of the delayed
posttest. Therefore, 110 participants (58 girls and 52 boys) underwent the pre, post, and
delayed posttest design. Participants’ scores in the epistemological dimension of Certain
and simple knowledge (see the Results section) were dichotomized on the basis of the
median and were used to create two mutually exclusive groups: one made up of students
with more advanced beliefs about the nature of knowledge (n ! 48; 25 in the experimental
condition and 23 in the control condition) and the other of students with less advanced
beliefs (n ! 62; 29 in the experimental condition and 33 in the control condition).
Conditions
Each class was randomly assigned to one of two conditions: experimental and control.
Students in the experimental condition read a refutational text about natural selection and
biological evolution. In the control condition, the students read a traditional text about the
topic (see next section).
Prereading Tasks
Epistemological beliefs. A shorter version of Schommer’s Epistemological Questionnaire

was administered to measure students’ beliefs about the nature and acquisition of knowl-
edge. This is a reduced and adapted version in Italian, which had already been used in a
previous study (Lando, 2003). It comprises 36 items to be rated on a 5-point Likert-type
scale (1 ! totally disagree; 5 ! totally agree). The items refer to the four epistemological
dimensions proposed by Schommer (1990). Nine items regard each of the four dimen-
sions: simplicity vs. complexity of knowledge, certainty vs. uncertainty of knowledge,
fixed vs. malleable ability to learn, and quick vs. gradual learning.
Prior knowledge. Participants’ preexisting conceptions about natural selection and bio-
logical evolution were ascertained by means of five open-ended and four multiple-choice
questions. All questions were of a “generative” nature (Vosniadou, 1994), which intro-
duced a problem to be solved by the construction of an appropriate representation of nat-
ural selection. The open-ended questions asked students to formulate an explanation of the
problem, while the multiple-choice questions asked them to choose one of the four expla-
nations. Some of these questions were devised by taking into account questions asked in
previous studies (e.g., Bandiera, 1991; Bishop & Anderson, 1990; Demastes-Southerland,
Good, & Peebles, 1995; Lawson & Thompson, 1988; Lawson & Weser, 1990) and the
grade level of our participants, who were younger than those involved in previous research.
Other questions, formulated by the authors, were new. The reliability of open-ended and
multiple-choice questions at pretest were .68 and .76, respectively. For each participant a
composite score for the explanations was calculated by summing the scores for answers to
the open-ended questions and multiple-choice questions.
Reading comprehension. Participants were individually administered an Italian stan-
dardized test for eighth grade to measure their expository text reading comprehension
skills (Cornoldi & Colpo, 1995).
Reading Text
The effects of learning text structure on conceptual change were examined by means of two
versions of the same text about natural selection and biological evolution according to
Darwin’s theory. In the control condition students were given a text taken from a common
science textbook (Braccini, Bosco, & Durante, 1998). In the experimental condition,
students were given a refutational text, which was prepared by modifying the structure of
the traditional text used in the control condition. Four parts were added and no other mod-
ifications were made. Each of these four parts was written to activate, by directly stating and
challenging, a specific alternative conception held by the participants, which had already
been identified in previous studies on the topic (e.g., Bandiera, 1991; Bishop & Anderson,
1990; Demastes-Southerland et al., 1995; Ferrari, & Chi, 1998; Lawson & Thompson,
1988). Each part added to the refutational text was also written in keeping with the con-
ceptual continuity and text flow (Beck et al., 1991). Text coherence was modified as little
as possible with the insertion of each part. The first part added considered the question of
the origin of a new species and aimed at refuting the creationist perspective. The second
underlined the importance of random intraspecies variability and aimed at undermining the
alternative conception that all exemplars of a single species are the same. The third exam-
ined the concept of “struggle for survival” underlining that it should not be considered in
physical terms. The fourth part added challenged the naïve conception of the individual’s
adaptation to the environment based on the need and/or will to change. This alternative
conception is confuted to introduce the theory of adaptation of the species to the environ-
ment. In Appendix an excerpt from the traditional text and another from the refutational are
presented.
Postreading Tasks
Text-based comprehension. Six open questions were asked to assess participants’ com-
prehension at the level of textbase. The answers to these questions required information
that was explicitly stated in the text.
Conceptual change as change in explanations. The same generative open-ended and

multiple-choice questions used to ascertain participants’ prior knowledge were asked again
at both the immediate and the delayed posttests to examine any changes in participants’
explanations and the stability of these changes. Answers to these questions should ascer-
tain text comprehension at the level of situation model (Kintsch, 1988), based on the inte-
gration between prior knowledge and knowledge provided by the text. At the immediate
posttest, reliability of open-ended and multiple-choice generative questions were .78 and
.82, respectively.
Metaconceptual awareness. Two questions were asked to examine participants’ meta-

conceptual awareness of the changes in their conceptual structures. They were first asked,
“Do you think that you have changed any conception after reading the text?” If they
answered positively, they were asked, “What was your conception before reading and what
is it now?”
Coding
Answers to the open-ended questions, for measuring comprehension at textbase level,

were scored 0–2 according to their degree of correctness and completeness (Table 13.4).
All answers were coded by two independent judges. Inter-rater agreement was
93%. Disagreements were resolved in conference through discussion in the presence of a
third judge.
Answers to the generative open-ended questions asked at pretest and immediate and
delayed posttests to measure knowledge revision were scored 0–3 according to their
degree of correctness and completeness (Tables 13.2 and 13.5). All answers were coded by
the same two independent judges. Inter-rater agreement was 94%. Disagreements were
resolved in conference through discussion in the presence of a third judge. Answers to the
multiple-choice questions asked at pretest and immediate and delayed posttests to meas-
ure knowledge revision were scored 0–2. Of the four choices, two were wrong, one cor-
rect but generic, and one correct and complete (Table 13.3).
Answers to the questions about metaconceptual awareness were assigned 1 point for
mentioning each of the referred concepts. For instance, the answer “I believed that the
species were always the same since the beginning, now I know that they evolved” scored
1. The answer “Before I thought like Lamarcke, that is, species acquire new features
through the use of their organs and then these features are transmitted. Now I think that
speciation is based on a mutation” scored 2. All answers were coded by the same two inde-
pendent judges. Inter-rater agreement was 98%. Disagreements were resolved in confer-
ence through discussion in the presence of a third judge.
Table 13.1: Items loading on the epistemological belief dimension “Certain and simple
Knowledge.”
Item Loading
Almost everything one reads is believable .488

If scientists try hard enough, they can find the truth to almost everything .548
The most important aspect of scientific work is precise measurement and
careful work .677
Scientists can ultimately arrive at the truth .603
Most words have one clear meaning .701
Truth is unchanging .567
It is a waste of time to work on problems that have no chance of arriving
at a clear-cut and unambiguous answer .424
The best thing about a science course is that most problems have only
one right answer .501
Today’s facts may be tomorrow’s fiction .432
If you try to combine new ideas from a textbook with what you already
know, you will get mixed up .457
Table 13.2: Frequency/percentage of responses to the open-ended questions at pretest by condition and epistemological beliefs.
Questions Response categories Condition/epistemological beliefs
Experimental Control
Less More Less More
advanced advanced advanced advanced
Effects of Epistemological Beliefs and Learning Text Structure

Q1. If a man has developed Not pertinent/unclear (0)a 2 (8%) 1 (3%) 3 (9%) 1 (4%)
strong muscles because of Yes, strong muscles are transmitted (0) 4 (16%) 5 (17%) 7 (21%) 5 (22%)
having done a lot of physical No, the man changed over time (1) 6 (24%) 3 (10%) 6 (18%) 4 (17%)
exercise, will his children be No, the man was not born with strong 8 (32%) 13 (45%) 13 (39%) 9 (39%)
born with stronger muscles? muscles and cannot transmit them (2)
No, strong muscles are not part of genetic 5 (20%) 7 (24%) 4 (12%) 4 (17%)
inheritance and cannot be transmitted (3)
Q2. The first cheetahs used to Not pertinent/unclear (0) 4 (16%) 3 (10%) 6 (18%) 2 (9%)
run at a maximum speed of Running increases running capacity (0) 12 (48%) 16 (55%) 20 (61%) 12 (52%)
only 20 km/h. Nowadays Need or will to run faster increases 7 (28%) 8 (28%) 6 (18%) 9 (39%)
cheetahs run 60 km/h when running capacity (0)
they are hunting prey. In It depends on the struggle for 2 (8%) 1 (3%) – –
your opinion, how can the survival (1)
evolution of their running It depends on the struggle for survival: – – 1 (3%) –
capacity be explained? The fastest cheetahs were favored (2)
Natural selection favored survival of the – 1 (3%) – –
most adaptable, those that were fast, the
winners in the struggle for survival (3)
Q3. In UK there is a species Not pertinent/unclear (0) 6 (24%) 6 (21%) 5 (15%) 4 (17%)
of butterfly called ‘Biston They will become dark as they 12 (48%) 7 (59%) 21 (64%) 16 (70%)
Betularia’. There are two need to be dark (0)
types of this butterfly. One Dark butterflies will survive (1) 5 (20%) 5 (17%) 4 (12%) 3 (13%)
type has light-colored wings, There will be more dark butterflies 2 (8%) 1 (3%) 2 (6%) –
the other dark-colored wings. than the light colored as
Most butterflies have light- they reproduce (2)
colored wings and land on Dark butterflies can camouflage – – 1 (3%) –
birch trees, which have light- themselves, so they adapt and
173
(Continued)
Table 13.2: (Continued)
174
Questions Response categories Condition/epistemological beliefs

Less More Less More
colored bark. In this way reproduce, while the light colored
are camouflaged and are not are destined to become extinct (3)
they eaten by birds. What
would happen in the long run
to the butterflies’ wings if
the bark of the birch became
darker because of pollution?
Q4. The arctic fox lives best at Not pertinent/unclear (0) 1 (4%) 3 (10%) 5 (15%) 2 (9%)
very low temperatures. It has Arctic foxes were created 11 (44%) 0 (34%) 14 (42%) 11 (48%)
a thick coat, which, of course, with a thick coat (0)
is very important for survival. Arctic foxes needed to have a 12 (48%) 15 (52%) 14 (42%) 10 (43%)
What do you think is the o thicker coat because of the
rigin of this thick coat? low temperature (0)
A mutation was the cause of 1 (4%) 1 (3%) – –
the thick coat (1)
Q5. Bacteria that infest the Not pertinent/unclear (0) 18 (72%) 16 (55%) 24 (73%) 13 (57%)
human body often cause No, bacteria will get used to – 4 (14%) 4 (12%) 3 (13%)
disease. Antibiotics are used antibiotics (0)
to kill them. However, some No, bacteria will build on their 5 (20%) 7 (27%) 4 (12%) 5 (22%)
of these bacteria manage to own resistance to survive (0)
resist antibiotics. In your No, bacteria are resistant (1) 2 (8%) – – 2 (9%)
opinion, in 20 years, for No, fewer and fewer bacteria will – 1 (3%) – –
instance, will current be killed by antibiotics, the others
antibiotics still be effective will become resistant (2)
against bacteria? No, if they are resistant, they – 1 (3%) 1 (3%) –
reproduce and the new bacteria
will also be resistant (3)
aScoring (0) means that no points were attributed to the answer, (1) indicates a correct but incomplete answer, (2) indicates a correct and more elaborated answer,
and (3) indicates a correct and complete answer.
Procedure
Data gathering took place in three sessions. The topic of natural selection and biological
evolution had not yet been dealt with in any of the classes. In the first session (pretest)
the order of tasks was: Epistemological Questionnaire, pretest generative open-ended and
multiple-choice questions, and reading comprehension test. The first session took about
2 hours. The second session took place about a week after the first. Participants in the
control condition were asked to study the traditional text, while participants in the exper-
imental condition were asked to study the refutational text. After reading the text (imme-
diate posttest), participants were asked to answer open-ended and multiple-choice
generative questions, and metaconceptual questions based on the text. This session also
took about 2 hours. The third session (delayed posttest) took place 2 months after the
posttest and lasted about 1.5 hours. Participants were asked all the same questions again.
Results
The initial 559 participants’ item ratings for each of the 36 items of the reduced and
adapted version of Schommer’s Epistemological Questionnaire were used as a variable in
the factor analysis. To be consistent with Schommer (1990), a principal-axis factoring with
varimax rotation was used for an exploratory factor analysis. A three-factor model
emerged with all eigenvalues "1.00 as cut-off. The three factors accounted for 45.3% of
the variance after rotation and were given descriptive titles on the basis of high-loading
items. The three factors are: “Externally controlled learning” (" = .90), “Quick learning”
(" = .84), and “Certain and simple knowledge” (" = .82). Weighted factor scores were cal-
culated for each belief dimension. Only scores for Certain and simple knowledge, that is,
the factor which was most relevant to the present study, were used for subsequent analy-
ses. It should be pointed out that also in Qian and Alvermann’s (1995) study using
Schommer’s revised 32-item questionnaire, the dimension “Simple and certain knowl-
edge” emerged. Table 13.1 shows the items that loaded ".40 in the belief dimension of
Certain and simple knowledge.
Reading Comprehension Skills
According to the scores in the standardized test of reading comprehension skills, each par-
ticipant could be considered as having high-, intermediate-, or low-level skills.
Prior Knowledge
Answers to the open-ended and multiple-choice questions revealed that the typical alterna-
tive conceptions, already identified in previous studies, were common among participants
(Tables 13.2 and 13.3). Overall, it can be said that students held naïve conceptions of an
organism’s adaptation to the environment when they explained the biological evolution
176
Table 13.3: Frequency/percentage of responses to the multiple-choice questions at pre, immediate, and delayed posttests by
condition and epistemological beliefs.
Questions Response Condition/epistemological beliefs

categories
Pretest Immediate posttest Delayed Posttest
Experimental Control Experimental Control Experimental Control

Less More Less More Less More Less More Less More Less More
advanced advanced advanced advanced advanced advanced advanced advanced advanced advanced advanced advanced
Q1
A lady with brown Incorrect – 1 (3%) 1 (3%) 1 (4%) – 4 (14%) 1 (3%) 3 (13%) 4 (16%) – 5 (15%) 1 (4%)
hair, married to a (choice c
man who also has and d) (0)a
brown hair, has Correct, 8 4 9 6 5 4 11 10 21 5 27 7
been dyeing her generic (32%) (14%) (27%) (26%) (20%) (14%) (33%) (43%) (84%) (17%) (82%) (30%)
hair blond since (choice a)
she was 18 years (1)
old. When she has Correct and 17 24 23 16 20 21 21 10 – 24 1 15
a child, will this complete (68%) (83%) (70%) (70%) (80%) (72%) (64%) (43%) (83%) (3%) (65%)
child have brown or (choice b)
blond hair? Why? (2)
Q2
Many years ago Incorrect 19 18 21 14 11 5 18 12 14 10 20 13
when insecticides (choice a (76%) (62%) (64%) (61%) (44%) (17%) (55%) (52%) (56%) (34%) (61%) (57%)
started to be used, and c) (0)
they were very Correct, 3 4 7 6 5 4 4 3 2 – 2 2
effective for generic (12%) (14%) (21%) (26%) (20%) (14%) (12%) (13%) (8%) (6%) (9%)
killing flies and (choice d)
mosquitoes. Now, (1)

after 40 years, Correct and 3 7 5 3 9 20 11 8 9 19 11 8
when the same complete (12%) (24%) (15%) (13%) (36%) (69%) (33%) (35%) (36%) (66%) (33%) (35%)
insecticides are (choice b)
sprayed, much (2)
fewer mosquitoes
are killed. How can
you explain this
phenomenon?
Q3
Why did geese Incorrect 21 24 29 18 15 8 24 14 17 12 25 16
develop palmate (choice b (84%) (83%) (88%) (78%) (60%) (28%) (73%) (61%) (68%) (41%) (76%) (70%)
feet? and c) (0)
Correct, 1 (4%) 3 (10%) – – 3 (12%) 4 (14%) 1 (3%) 1 (4%) 2 (8%) 5 (17%) 2 (6%) –
generic
(choice d)
(1)
Correct and 3 2 4 5 7 17 8 8 6 12 6 7
complete (12%) (7%) (12%) (22%) (28%) (59%) (24%) (35%) (24%) (41%) (18%) (30%)
(choice a)
(2)
(Continued)
177
178

categories
Pretest Immediate posttest Delayed Posttest
Experimental Control Experimental Control Experimental Control

Less More Less More Less More Less More Less More Less More
advanced advanced advanced advanced advanced advanced advanced advanced advanced advanced advanced advanced
Q4
If a population of Incorrect 17 25 27 14 18 14 23 14 18 17 22 15
geese were forced (choice a (68%) (86%) (82%) (61%) (72%) (48%) (70%) (61%) (72%) (59%) (67%) (65%)
to live in an and b) (0)
environment Correct, 2 2 2 1 – 2 4 1 1 5 3 –
where water to generic (8%) (7%) (6%) (4%) (7%) (12%) (4%) (4%) (17%) (9%)
swim in was not (choice c)
available, what (1)
would happen? Correct 6 2 4 8 7 13 6 8 6 7 8 8
and (24%) (7%) (12%) (35%) (28%) (45%) (18%) (35%) (24%) (24%) (24%) (35%)
complete
(choice d)
(2)
aScoring (0) was attributed to the wrong choice, (1) was attributed to the correct but generic choice, and (2) was attributed to the correct choice.
mentioned in the questions. According to these conceptions, adaptation is due to the need
or will to survive, or to the use or exercise of a feature. The creationist conception was
detected only in the explanations regarding the question of the coat of the artic fox, which
asked explicitly about its origin.
Overall, answers to the multiple-choice generative questions at pretest substantially
confirmed the presence of naïve conceptions about adaptation to the environment.
Preliminary analyses through independent-sample t-tests revealed that the students in
the two conditions, experimental and control, did not differ significantly for prior know-
ledge, t (123) ! .98, p ! .32, reading comprehension skills, t (123) ! .61, p ! .54 or epis-
temological beliefs, t (123) = .99, p = .32.
Text-Based Comprehension
A repeated-measure ANCOVA, with epistemological beliefs (more and less advanced) and
condition (experimental, control) as between-subject variables, time (immediate posttest
and delayed posttest) as the within-subject variable, and reading comprehension as the
covariate, was carried out. It revealed the main effect of testing time, F (1, 105) ! 11.83,
p = .001, as well as the interaction time # condition, F (1, 105) ! 4.70, p $ .05. Two
months after studying the text, students’ overall scores decreased, but the scores of partic-
ipants who had read the refutational text decreased significantly less.
In this case, the text structure affected the stability of comprehension at the textbase
level (Table 13.4). No differences related to epistemological beliefs emerged. Figure
13.1 shows mean scores for text-based comprehension at the immediate and delayed
posttests.
Immediate posttest
Delayed posttest
9
8
7
6
5
4
3
2
1
0
Less More Less More
Figure 13.1: Adjusted mean scores of text-based comprehension by condition

and epistemological beliefs.
Table 13.4: Frequency/percentage of responses to the text-based comprehension questions at immediate and delayed posttests by con-
180
dition and epistemological beliefs.

categories
Immediate posttest Delayed posttest
Experimental Control Experimental Control

Less More Less More Less More Less More
advanced advanced advanced advanced advanced advanced advanced advanced
Q1. What should The parents (0)a 18 (72%) 15 (52%) 19 (58%) 16 (70%) 23 (92%) 25 (86%) 25 (76%) 18 (78%)
the variability Chance mutations 4 (16%) 2 (7%) 4 (12%) 2 (9%) 2 (8%) 4 (14%) 8 (24%) 5 (22%)
of individual or “mixing” of
characteristics parents’
be attributed characteristics (1)
to? Chance 3 (12%) 12 (41%) 10 (30%) 5 (22%) – – – –
mutations and
“mixing” of
parents’
characteristics (2)
Q2. What does Fight between 9 (36%) 4 (14%) 9 (27%) 8 (35%) 19 (76%) 19 (66%) 27 (82%) 16 (70%)
“struggle for animals (0)
survival” Competition 10 (40%) 12 (41%) 8 (24%) 4 (17%) 6 (24%) 10 (34%) 5 (15%) 5 (22%)
mean? for food (1)
Competition 6 (24%) 13 (45%) 16 (48%) 11 (48%) – – 1 (3%) 2 (9%)
for food and
territory, to
avoid predators,
to overcome
diseases (2)
Q3. Which The strongest 9 (36%) 4 (14%) 12 (36%) 4 (17%) 16 (64%) 12 (41%) 32 (97%) 18 (78%)
individuals survive and
are more the weakest
advantaged die (0)
and which are Animals 7 (28%) 11 (38%) 7 (21) 3 (13%) 8 (32%) 7 (24%) – 3 (13%)
destined to die with good
in the struggle characteristics
for survival? for the environment

survive, the
others die (1)
Animals with 9 (36%) 14 (48%) 14 (42%) 16 (70%) 1 (4%) 10 (34%) 1 (3%) 2 (9%)
the characteristics
fitting the
environment
survive and have
offspring, the
others do not live
long (2)
Q4. What favors The strongest 6 (24%) 3 (10%) 6 (18%) 5 (22%) 19 (76%) 17 (59%) 32 (97%) 23 (100%)
natural or the
selection? most clever
animals (0)
Survival of the 3 (12%) 2 (7%) 2 (6%) 1 (4%) 1 (4%) 5 (17%) – –
animals that can
adapt to the
environment (1)
Survival of the 16 (64%) 24 (83%) 25 (76%) 17 (74%) 5 (20%) 7 (24%) 1 (3%) –
fittest who
transmit their
characteristics to
offspring (2)
181
(Continued)
182

categories

Q5. What are Changes in 2 (8%) 4 (14%) 8 (24%) 2 (9%) 6 (24%) 10 (34%) 16 (48%) 10 (43%)
mutations? animals’
appearance (0)
Variation in 6 (24%) 2 (7%) 1 (3%) 5 (22%) 10 (40%) 12 (41%) 12 (36%) 8 (35%)
genetic
inheritance (1)
Positive or 17 (68%) 23 (79%) 24 (73%) 16 (70%) 9 (36%) 7 (24%) 5 (15%) 5 (22%)
negative chance
variations in
genetic
inheritance (2)
Q6. What is Reproduction (0) 2 (8%) 2 (7%) 1 (3%) 2 (9%) 18 (72%) 19 (66%) 29 (88%) 19 (82%)
speciation? Origin of a 15 (60%) 7 (24%) 19 (58%) 9 (39%) 6 (24%) 8 (28%) 2 (6%) 2 (9%)
new species (1)
Origin of a new 8 (32%) 20 (69%) 13 (39%) 12 (52%) 1 (4%) 2 (7%) 2 (6%) 2 (9%)
species when
mutations
accumulate over
generations (2)
Conceptual Change
A repeated-measure ANCOVA, with epistemological beliefs (less and more advanced) and
condition (experimental, control) as between-subject variables, conceptual knowledge at the
three different times (composite scores for explanations at pretest, immediate, and delayed
posttests) as the within-subject variable, and reading comprehension as the covariate, was
carried out. From this analysis the main effects of testing time, F (2, 104) = 3.16, p <.05
and the interactions time # epistemological beliefs, F (2, 104) = 3.75, p < .05, time # con-
dition, F (2, 104) ! 17.50, p < .001, and time # epistemological beliefs # condition,
F (2, 104) ! 5.72, p ! .01, emerged. The covariate also correlated significantly with the
three scores, F (2, 104) ! 9.93, p ! .001. Overall, participants’ scientific knowledge about
natural selection and biological evolution increased from the pre to the immediate posttest.
However, participants’ scores decreased slightly from the immediate to the delayed posttest.
Regarding epistemological beliefs, students with more advanced convictions about the
nature of knowledge progressed conceptually from the pre to the immediate posttest much
more than those with less advanced convictions. As far as the type of text is concerned, stu-
dents who received information from the refutational text were facilitated in conceptual
change much more than students who read the traditional text (Tables 13.3 and 13.5).
From the interaction testing time # beliefs # condition, it emerged that participants
who changed their alternative prior conceptions more at the immediate posttest were those
with more advanced epistemological convictions and who read the refutational text.
Moreover, students with less advanced epistemological beliefs who read the refutational
text outscored those at the same level of epistemic cognition who read the traditional text.
Furthermore, students with the highest outcomes at the immediate posttest were also
those whose scores decreased more at the delayed posttest. However, they still signifi-
cantly outscored those who progressed less at the immediate posttest and whose scores
did not significantly decrease at the delayed posttest. Figure 13.2 shows means of com-
posite scores for generative questions at pre, immediate, and delayed posttests. It is worth
noting that the covariate prior knowledge correlates with conceptual change measures.
Conceptual Change in Explanations: Examples
To illustrate cases of conceptual change due to text reading, some examples of expla-
nations that show a revision of personal conceptions from pre to immediate posttest fol-
low. For the question about cheetahs’ increased running speed, the pretest explanation
given by a student in the experimental condition (signed as P6) expressed a naïve pre-
conception about adaptation, which was then successfully revised. The explanation
given at the delayed posttest referred to the crucial ideas of intraspecies variation and
natural selection.
Pretest:
The ability to run faster developed over time, over some decades in fact, so
muscles and bones changed allowing cheetahs to increase their speed.
Cheetahs’ running faster and faster influenced the speed. (P6, exp.)
184
Table 13.5: Frequency/percentage of responses to the open-ended questions at immediate and delayed posttests by condition and
epistemological beliefs.

categories

Q1
If a man has Not pertinent/ 1 (4%) 1 (3%) 5 (15%) 3 (13%) 3 (12%) 2 (7%) 4 (12%) 4 (17%)
developed strong unclear(0)a
muscles because of Yes, strong muscles 3 (12%) 2 (7%) 8 (24%) 9 (39%) 3 (12%) 3 (10%) 5 (15%) 6 (26%)
having done a lot are transmitted (0)
of physical No, the man changed 2 (8%) 2 (7%) 7 (21%) 3 (13%) 2 (8%) 4 (14%) 8 (24%) 2 (9%)
exercise, will his over time (1)
children be born No, the man was 8 (32%) 5 (17%) 9 (27%) 6 (26%) 7 (28%) 7 (24%) 11 (33%) 9 (39%)
with stronger not born with
muscles? strong muscles and
cannot transmit
them (2)
No, strong muscles 11 (44%) 19 (66%) 4 (12%) 2 (9%) 10 (40%) 13 (45%) 5 (15%) 2 (9%)
are not part of
genetic inheritance
and cannot be
transmitted (3)
Q2
The first cheetahs Not pertinent/
used to run at a unclear (0) 1 (4%) – 1 (3%) – 1 (4%) – 2 (6%) 1 (4%)
maximum speed Running increases 7 (28%) 5 (17%) 11 (33%) 8 (35%) 8 (32%) 2 (7%) 8 (24%) 6 (26%)
of only 20 km/h. running capacity (0)
Nowadays cheetahs Need or will to run 9 (36%) 7 (24%) 14 (42%) 12 (52%) 10 (40%) 12 (41%) 15 (45%) 14 (61%)
run 60 km/h when faster increases
they are hunting running capacity (0)

prey. In your It depends on the 4 (16%) 5 (17%) 4 (12%) 1 (4%) 2 (8%) 4 (14%) 4 (12%) 2 (9%)
opinion, how can struggle for
the evolution of survival (1)
their running It depends on 2 (8%) 1 (3%) 2 (6%) – 1 (4%) 3 (10%) 1 (3%) –
capacity be the struggle for
explained? survival: The fastest
cheetahs were
favored (2)
Natural selection 2 (8%) 11 (38%) 1 (3%) 2 (9%) 3 (12%) 8 (28%) 3 (9%) –
favored survival
of the most
adaptable, those
that were fast,
the winners in
the struggle for
survival (3)
Q3
In UK there is a Not pertinent/ 5 (20%) 3 (10%) 11 (33%) 4 (17%) 7 (28%) 5 (17%) 11 (33%) 5 (22%)
species of butterfly unclear (0)
called ‘Biston They will become 12 (48%) 10 (34%) 18 (55%) 13 (57%) 8 (32%) 7 (24%) 13 (39%) 12 (52%)
Betularia.’ dark as they need
to be dark (0)
(Continued)
185
186

categories

There are two types Dark butterflies 5 (20%) 3 (10%) 3 (9%) 4 (17%) 2 (8%) 8 (28%) 5 (15%) 5 (22%)
of this butterfly. will survive (1)
One type has light- There will be 1 (4%) 3 (10%) 1 (3%) 1 (4%) 6 (24%) 6 (21%) 4 (12%) –
colored wings, the more dark
other dark colored butterflies than
wings. Most as they reproduce (2)
butterflies reproduce, Dark butterflies can 2 (8%) 10 (34%) – 1 (4%) 2 (8%) 3 (10%) – 1 (4%)
while have light- camouflage
colored wings and themselves,so
land on birch trees, they adapt and
which have light- the light colored
colored bark. In are destined to
this way they are become extinct (3)
camouflaged and are
not eaten by birds.
What would happen
in the long run to the
butterflies’ wings if
the bark of the
birch became darker
because of pollution?
Q4
The arctic fox Not pertinent/ 2 (8%) – 1 (3%) 1 (4%) 1 (4%) 1 (3%) 2 (6%) –
lives best at very unclear (0)
low temperatures. Arctic foxes were 12 (48%) 9 (31%) 20 (61%) 14 (61%) 15 (60%) 13 (45%) 16 (48%) 16 (70%)
It has a thick coat, created with a
which, of course, thick coat (0)
is very important Arctic foxes 8 (32%) 3 (10%) 8 (24%) 7 (30%) 8 (32%) 3 (10%) 10 (30%) 7 (30%)
for survival. needed to have

What do you think a thicker coat
is the origin of this because of the
thick coat? low temperature (0)
A mutation was the 2 (8%) 7 (24%) 1 (3%) 1 (40%) – 1 (3%) 2 (6%) –
cause of the thick
coat (1)
A mutation favored – 1 (3%) – – – 5 (17%) 1 (3%) –
the adaptation
of foxes with a
thick coat (2)
By chance mutation 1 (4%) 9 (31%) 3 (9%) – 1 (4%) 6 (21%) 2 (6%) –
foxes with thick
coats adapted,
survived and
reproduced (3)
Q5
Bacteria that infest Not pertinent/ 14 (56%) 15 (52%) 19 (58%) 14 (61%) 13 (52%) 11 (38%) 17 (52%) 12 (52%)
the human body unclear (0)
often cause disease. No, bacteria will
Antibiotics are used get used to 5 (20%) 6 (21%) 9 (27%) 6 (26%) – – – –
to kill them. antibiotics (0)
However, some of No, bacteria will 2 (8%) 1 (3%) 2 (6%) 3 (13%) 8 (32%) 8 (28%) 13 (39%) 8 (35%)
these bacteria build on their
manage to resist own resistance
187
antibiotics. In your to survive (0)
(Continued)
188

categories

opinion, in 20 years, No, bacteria 2 (8%) – 2 (6%) – 1 (4%) 2 (7%) 1 (3%) 3 (13%)
for instance, are resistant (1)
will current No, fewer and fewer 2 (8%) – 2 (6%) – 3 (12%) 1 (3%) ^1 (3%) –
antibiotics still be bacteria will be
effective against killed by
bacteria? antibiotics,
the others will
become
resistant (2)
No, if they are 2 (8%) 7 (24%) 1 (3%) – – 7 (24%) 1 (3%) –
resistant, they
reproduce and
the new bacteria
will also be
resistant (3)
Pretest
Immediate posttest
Delayed posttest
14
12
10
0
Less More Less More
Figure 13.2: Adjusted mean of composite scores for conceptual change by condition and
epistemological beliefs.
Delayed posttest:
The first cheetahs had different characteristics. Some ran faster than oth-
ers, thus this advantageous characteristic allowed faster cheetahs to pre-
vail over the slower ones, and over time the slow cheetahs became extinct.
(P6, exp.)
For the question about the origin of the thick coat of the arctic fox, a change in expla-
nation from pre to the posttest is evident in the following answers given by another student
in the experimental condition (signed as P92). Her naïve preconception was based on the
idea that the need for survival leads to adaptation to the environment. The new conception
she constructed appealed to the notions of intraspecies variation and natural selection. She
understood that the latter favors only the species members with the most suitable charac-
teristics.
Pretest:
I think that the arctic fox, in the beginning, was a fox with a normal coat
but over time its coat changed because of the very low temperatures. It
therefore became thicker and thicker, allowing it to survive. (P92, exp.)
Delayed posttest:
These foxes with a thicker coat are the descendants of the first foxes that
inhabited the earth. They were actually “privileged” because they could
adapt to living at low temperatures, while at the same time there were foxes
with less thick coats. Foxes with a less thick coat disappeared, while the
“privileged” survived and had offspring that had offspring. Thus, after mil-
lions of years the entire species has adapted to the environment. (P92, exp.)
Metaconceptual Awareness
An ANCOVA with epistemological beliefs and condition as the between-subject factors,

reading comprehension as the covariate, and the immediate posttest score for metacon-
ceptual awareness as the outcome measure revealed the effects of both independent vari-
ables, F (1, 105) ! 5.58, p $ .05, and F (1, 105) ! 8.89, p $ .01, respectively. Students
with more advanced beliefs about the nature of knowledge (M ! .69, SE ! .10) showed
greater metaconceptual awareness of the changes in their conceptual structures after text
reading than students with less advanced beliefs (M ! .34, SE ! .10). Participants who
read the refutational text (M ! .73, SE ! .10) were also metaconceptually more aware
than students who read the traditional text (M ! .29, SE ! .10). No interaction between
the two variables emerged. The following are examples of the different degrees of parti-
cipant awareness of the change in their personal conceptions, having read the text, as
expressed at the immediate posttest:
Before, I thought that cheetahs had developed a skeleton suitable to the

environment and that foxes had developed their coats to adapt to the envi-
ronment. Now I know that it was only possible for members of the species
to survive if they had a suitable skeleton or coat. (P8, exp.)
Before, I thought animals changed because they tried to adapt to the envi-
ronment so that they would not die. I have learned now that their charac-
teristics changed because of natural selection not because of wanting to
adapt to the environment. Through natural selection only those who have
some advantages for survival in fact survive. The example of giraffes
in the text led me to think a lot about the evolution of some animals.
(P58, exp.)
These examples illustrate the students’ metacognitive ability to think about their past
and current thinking, although at increasing levels of awareness.
Discussion and Implications
This study aimed at extending current understanding of the role of two variables that
affect conceptual change, which had been investigated separately, that is, epistemologi-
cal beliefs and the structure of the text to be read. The former is a learner’s variable that
reflects a more or less advanced position about the nature of knowledge. In this study, it
reflects convictions about knowledge as simple and certain vs. complex and continu-
ously evolving. The latter variable is a factor regarding instructional material that
reflects the type of text from which new information can be constructed. In this study
two types of text were used: an ordinary expository science text whose primary function
was to give new information, and a refutational text that not only gives new, correct
information but also explicitly states and refutes alternative conceptions by presenting
the scientific conceptions as viable alternatives.
Our first research question asked if participants’ comprehension at the textbase level
would be affected by epistemological beliefs, text structure, and/or their interaction.
Findings partially confirm our hypothesis, since epistemological beliefs did not affect this
level of text comprehension but text structure did. Students’ scores decreased at the
delayed posttest and the decrease was less evident for students in the experimental condi-
tion, who read the refutational text.
The second research question asked if students’ conceptual change about the topic —
reflected in explanations given in answer to open-ended questions and in the choice of
explanations for the multiple-choice questions — would be affected by epistemological
beliefs and text structure. We also wondered if reading a refutational text could compen-
sate less advanced epistemological understanding. Findings partially confirm our hypoth-
esis and show that, overall, students’ scientific knowledge of the topic improved, but those
who believed more in complex and uncertain knowledge generated more conceptual
change than those who believed more in simple and certain knowledge. Similarly, students
who read the refutational text changed their preconceptions more than students who read
the traditional text.
Our data on the effects of beliefs about knowledge and text structure confirm the find-
ings of previous studies on these two variables, which were investigated separately.
On the one hand, all studies that addressed the relationship between epistemological
beliefs (measured by the revised version of Schommer’s questionnaire) and conceptual
change found that less advanced epistemological beliefs, that is belief in knowledge as
absolute, simple, stable, and transmitted by authority, are associated with lower perform-
ances in knowledge revision. Conversely, more sophisticated beliefs, that is belief in
knowledge as complex, uncertain, and derived from reason, are associated with higher
performances (e.g., Mason, 2000; Qian & Alvermann, 1995; Sinatra, Southerland,
McConaughy, & Demastes, 2003, Windschitl & Andre, 1998). As previously proposed
(Mason, 2003), epistemological beliefs mediate cognitive functions in the process of inten-
tional conceptual change (Sinatra & Pintrich, 2003a). They may or may not guide students
toward the goal of learning through knowledge revision, starting from recognizing a prob-
lem of knowledge, such as lack of knowledge or contradictory information, in their con-
ceptual structures. Only beliefs in complex, hypothetical, and evolving knowledge are
conducive to that recognition, which requires that students be engaged in learning as prob-
lem solving to be able to intentionally produce changes in knowledge (Mason, 2003).
On the other hand, it emerges from almost all studies on the effects of refutational texts
that they are a powerful means of promoting understanding of complex scientific knowl-
edge at the situation model level (e.g., Alvermann & Hynd, 1989; Guzzetti et al., 1993;
Diakidoy et al., 2003). This positive effect may be explained in terms of the creation
or refinement of metacognitive awareness of one’s current conceptions and the new con-
ceptions to be learned. While the former are introduced as limited and then refuted, the
latter are introduced as viable representations, that is, intelligible, credible, and fruitful.
The power of refutational texts can also be explained in terms of deeper text processing
stimulated by a comparison between the readers’ preconceptions and the new conceptions
to be learned.
The interaction between testing time, epistemological beliefs, and text type is partic-
ularly interesting. A powerful combination of the personal and instructional variables to
support conceptual change emerged. Students with higher levels of epistemological
thinking are those who are better able to take advantage of reading a text that explicitly
states and refutes their preconceptions. A compensation effect between less mature
beliefs about the nature of knowledge and refutational text also emerged since students
who believed less in the complexity and uncertainty of knowledge and read the refuta-
tional text outscored students at the same level of epistemological thinking who read the
traditional text.
The correlation between reading comprehension skills and conceptual change measures
leads us to reflect, once again, on the importance of these skills in academic learning.
Improving students’ reading comprehension is, in any case, an essential way to sustain
their conceptual learning from a text.
The third research question asked if at posttest, participants’ metaconceptual awareness
of the changes in their cognitive structures would also be affected by epistemological
beliefs and text structure. We also wondered if reading a refutational text could compen-
sate less advanced epistemological understanding. Findings add to the literature that both
factors help students think about their thinking, as hypothesized. To believe more that
knowledge is complex and uncertain enhances metaconceptual awareness. Reading a refu-
tational text that states and refutes alternative preconceptions also creates or refines aware-
ness of the changes in one’s cognitive structures. These data are in line with Vosniadou’s
(1994, 2003) recommendation for the creation of a powerful learning environment for con-
ceptual change as well as with Wiser and Amin’s (2001) positive effects of metaconcep-
tual teaching about thermal phenomena. To some extent, our outcomes also confirm the
data from a previous study on the understanding of evolution by Lawson and Worsnop
(1992), who pointed out that reflective reasoning was a strong predictor of learning about
the topic. No interaction between the personal and instructional variables emerged for
metaconceptual awareness.
We should underline that in this study we did not examine students’ religious beliefs, as
in other research in science education (e.g., Dagher & BouJaoude, 1997; Demastes-
Southerland et al., 1995). Since the topic was animal, and not human, evolution, we did
not take into account the consideration that this variable may mediate acceptance of sci-
entific knowledge, which may be understood, but not believed as truth. More recently,
Sinatra and her associates (2003) have showed not only the role of beliefs about knowl-
edge but also cognitive dispositions, such as the disposition to engage in effortful and
open-minded thinking, and not to identify with one’s beliefs but rather to weigh up new
evidence in the understanding and acceptance of human evolution.
Furthermore, it should be pointed out that we investigated only short-term conceptual
change. Learning was only stimulated through the reading of one text. On one hand, this
is in line with what takes place in regular classroom settings, at least in Italy, from the last
years of primary school. New knowledge comes mainly from text reading, especially when
abstract concepts are acquired, so that observation and experimentation can be scarcely
applied to learning, such as in the case of natural selection and biological evolution. On
the other hand, we know that reading a refutational text accompanied by group or pair dis-
cussion can increase the positive effects of the text itself (Hynd, Qian, Ridgeway, & Pickle,
1991). Further research is needed to examine the role and contribution of argumentation,
promoted by social-cognitive interaction, in the construction of more advanced explana-
tions in different teaching conditions.
From the educational point of view, our findings lead us to draw two main implications
for teaching for conceptual change. First of all, since students’ representations about
knowledge affect the process of knowledge construction and reconstruction, it is important
to ascertain their epistemological beliefs. To acquire knowledge is the main, or one of the
main, reasons for attending school. What do students think knowledge is and where does
it come from? (Kuhn, 2000). It is also crucial to implement educational interventions
aimed at changing convictions that are not conducive to deepening conceptual under-
standing (e.g., Conley, Pintrich, Vekiri, & Harrison, 2004). The way in which scientific
knowledge is presented can itself foster or hinder the refinement of thinking about knowl-
edge and knowing.
The second implication regards the instructional material to be studied. Adequate space
should be given, especially for complex topics, to texts that explicitly demonstrate the lim-
itations of alternative conceptions and the fruitfulness of intelligible and credible ones.
This type of text should be considered not only as a source of scientific knowledge but also
as a tool for promoting the refinement of students’ metacognitive awareness. A careful and
critical analysis of the text content and a successful integration of the new information into
conceptual structures are related to students’ awareness of the differences between their
own, and scientific conceptions. All this means supporting students’ deep understanding,
that is, a knowledge construction process which is not fictitious or pseudo, as happens
when new knowledge is ignored, distorted, or remains inert.
Appendix
Excerpt from the Traditional Expository Text
Individuals of the same species are not all exactly the same. There is a great variability
in characteristics, which are partially inheritable, that is, they can be transmitted to the
offspring.
Although Darwin considered the variability in characteristics of the exemplars of a
species a fundamental aspect of his theory, he did not know how it originated.
Nowadays, this variability is attributed to mutations in genetic inheritance, which are
totally a matter of chance, and to a “mixing” of the parents’ characteristics that occurs
through reproduction.
Excerpt from the Refutational Text
Variability is fundamental in Darwin’s theory. Why is variability so important? Imagine

that animals of the same species are exactly the same and perfectly adapted to a given
environment. Imagine that that environment changes. No animal is adaptable to the new
environment; they all die and therefore the species ends. Conversely, if one member of
the species is different from the rest, it is more likely that, by chance, it has some char-
acteristics that are appropriate for the new environment. This “lucky” animal survives,
leaves offspring behind, and the species it belongs to continues. Individuals of the same
species are therefore different. Try and look at a baby. The baby only looks like its father
and mother in part. This is due to a “mixing” of the parents’ characteristics during repro-
duction. Variability is also due to mutations of the genetic inheritance, which are totally
a matter of chance.
References
Alvermann, D., & Hynd, C. (1989). Effects of prior knowledge activation modes and text structure
on nonscience majors’ comprehension of physics. Journal of Educational Research, 83, 97–102.
Alvermann, D., Hynd, C., & Quin, G. (1997). Preservice elementary school teachers’ conceptual
change about projectile motion: Refutation text, demonstration, affective factors, and relevance.
Science Education, 1, 1–27.
Bandiera, M. (1991). Genetica [Genetics]. In: N. Grimellini Tomasini, & G. Segrè (Eds), Conoscenze
scientifiche: Le rappresentazioni mentali degli studenti [Scientific knowledge: Students’ mental
representations] (pp. 313–351). Florence, Italy: La Nuova Italia.
Beck, I. L., McKeown, M. G., Sinatra, G. M., & Loxterman, J. A. (1991). Revising social text from
a text-processing perspective: Evidence of improved comprehensibility. Reading Research
Quarterly, 26, 251–276.
Bishop, B. A., & Anderson, C. W. (1990). Student conceptions of natural selection and its role in
evolution. Journal of Research in Science Teaching, 27, 415–427.
Boscolo, P., & Mason, L. (2003). Prior knowledge, text coherence, and interest: How they interact
in learning from instructional texts. Journal of Experimental Education, 71(2), 126–148.
Braccini, G., Bosco, C., & Durante, T. (1998). Scienze per Temi [Sciences by Themes]. Florence,
Italy: Le Monnier.
Chi, M. T. H., Slotta, J. D., & de Leeuw, N. (1994). From things to processes: A theory of concep-
tual change for learning science concepts. Learning and Instruction, 4, 27–43.
Chinn, C. A., & Brewer, W. F. (1993). The role of anomalous data in knowledge acquisition: A the-
oretical framework and implications for science education. Review of Educational Research, 63,
1–49.
Chinn, C. A., & Brewer, W. F. (1998). An empirical text of a taxonomy of responses to anomalous
data in science. Journal of Research in Science Teaching, 35, 623–654.
Conley, A. M., Pintrich, P. R., Vekiri, J., & Harrison, D. (2004). Changes in epistemological beliefs
in elementary science students. Contemporary Educational Psychology, 29, 186–204.
Cornoldi, C., & Colpo, G. (1995). Nuove Prove di Lettura MT per la Scuola Media Inferiore [New
MT reading tests for the middle school]. Florence, Italy: Organizzazioni Speciali.
Dagher, Z., & BouJaoude, S. (1997). Scientific views and religious beliefs of college students: The
case of biological evolution. Journal of Research in Science Teaching, 34, 429–445.
Demastes-Southerland, S., Good, R. G., & Peebles, P. (1995). Students’ conceptual ecologies and the
process of conceptual change in evolution. Science Education, 79(6), 637–666.
Diakidoy, I. N., Kendeou, P., & Ioannides, C. (2003). Reading about energy: The effects of text struc-
ture in science learning and conceptual change. Contemporary Educational Psychology, 28,
335–356.
Dole, J. A., & Sinatra, G. M. (1998). Reconceptualizing change in the cognitive construction of
knowledge. Educational Psychologist, 33(2/3), 109–128.
Ferrari, M., & Chi, M. T. H. (1998). The nature of naive explanations of natural selection.
International Journal of Science Education, 20, 1231–1256.
Gregoire-Gill, M., Ashton, P. T., & Algina, J. (2004). Changing preservice teachers’ epistemological
beliefs about teaching and learning in mathematics: An intervention study. Contemporary
Guzzetti, B. J., Snyder, T. E., Glass, G. V., & Gamas, W. S. (1993). Promoting conceptual change in
science: A comparative metaanalysis of instructional interventions from reading education and
science education. Reading Research Quarterly, 28, 117–159.
knowledge and knowing and their relation to learning. Review of Educational Research, 67,
88–140.
Hofer, B. K., & Pintrich, P. R. (Eds). (2002). Personal epistemology: The psychology of beliefs about
knowledge and knowing. Mahwah, NJ: Laurence Erlbaum Associates.
Hynd, C. (2003). Conceptual change in response to persuasive messages. In: G. M. Sinatra, & P. R.
Pintrich (Eds), Intentional conceptual change (pp. 291–315). Mahwah, NJ: Lawrence Erlbaum
Associates.
Hynd, C., McWhorter, Y., Phares, V., & Suttles, W. (1994). The role of instructional variables in con-
ceptual change in high school physics topics. Journal of Research in Science Teaching, 31,
933–946.
Hynd, C., Qian, G., Ridgeway, V., & Pickle, M. (1991). Promoting conceptual change with science
texts and discussion. Journal of Reading, 34, 596–601.
Kintsch, W. (1988). The use of knowledge in discourse comprehension: A construction-integration
model. Psychological Review, 95, 163–182.
Kuhn, D. (2000). Theory of mind, metacognition, and reasoning: A life-span perspective. In: P.
Mitchell, & K. J. Riggs (Eds), Children’s reasoning and the mind (pp. 301–326). Hove, UK:
Psychology Press.
Lando, M. (2003). Credenze epistemologiche generali e dominio-specifiche [Domain-general and
domain-specific epistemological beliefs]. Unpublished degree dissertation. University of Padua, Italy.
Lawson, A. E., & Thompson, L. D. (1988). Formal reasoning ability and misconceptions concerning
genetics and natural selection. Journal of Research in Science Teaching, 25, 73–746.
Lawson, A. E., & Weser, J. (1990). The rejection of nonscientific beliefs about life: Effects of
instruction and reasoning skills. Journal of Research in Science Teaching, 27, 589–606.
Lawson, A. E., & Worsnop, W. A. (1992). Learning about evolution and rejecting a belief in special
creation: Effects of reflective reasoning skill, prior knowledge, prior belief and religious commit-
ment. Journal of Research in Science Teaching, 29, 143–166.
Limón, M., & Mason, L. (Eds). (2002). Reconsidering conceptual change. Issues in theory and prac-
tice. Dordrecht, The Netherlands: Kluwer.
Mason, L. (2000). Role of anomalous data and epistemological beliefs in middle students’ theory
change on two controversial topics. European Journal of Psychology of Education, 15, 329–346.
Mason, L. (2001). Responses to anomalous data on controversial topics and theory change. Learning
and Instruction, 11, 453–483.
Mason, L. (2003). Personal epistemologies and intentional conceptual change. In: G. M. Sinatra, &
P. R. Pintrich (Eds), Intentional conceptual change (pp. 199–236). Mahwah, NJ: Lawrence
Erlbaum Associates.
McNamara, D. S., Kintsch, E., Songer, N. B., & Kintsch, W. (1996). Are good texts always better?
Interactions of text coherence, background knowledge, and levels of understanding in learning
from text. Cognition and Instruction, 14, 1–43.
Mikkilä-Erdmann, M. (2002). Science learning through text: The effect of text design and text com-
prehension skills on conceptual change. In: M. Limón, & L. Mason (Eds), Reconsidering con-
ceptual change. Issues in theory and practice (pp. 337–356). Dordrecht, The Netherlands: Kluwer.
conception: Toward a theory of conceptual change. Science Education, 66, 211–227.
Qian, G., & Alvermann, D. (1995). Role of epistemological beliefs and learned helplessness in sec-
ondary school students’ learning science concepts from text. Journal of Educational Psychology,
87, 282–292.
Qian, G., & Alvermann, D. E. (2000). The relationship between epistemological beliefs and con-
ceptual change learning. Reading & Writing Quarterly, 16, 59–74.
Qian, G., & Pan, J. (2002). A comparison of epistemological beliefs and learning from science text
between American and Chinese high school students. In: B. K. Hofer, & P. R. Pintrich (Eds),
Personal epistemology. The psychology of beliefs about knowledge and knowing (pp. 365–385).
Schnotz, W., Vosniadou, S., & Carretero, M. (Eds). (1999). New perspectives on conceptual change.
Amsterdam: Pergamon.
Schommer, M. (1993). Epistemological development and academic performance among secondary
students. Journal of Educational Psychology, 85, 406–411.
Sinatra, G. M. (2005). The “warming trend” in conceptual change research: The legacy of Paul R.
Pintrich. Educational Psychologist, 40, 107–115.
Sinatra, G. M., & Pintrich, P. R. (Eds). (2003a). Intentional conceptual change. Mahwah, NJ:
Lawrence Erlbaum Associates.
Sinatra, G. M., & Pintrich, P. R. (Eds). (2003b). The role of intentions in conceptual change learn-
ing. In: G. M. Sinatra, & P. R. Pintrich (Eds), Intentional conceptual change (pp. 1–18). Mahwah,
NJ: Lawrence Erlbaum Associates.
Sinatra, G. M., Southerland, S. A., McConaughy, F., & Demastes, J. (2003). Intentions and beliefs
in students’ understanding and acceptance of biological evolution. Journal of Research in Science
Teaching, 40, 510–528.
Southerland, S. A., & Sinatra, G. M. (2003). Learning about biological evolution: A special case of
intentional conceptual change. In: G. M. Sinatra, & P. R. Pintrich (Eds), Intentional conceptual
change (pp. 317–345). Mahwah, NJ: Lawrence Erlbaum Associates.
Vosniadou, S. (1994). Capturing and modeling the process of conceptual change. Learning and
Vosniadou, S. (2003). Exploring the relationships between conceptual change and intentional learn-
ing. In: G. M. Sinatra, & P. R. Pintrich (Eds), Intentional conceptual change (pp. 377–406).
Windschitl, M., & Andre, T. (1998). Using computer simulations to enhance conceptual change: The
roles of constructivist instruction and student epistemological beliefs. Journal of Research in
Science Teaching, 35, 145–160.
Wiser, M., & Amin, T. (2001). “Is heat hot?” Inducing conceptual change by integrating everyday
and scientific perspectives on thermal phenomena. Learning and Instruction, 11, 331–355.
Chapter 14
Conceptual Change Ideas: Teachers’ Views

and their Instructional Practice
Reinders Duit, Ari Widodo and Christoph T. Wodzinski
Objectives
Conceptual change views have played a key role in constructivist-oriented attempts to

improve science teaching and learning since the 1980s (Duit & Treagust, 1998, 2003;
Mason, 2001; Schnotz, Vosniadou, & Carretero, 1999). Whereas constructivist ideas pro-
vide the epistemological frame for conceptualizing teaching and learning processes, the
term conceptual change denotes that learning science includes major reconstruction of the
already existing conceptual structures (Duit, 1999). Conceptual change views play a
major part in recent programs to improve the quality of science instruction (Beeth et al.,
2003). It seems, however, that constructivist ideas and findings of research on conceptual
change are not familiar to most science teachers. This lack of teachers’ familiarity with
the recent state of research on efficient teaching and learning is seen as the key barrier
towards improving science teaching and learning (Anderson & Helms, 2001). Teachers’
epistemological beliefs about teaching and learning as well as their views about efficient
instruction have to undergo major conceptual changes. Fostering these changes has
played a significant role in approaches of teacher professional development (West &
Staub, 2003).
The study presented here does not address the issue of changing teachers’ views. It is
investigating whether a sample of German physics teachers is familiar with conceptual
change ideas, on the one hand, and whether their instruction meets key characteristics of
such approaches, on the other. Subsequent studies drawing on these results attempt to
develop teachers’ epistemological beliefs and views about instruction as well as their
instructional behavior.
An analysis of the literature has resulted in two sets of categories denoting key issues
of conceptual change ideas. The first category system, COSC (constructivist-oriented sci-
ence classrooms), is designed to identify the appearance of key characteristics of
constructivist learning environments (Phillips, 2000). The second category system, CTS

ISBN: 0-08-045355-4
198 Reinders Duit et al.
(constructivist teaching sequences), allows one to investigate whether the phase structure
of instruction follows suggestions for constructivist teaching sequences in the literature
(e.g., Biemans, Deel, & Simons, 2001; Driver, 1989).1 Additional data on teachers’ think-
ing about teaching and learning science are available from teachers’ interviews.2
The study aims at investigating the appearance of key characteristics of construc-
tivist learning environments and constructivist teaching sequences in ordinary science
classrooms. It is not the intention to use constructivism to judge teachers’ teaching,
rather the main objectives of the study are to explore the appearance of constructivist
ideas in ordinary classroom practice, to identify teachers’ difficulties to implement con-
structivist ideas, and to explore the possibilities of using constructivist ideas to improve
practice.
The study presented is part of the first phase of a larger project that aims at investigat-
ing physics teachers’ and students’ scripts of introductory physics instruction.3 The study
shares major features of the TIMS Video Study on mathematics instruction in the USA,
Japan, and Germany (Stigler, Gonzales, Kawanaka, Knoll, & Serrano, 1999) and of the
more recent TIMSS-R video studies on science instruction (Roth et al., 2001).
Constructivist-Oriented Science Classrooms
Constructivist principles of teaching and learning (e.g., Driver, 1989; Duit & Treagust, 1998,
2003; Matthews, 2000; Phillips, 2000; Watts, 1994) provide the theoretical framework for the
development of the coding system to identify key characteristics of constructivist informed
learning environments. Constructivism suggests that first, knowledge is human construction.
From the constructivist point of view, knowledge is not an objective representation of the
world; rather it is a human construct. Natural objects or phenomena are themselves “objec-
tive” and “real” but the observations and interpretations of them are affected by the subjec-
tive interpretation schemes of the observer. Second, knowledge is constructed within certain
social and material contexts and, consequently, it is affected by issues such as ideologies,
religion, politics, economics, human interests, and by the particular material features of the
learning environment. Third, knowledge is tentative. Our knowledge about the world is not
a mere copy of the reality outside but it is our tentative construction about it. Scientific truth
is not absolute but it is relative and may change over time.
1
Ari Widodo developed the coding instruments COSC and CTS. For more details of the instruments and the
results of his study see Widodo (2004). The Coding Manuals are also available from the first author of the pres-
ent chapter (duit@ipn.uni-kiel.de).
2
Christoph T. Wodzinski (née Müller) was responsible for developing methods to analyze the interviews and for
interpreting the results (see Müller, 2004)
3
The team of the first phase of this project comprises: Manfred Prenzel, Tina Seidel, Reinders Duit, Manfred
Lehrke, Rolf Rimmele, Christoph T. Müller, Maike Tesch, Lena Meyer, Inger Marie Dalehefte and Ari Widodo.
The project is part of the priority program “BIQUA - The Quality of School: Studying Students’ Learning in Math
and Science and Their Cross-Curricular Competencies Depending on In-School and Out-of-School Contexts”
sponsored by the German Science Foundation that includes a total of 23 projects.
Conceptual Change Ideas 199
Since constructivism includes views of what knowledge is and how knowledge is

acquired, principles emerge from both perspectives that need to be taken into account in
developing criteria for “constructivist learning environments.” Science learning should
provide students with opportunities to experience science as a body of knowledge and as
a knowledge generation process (Duschl & Gitomer, 1991). Students should learn facts,
laws, and theories and how science works and knowledge is developed. It suggests that
principles of the nature of science (McComas, 1998; OECD PISA, 1999) should also be
considered in teaching science.
Key characteristics of constructivist-oriented learning environments have been pro-
vided in a number of research projects. In developing our COSC instrument we draw espe-
cially on the two versions of CLES (constructivist learning environment scale; Aldridge,
Fraser, Taylor, & Chen, 2000; Fraser, 1998; Taylor, Fraser, & Fisher, 1997; Taylor &
Fraser, 1991) and STAM (secondary teacher analysis matrix —science version; Gallagher
& Parker, 1995). Furthermore, criteria for constructivist learning environments developed
by Labudde (2000) as well as Tenenbaum, Naidu, Jegede, and Austin (2001) are taken into
account. The COSC consists of five categories and each is developed further into three to
six subcategories (Table 14.1).
The first category, “Facilitating Knowledge Constructions,” represents that knowledge
is seen as human construction, that learners have pre-instructional conceptions, that learn-
ing is an active process of knowledge construction, and that learning is a change in learn-
ers’ conceptions, that is, a conceptual change. It identifies the extent to which students’
prior knowledge and conceptual change strategies are explored and employed to facilitate
students’ knowledge constructions.
The second category, “The Relevance and the Meaningfulness of the Learning
Experience,” represents the view that knowledge construction is embedded within a par-
ticular social and material context that may support or hamper conceptual change, that is,
students’ construction processes. This category identifies the extent to which students’
learning needs are addressed and how resources are utilized to provide relevant and mean-
ingful learning experiences to the students.
The third category, “Social Interactions,” focuses on the issue that knowledge is socially
constructed. It identifies the extent to which students are given opportunities to socially
interact with each other and the teacher, in different forms of social organizations. Different
forms of social organizations provide different learning experiences to the students. Social
interaction is also essential for conceptual change since students need to be informed about
the ideas of other students and realize that their point of view may not be shared by others.
The fourth category, “Fostering Students to be Independent Learners,” represents con-
structivist views that learners are purposive and ultimately responsible for their own learn-
ing. It identifies the extent to which students are given some freedom to organize their own
learning and are fostered to be independent learners (see Sinatra and Pintrich (2003) on the
importance of independent learning for intentional conceptual change).
The fifth category, “Science, Scientific Knowledge, and Scientists,” represents the con-
structivist views that science knowledge is human construction and that science knowledge
is tentative (see McComas, 1998). It identifies the extent to which science lessons provide
the students with opportunities to experience how scientific knowledge is constantly being
developed and revised.
Table 14.1: Constructivist-oriented science classrooms (COSC).
200
A. Facilitating B. Relevance and C. Social D. Fostering students E. Science, scientific
knowledge meaningfulness of the interactions to be independent knowledge, and
Reinders Duit et al.

constructions learning experience learners scientists
1. Making the students 1. Exploring students’ 1. Student–student 1. Providing the students 1. Acknowledging
aware of the status of interests, attitudes, and interactions with some freedom to the tentativeness of
their learning within feelings organize their own science
the whole subject learning
2. Exploring students’ 2. Addressing students’ a. Simple interactions 2. Encouraging the 2. Acknowledging
prior knowledge or learning needs among the students students to rethink their differences in
ideas own ideas theories or views
3. Exploring students’ 3. Addressing real-life b. Students exchange 3. Encouraging the 3. The roles of
ways of thinking events, phenomena, or ideas with other students to be self- observation and
examples students regulative and reflective evidence, hypotheses,
theories, and laws
in science
4. Providing thinking- 4. Using resources 2. Student–teacher 4. Taking into account 4. Acknowledging
provoking problems from everyday life interactions students’ critical voices differences in the
ways to do science
5. Addressing students’ 5. Discussing a. Simple interactions 5. Acknowledging
conceptions applications of the between students and the limitations of
concepts learned the teacher science explanations
a. Using evolutionary b. Students exchange
ways ideas with the teacher
b. Using evolutionary 3. Social organization
ways of the class
a. Individual setting
b. Group setting
c. Classroom setting
Constructivist Teaching Sequences
Major characteristics of the previously described COSC instrument are the way teachers
address students’ conceptions when developing physics concepts, and whether students’
interests, attitudes, and needs are explicitly taken into account, in short, whether a learn-
ing environment is provided that persistently supports students’ construction processes,
that is, fosters conceptual change. In order to allow a more fine-graded investigation of
teachers’ attempts to guide students from their pre-instructional conceptions to the physics
concepts, we have developed an additional coding system. The sequencing of instruction
observed is analyzed from the perspective of constructivist teaching and learning
sequences which have played a significant role in conceptual change approaches. For con-
structing the “reference model” for steps of constructivist teaching sequences in Figure
14.1, we draw on major approaches in the literature (Biemans & Simons, 1999; Cosgrove
& Osborne, 1985; Driver, 1989; Hodson, 1993; Lawson, Abraham, & Renner, 1989;
Nunez-Oviedo, Clement, & Rea-Ramirez, 2002). According to our model, constructivist
teaching progresses in a spiral sequence of five steps:
1. Introduction: It identifies efforts to prepare students for the topic, to promote students’
readiness, and to generate students’ interests in the lesson.
2. Exploring students’ prior knowledge: It identifies teachers’ efforts to explore students’
prior knowledge related to the topic.
3. Restructuring students’ conceptions: It identifies attempts to facilitate conceptual
change.
4. Applying the newly constructed ideas: It identifies attempts to apply the concepts
learned to other contexts or to real life.
5. Reviewing the new ideas: It identifies attempts to encourage students to compare the
newly achieved and the previous conceptions.
Exploring
students’
12 Introduction 22 conceptions as
developed so far
Exploring
2 students’ 32 Restructuring
the conceptions
1 Introduction pre-conceptions
Reviewing and Restructuring

5 evaluating 3 the conceptions
the new conceptions
Applying
4 the new
conceptions
Figure 14.1: Steps of the constructivist teaching sequences for conceptual change.
Design of the Study
Thirteen classes of schools in two German states participating in a national quality deve-
lopment program comprise the sample (13 teachers; 344 grades 7 and 8 students). Two
three-lessons sequences on introduction to the electric circuit and the force concept were
video-documented for each teacher (Figure 14.2).
Major data sources are the videos of the lessons. Further data sources are student ques-
tionnaires. In the beginning and at the end of the school year, a questionnaire was used to
investigate the development of students’ affective variables (e.g., interests, self-concept,
and meta-cognitive views) and achievement (regarding the two topics video-documented).
After each lesson, video-documented brief questionnaires — especially on students’ views
of instructional activities and learning motivation during instruction — were provided. On
the teacher side, there was a brief questionnaire on their views of learning in the beginning
of the school year. Teachers were also asked to provide drafts of the planned instruction on
the two topics documented. After videotaping the second topic, teachers were interviewed
about various facets of teaching and learning physics. They were also asked to comment
on three video-documented episodes from their instruction.
Various quantitative and qualitative data are available. Accordingly, quantitative and
qualitative research methods are employed (see for a similar approach the “video-survey
methodology” of TIMSS-R Video; Roth et al., 2001).
Two video cameras were used. One of them followed the actions of the teachers, the other
documented the whole class. Every lesson was fully transcribed and coded in various ways
(Prenzel, Duit, Euler, Lehrke, & Seidel, 2001); for a summary Duit, Müller, Tesch, &
Widodo, 2004) by using the software “Videograph” (Rimmele, 2003) developed by one of
our team members.4 For each of the coding systems used the interrater-reliabilty was
carefully examined. In the following, we refer only to the coding systems COSC and CTS
September 2000 July 2001
Video Topic 1 Video Topic 2

3 periods 3 periods
Q1 Qca Qca TI Q 2
Q1/Q2 : Students' questionnaires on interest and assessment
of knowledge
Qca : Students' questionnaires with regard to the videotaped-
lessons
TI : Teachers' interviews withregardtothe videotaped-
lessons
Figure 14.2: Design of the study.
4
Available from: http://www.ipn.uni-kiel.de/projekte/video/videostu.htm
presented above. For them, on average two independent coders agreed in more than 90% of
the decisions.
The COSC and the CTS are designed as time-based category systems. In this study the
coding time unit is 10 seconds. The COSC is designed as a two-option category system. In
coding practice, after observing a 10-second unit, coders decide whether the unit meets the
criteria for the related subcategories. When the unit meets the criteria, it is coded “yes” and
when it does not meet the criteria it is coded “no.” The CTS is designed as a five-option
category system.
Results: Learning Environments of the Lessons (COSC)

In general, the results from the COSC show that the lessons meet the characteristics of
constructivist-oriented science classrooms only to a certain extent. Most subcategories
were observed for less than 10 minutes. A number of characteristics were not observed at
all, namely subcategories D-3 and D-4 as well as all subcategories of category E.
Facilitating Knowledge Constructions
Two important characteristics for facilitating students to construct knowledge, namely

making the students aware of the status of their learning and exploring students’ ways of
thinking, were very seldom observed, while the other three subcategories were observed
more often (Figure 14.3). Regarding the issue of making students aware of the status of
their learning, a number of teachers in the interview claimed that they do not provide a
“preview” of the lesson because they do not like to lose moments of surprise.
Although exploring students’ prior knowledge was observed relatively often, fre-
quently little was obtained from these attempts. Usually teachers asked the students to
think about what they already know about the topic in question, sometimes requesting
them to write down what they think. However, the questions asked by the teachers were
usually very general and the identified prior knowledge not specific enough to allow the
10
Observed(in minutes)
8
6
4
2
0
A-1 A-2 A-3 A-4 A-5
subcategory
A-1: Making students aware of the status of their learning.

A-2: Exploring students’ prior knowledge or ideas.
A-3: Exploring students’ ways of thinking.
A-4: Providing thinking-provoking problems.
A-5: Addressing students' conceptions.
Figure 14.3: Results of subcategories A-1 to A-5.

teachers to detect students’ pre-conceptions. Additionally, after exploring students’ prior

knowledge, often no further actions were taken to address it. It seems that the teachers
did not have clear ideas of how to make use of students’ prior knowledge. This finding
is similar to Hewson, Tabachnick, Zeichner, and Lemberger’s (1999) results that
prospective teachers explored students’ prior knowledge, but only few were able to use
it to plan their teaching.
At first sight, it is rather pleasing that a substantial amount of time (in average about
20%) was spent to address students’ conceptions — which is a key issue in constructivist
conceptual change approaches. Only seldom observed were the use of analogies and cog-
nitive conflicts. “Evolutionary” strategies (drawing on step by step development of ideas)
were more frequent than “revolutionary” strategies (drawing on sudden insight facilitated
by cognitive conflicts). As will be discussed later (see the results of the CTS instrument),
the use of conceptual change strategies as proposed in the literature (Duit, 1999) was rather
infrequent. This finding is not surprising taking into account that the interviews revealed
that most teachers were not well informed about the literature on conceptual change
(Müller, 2004). The most frequent kind of addressing students’ conceptions was embedded
in the dominant “questioning–developing” mode of classroom discourse. The teacher
opened the discussion with a question and in a series of further questions the science expla-
nation was developed. This attempt to actively include students in the development of the
concept in question was often reduced to a dyadic question–answer game in which the
teacher asked a question, a student gave an answer, and the teacher only took the answers
on board that led to the intended solution.
Briefly summarized, from the perspective of conceptual change strategies, there were
rather limited opportunities for the students to become aware of their ideas and in which
way they differed from the science view. There was further rather limited support for
students to construct the new science view.
Relevance and Meaningfulness of The Learning Experience
As shown in Figure 14.4, lessons seldom addressed issues related to students’ interests
and learning needs (B-1, B-2). However, the lessons were quite often situated within
students’ everyday life as they frequently addressed real-life phenomena or real-life
examples (B-3) and used resources from everyday life (B-4). These results suggest that
there were significant efforts by the teachers to provide learning experiences that were
relevant to students’ everyday life experiences. Although it was less frequently observed
than subcategories B-3 and B-4, application of knowledge (B-5) was also given certain
attention. Included were applications addressing issues of real life.
Social Interactions
Simple interactions were more common than intensive interactions that include exchange
of ideas5 (Figure 14.5). Simple interaction denotes primarily the common dyadic style
5
Due to the difficulties to code student–student interactions, no results for subcategories C-1a and C-1b are available.
where the teacher asks a question, a student gives an answer, the teacher asks the next
question and so forth. Intensive interactions concern classroom discourses in which
students have a voice and that include true exchanges of ideas. Further, students had only
limited opportunities to work in groups and to work individually. In some lessons only
classwork but no student individual or student group work occurred. Since lessons were
generally rather classwork-oriented and less individual-oriented, it was unlikely that
lessons catered to students’ individual learning needs and speeds.
Fostering Students to be Independent Learners
As shown in Figure 14.6, teachers put remarkable efforts on fostering students’ independ-
ence (see D-1). However, other aspects required for students to be independent learners
(especially D-3 and D-4) were seldom observed. For example, one of the important features
of being independent learners is that students should be able to learn how to learn, how to
control and to monitor their own learning. Unfortunately, these issues were not well
addressed by the teachers. It was rarely observed that students’ critical voices were taken
into account. Taylor et al. (1997) claim that it is an essential feature of efficient teaching and
Observed (in minute)
20
15
10
5
0
B-1 B-2 B-3 B-4 B-5
subcategory
B-1: Exploring students’ interests, attitudes, and feelings.

B-2: Addressing students’ learning needs.
B-3: Addressing real-life events, phenomena, or examples.
B-4: Using resources from everyday life.
B-5: Discussing applications of the concepts learned.
Figure 14.4: Results of subcategories B-1 to B-5.

Observed (in minute)
40
30
20
10
0
C-2a C-2b C-3a C-3b C-3c
subcategory
C-2a: Simple interactions between students and the teacher

C-2b: Students exchange ideas with the teacher
C-3a: Individual setting
C-3b: Group setting
C-3c: Classroom setting.
Figure 14.5: Results of subcategories C-2 and C-3.

Observed (in minutes)

12
10
8
6
4
2
0
D-1 D-2 D-3 D-4
subcategory
D-1: Providing students with some freedom to organise their own learning.
D-2: Encouraging students to re-think their own ideas.
D-3: Encouraging students to be self-regulative and reflective.
D-4: Taking into account students’ critical voices.
Figure 14.6: Results of subcategory D-1 to D-4.
learning that students have a voice in instruction, that is, that students’ contributions to
classroom discourse are taken seriously by the teacher. Hanrahan (1998) argued that this
may only be expected if students are explicitly encouraged to do that.
Science, Scientific Knowledge, and Scientists
The subcategories denoting views of the nature of science could not be observed. In other
words, they did not play any role in the introductory physics instruction video-docu-
mented. In the interviews, some teachers argued that it is too early for such considerations
at introductory level. But it became also evident that many teachers were not very famil-
iar with views of the nature of science addressed by the category E.
Results: Constructivist Teaching Sequences (CTS)
As presented in Figure 14.7, all five steps were observed in the lessons. However, this does
not mean that each lesson always included all five steps. The sum of the length of all steps
reveals that in only 21 minutes of the time, steps of the constructivist teaching sequences
could be observed. This suggests that about 50% of teaching time was spent for other
issues, such as addressing discipline, taking notes, or doing activities that do not clearly
refer to any of the five steps. Analyses of the individual lessons show that many lessons
had not progressed in steps as suggested by the constructivist teaching sequences.
To provide examples, Figure 14.8 displays the appearance of the steps for two teachers’
introductory lessons of the topic “electric circuit.” A number of lessons did not show com-
plete steps. T-7’s lesson in Figure 14.8, for instance, did not include application and review.
Further, a number of lessons did not progress in steps as suggested by the constructivist
teaching sequences. T-1’s lesson in Figure 14.8 to a certain extent meets the CTS. It began
with a short introduction, followed by extended explorations of students’ prior knowledge.
By the end of the lesson a short restructuring was done followed by application and review.
In contrast, T-7’s lesson included only introduction, exploration, and restructuring.
Length (in minute)

Steps:
6 1. Introduction
2. Exploring
4
3. Restructuring
2 4. Applying
5. Review
0
1 2 3 4 5
Steps
Figure 14.7: Average time of each step of the CTS.
T-1/E-1
5
4 Steps:
1. Introduction
3
Steps
2. Exploring
2 3. Restructuring
1 4. Applying
0 5. Review
0 5 10 15 20 25 30 35 40 45
Minute
T-7/E-1
5
Steps:
4 1. Introduction
3 2. Exploring
Steps
2 3. Restructuring
4. Applying
1
5. Review
0
0 5 10 15 20 25 30 35 40 45
Minute
Figure 14.8: Analyses of the sequences of lessons.
Analyses of the sequences reveal the following findings. First, in 26 lessons (38%), one or
more steps were missing. The fourth step (applying the conceptions) and the fifth step
(reviewing) were the two most frequently missing steps. Second, in many lessons one or
more steps were skipped or they progressed in unclear sequences.
Patterns of Instruction and the Development of Achievement and

Interests
In order to investigate relations between certain patterns of instruction and the develop-
ment of student cognitive and affective variables within the school year of videotaping
(Figure 14.2), correlations between the frequencies of certain codings, on the one hand,
and the development of student achievement and affective variables (like interests) meas-
ured by a pre- and a post-questionnaire, on the other, were calculated. Due to the small
number of classes participating in the study these measures have to be interpreted with
great care. At best they provide preliminary hypotheses that have to be further investigated
in subsequent studies.
For the coding system COSC there are two subcategories where substantially high
correlations (we used partial correlations) with student achievement occurred — for the
length of time students are challenged with thinking-provoking problems (A-4) and the
time spent in addressing students’ conceptions (A-5). These results are in accordance with
findings from other studies (e.g., Ditton, 2002) that constructivist learning environments
may promote students’ learning. There are also somewhat weak indications that embed-
ding instruction in everyday life contexts is linked to advanced achievement. Interestingly,
there is no substantial correlation between achievement gains and subcategory A-2
“Exploring students’ prior knowledge or ideas.” As argued above, the limited scope of the
teachers’ exploration attempts may be responsible for this result, which at first sight
appears to be in contradiction to arguments in the constructivist literature that exploring
students’ ideas is essential. Substantial correlations between certain codings (denoting cer-
tain instructional patterns) and the development of affective variables could not be
observed (for a detailed discussion see Widodo, 2004).
For the coding system CTS there is a significant correlation between the total time in
which steps of the CTS were observed for the 13 teachers and student achievements gains
(Widodo, 2004). This finding appears to indicate that the five coded steps (Figure 14.1)
denote the essential phases of effective instruction. This result is in accordance with find-
ings from other studies that orienting instruction at constructivist teaching sequences
promotes students’ learning (e.g., Biemans et al., 2001).
Teacher Interviews
In-depth interviews with each of the 13 teachers provide the data for investigating teach-
ers’ epistemological beliefs about teaching and learning as well as their views about effi-
cient instruction. The study presented here also explores the match between teachers’
beliefs and views and their actual classroom behavior. Further attention is given to the rela-
tions between teachers’ views and teaching behavior on the one hand, and the development
of student achievement and affective variables on the other.
The first part of the interviews explored teachers’ ideas about the key issues of the
study, namely:
• the role of experiments,
• the significance of various science processes,
• views of the nature of science,
• their preferred aims, and
• their epistemological beliefs of teaching and learning.
In the second part of the interview teachers were presented with three video sequences
from their instruction: (1) introduction into one of the topics taught, (2) a typical whole
class activity, and (3) a typical experiment. The semistructured interviews lasted between
60 and 90 minutes. They were carried out by the first author. They were audio-documented
and fully transcribed.
The interviews were analyzed from two different but complementary perspectives.
First, a qualitative analysis drawing on various theoretical perspectives (like theoretical
frames on the role of experiments and processes or constructivist views) was carried out.
This analysis was based on the “qualitative content analysis” as provided by Mayring
(2000). This analysis proceeds in three steps. As a first step, the utterances documented in
the transcripts are paraphrased in such a way that key features of the meaning become
apparent. As a second step, these key features are thematically organized and summarized
as “prototypical statements.” These statements — as a third and last step — are systemat-
ically categorized. This analysis resulted in a brief summary description of teachers’ views
concerning the above-mentioned key issues.
Second, a system of categories was developed to carry out a turn-to-turn coding of
the interviews. These categories were oriented at the categories used for coding the les-
sons in order to allow an analysis of what teachers think and what they do in class-
rooms. Due to the small sample of 13 teachers, statistical analysis can only be used to
generate hypothetical information on eventually relevant interdependencies. For this
attempt, a cluster analysis was conducted on the quantitative data of the coding process.
It resulted in four types of teachers’ implicit theories. These types matched the results
of the qualitative analysis.
Teachers’ implicit theories were compared with their actual (videotaped) classroom
practice. Finally, relations between the types of implicit theories on the one hand and stu-
dent achievement gains over one school year and the development of students’ affective
variables on the other were investigated.
Findings on Teachers’ Views of Teaching and Learning Physics
Teachers’ Familiarity with the Recent Literature on Teaching and Learning Science
and Their Views of Learning Science
Most teachers are not well informed about key ideas of conceptual change research, nei-
ther of constructivist theories nor of conceptions of teaching and learning in general. Their
views of their students’ learning usually do not meet the state of recent theories of teach-
ing and learning. In general, most teachers appear to lack an explicit view of learning.
Some of the teachers hold implicit theories, that contain some “intuitive” constructivist
aspects, for instance, they want to be learning counselors, and they are aware of the impor-
tance of students’ cognitive activity and the interpreting character of students’ observations
and understanding. However, we also find teachers who characterize themselves as medi-
ators of facts and information and who are not aware of students’ interpretational frames
and the role of students’ pre-instructional conceptions. They mostly think that giving good
instructions is a guarantee for successful learning. Only one teacher was well informed
about current theories of teaching and learning, namely constructivist theories. Ironically,
the instruction of this teacher was the least constructivist.
The analyses of the interviews also revealed that the following way of thinking about
instruction is predominating. Major concerns are considerations on the science topic in
question. Considerations on students’ learning capabilities and difficulties are often mar-
ginal or nearly totally missing. These findings clearly have to do with the previously men-
tioned lack of explicit views about students’ learning.
Types of Teachers’ Implicit Theories
The cluster analysis of quantitative coding of the teachers’ interviews resulted in four types
of teachers’ implicit theories (Table 14.2).
Table 14.2: Types of teachers’ implicit theories based on the analysis of teachers’ interviews.
Cognitive apprenticeship Cognitive self-effort
Learning viewed as stepwise Topic-oriented learning

supported development in opportunities should be
argumentative discourse provided — but support is not
explicitly mentioned
Mistakes are learning opportunities
Conceptual Type C Type B

Growth View themselves primarily as View themselves as
physicists pushing forward, to
Aim of instruction: introduction overcome learning resistance
into the physics view, including Clear flow of instruction
understanding concepts, principle (order, precision)
and processes Aim of instruction: Incite
interest to understand
everyday life issues
Conceptual Type A Type D

Change
View themselves as teachers View themselves as
providers of learning
Teaching oriented to student
opportunities
understanding; provide students
with chances to find out Expect/miss interest &
themselves and to think student initiatives
Aims: modeling, understanding Students shall find out
everyday life themselves
Instruction open but not
substantially structured
The succession in which the clusters occurred in the analysis was type A, B, C, and D.
These types could be described by certain characteristic items included in the cluster
analysis. Based on these items, two theoretically founded dimensions were constructed.
Each type of teachers’ implicit theories could be allocated to one of the cells in this two-
dimensional framework. For the first dimension, one pole represents the idea that learning
can be viewed as stepwise supported development in argumentative discourse, while the
other pole represents the idea that teaching means to provide topic-oriented learning
opportunities, whereby the need to support the use of the learning opportunities is not
explicitly mentioned. The poles of this dimension could be characterized by “cognitive
apprenticeship” and “cognitive self-effort.” For the second dimension, one pole represents
the idea that teaching means to provide opportunities to change students’ scientific con-
cepts, while the other pole represents the idea that teaching physics means to present phys-
ical phenomena and to mediate the structure of the topic. Here, the poles of the dimension
could be called “conceptual change” and “conceptual growth.” Taking these two dimen-
sions together, we can identify four types of teachers as described below.
Statistical analysis reveals substantial differences between the classes allocated to the
types of implicit theories concerning student achievement gains over one school year. Due
to the small sample of 13 teachers, the following results have to be interpreted with cau-
tion. They may be viewed as hypotheses only. Between the type of teacher (being a mem-
ber of one type of implicit theories) and students’ achievement gains (pre–post, i.e., from
the beginning to the end of the school year) we found a positive and significant correlation
(r ! 0.697, p ! 0.008 for Rasch parameters; r ! 0.755, p ! 0.003 for residuals; Figure
14.9). Between the type of teacher and students’ affective variables we found no signifi-
cant correlations. Nevertheless, we can observe substantial differences in the development
of interest, self-concept, and self-evaluated competence between students of the different
types of teachers.
2.5 10
achievement
gains (Rasch) 8
achievement
2 gains (residual) 6
achievement gains (residual)
4
achievement gains (Rasch)
1.5 2
1 -2
-4
0.5 -6
-8
0 -10
type A type B type C type D
Figure 14.9: Achievement gains (Rasch parameters and residuals) for types A to D.
Regarding achievement gains, the average gain for classes of the “cognitive appren-
ticeship” teachers is significantly higher than for the “cognitive self-effort” teachers
(Figure 14.9). There is also a somewhat marginal difference between the “conceptual
growth” and “conceptual change” poles of the “cognitive apprenticeship” dimension with
the “conceptual change” teachers being ahead.
From our point of view, these results indicate that to foster students’ learning in the
sense of constructing their own knowledge it is necessary not only to provide opportuni-
ties for students to think and work independently but to structure learning processes.
Students seem to need stepwise guidance in the sense of cognitive apprenticeship to con-
nect new knowledge with existing structures. The role of the teacher in this process should
be a kind of learning counselor, not only somebody providing learning opportunities with-
out telling students how to use them.
The results also show a strong relationship between teachers’ perspectives on issues of
instruction and their own role definition. Therefore, developing teachers’ content-specific
pedagogical knowledge is necessary not only to foster their methodological competencies
but also to develop a new view of their role as teachers.
Teachers’ Views in a Nutshell
Seen from the constructivist conceptual change perspective underlying the present chap-
ter, teachers’ thinking about teaching and learning physics meets characteristics of the con-
structivist view only marginally. Most teachers are not well informed about recent views
of teaching and learning science, they do not hold explicit views of student learning, and
their ways of thinking about instruction is rather topic-oriented only. However, results also
show that a number of teachers implicitly hold views that are in accordance with con-
structivist ideas. And if they implicitly hold such views the achievement gains appear to be
better as compared to teachers who do not hold them.
Summary of Findings
Findings of the Whole Video-Study
The findings presented in this chapter are part of a larger set of studies within the first
phase of the IPN Videostudy Physics. It turned out (Duit et al., 2004; Prenzel et al., 2002;
Seidel et al., 2002) that the physics instruction of our 13 teachers is strongly teacher dom-
inated. The teachers prefer a script in which a straight path to a certain concept or to the
results of a certain experiment is used. Our data also show that there is substantial vari-
ance between the 13 teachers participating. Therefore, we observe 13 scripts that are indi-
vidual and characteristic of the single teachers. There is also a broad range of teaching
methods — although teacher-dominated methods prevail. We see, for instance, also teach-
ing and learning methods that allow students some freedom to follow their own ideas.
However, certain narrowness is to be seen in the scripts of most teachers. This becomes
apparent, for instance, in the dominating use of the questioning– developing approach.
Rather seldom does it happen that the dyadic dialogue between the teacher and students
is widened to a discourse including interactions between students. There are also several
examples in our data that the questioning–answering game turns out to be rather ineffi-
cient. This is, for instance, the case when teachers try to develop characteristics of physics
concepts. Often teachers ask questions students cannot answer at that time. Hence, they
start a guessing game.
The teachers’ views about “good” physics teaching and learning as revealed by the
teacher interviews also show a rich repertoire of thinking patterns about instruction, on the
one hand, and a certain narrowness, on the other. Many teachers hold elaborate ideas about
their way of teaching. However, considerations about the content in question predominate
planning. Reflections about students’ perspectives and their role in the learning process
play a comparably minor role. It turned out that most teachers do not hold elaborate views
about teaching and learning. Further, most teachers are not informed about research find-
ings on teaching and learning physics. Briefly summarized, two general orientations of
instruction may be distinguished:
1. Instructional: Oriented to physics; focus on physics concepts; learning viewed as

knowledge transmission.
2. Constructivist: Focus is on student learning, in particular on the conditions necessary
to support learning; learning viewed as student construction.
The instructional orientation predominates teaching behavior and teachers’ beliefs.

There is a large gap between the kind of thinking about efficient teaching and learning
physics, as discussed in the research-based literature, and the thinking of the teachers in
our study and their instructional practices.
Findings Concerning Constructivist-oriented Science Instruction
As presented in the previous sections, the instructional patterns observed meet patterns
characteristic of constructivist-oriented science instruction only to a limited extent. Most
teachers’ views of teaching and learning science are also quite far from constructivist
ideas. Furthermore, most teachers are not informed about publications addressing
constructivist views. This also holds for publications written for teachers in teachers’
journals. It is even true for publications in which the role of pre-instructional (alternative)
conceptions (such as conceptions on the electric circuit or the force concept) are dis-
cussed — although a substantial number of such publications appeared since the 1980s in
Germany. If instructional patterns and teachers’ views to a certain extent meet construc-
tivist ideas, achievement seems to be better. Two major facets are included in what is
called constructivist here. First, learning should be initiated with students’ conceptions
that might be in contrast to the physics views and need to be developed as described in
the constructivist literature on conceptual change. Second, it is not sufficient to provide
learning opportunities but learning has to be continuously supported, for example, by
means of cognitive activation.
Significance of the Findings
Owing to the small and particularly composed sample, the results presented here may only
be viewed as preliminary hypotheses. In the second phase of the IPN video study we will
investigate whether they also hold for the larger and somewhat more representative sam-
ple of 50 teachers. A video study in Switzerland which is a sister project of the IPN study
(Labudde, Gerber, & Knierim, 2003) provides an additional chance to further develop the
hypotheses of the first phase.
Concerning research methods, the first phase has confirmed that video analysis is a
rather powerful means to investigate instructional practice. This is especially true for a
design that is not restricted to just documenting videos (which was the case for the
TIMS Video Study in mathematics; Stigler et al., 1999). The set of student and teacher
questionnaires as well as interviews used, on the one hand, and coding the video-doc-
umented lessons on the other, have provided deep insights into the physics classroom
that go far beyond results that may be gained by the single research methods (cf. Roth
et al., 2001). The interplay of quantitative and qualitative interpretation modes has
proven rather valuable.
Conclusions
Our studies have shown that teachers’ views about their students’ learning are limited —
if the current state of research in teaching and learning is taken as reference. It further
became apparent that teachers’ thinking about instruction is predominantly oriented to the
content in question. It appears that teacher professional development needs to address
these two issues. It seems that teachers’ epistemological beliefs and their thinking about
instruction, on the one hand, and their instructional practice, on the other, have to be devel-
oped “in parallel.” Recent approaches of teacher professional development take this into
account. West and Staub (2003), for instance, developed an approach called “content-
focused coaching.” They attempt to make teachers familiar (a) with the recent state of
research knowledge about teaching and learning and (b) with the view of a fundamental
interplay of all variables of instructional planning (content, aims, teaching methods, and
media as well as students’ views). In a number of studies these changes of teachers’ epis-
temological beliefs and views are explicitly modeled as conceptual changes (Beeth &
Hewson, 1999; Duit & Treagust, 2003).
In the third phase of the IPN Video Study we will investigate in which way instructional
videos may support guiding teachers from their views and practices towards overcoming
the above limitation of their primarily content-oriented thinking. In a project to improve
physics instruction in Germany (Physics in Context; Duit, Euler, Friege, Komorek, &
Mikelskis-Seifert, 2003) development of teachers’ thinking about instruction is a key issue.
A final remark concerns the significance of research on teaching and learning on
improving practice. The state of theory building on conceptual change has become more
and more sophisticated and the teaching and learning strategies developed have become
more and more complex during the past 25 years. These developments are, of course,
necessary in order to address the complex phenomena of teaching and learning (science)
more and more adequately. However, the gap between what is necessary from the
researcher perspective and what may be set into practice by “normal” teachers has also
increased more and more. In other words, there is the paradox that in order to adequately
address teaching and learning processes, research alienates the teachers and hence
widens the “theory–practice” gap. The major message of the present chapter is that it is
necessary to close the gap between theory and practice at least to a certain extent. What
research on conceptual change has to offer classroom practice is not set into actual prac-
tice to a substantial extent. Of course, teacher development programs are essential in
order to change teachers’ views of teaching and learning and their practice. However, it
also appears to be necessary to “simplify” instructional theories and conceptual change
strategies in such a way that they may become part of teachers’ routines.
Acknowledgments
Ari Widodo is a lecturer at the Indonesian University of Education in Bandung. His Ph.D.
studies at the IPN were supported by a scholarship of DAAD (German Academic
Exchange Service). Christoph T. Wodzinski’s work was made possible by a grant from
DFG — Deutsche Forschungsgemeinschaft (German Science Foundation). We are also
grateful for the contributions of the other team members of the IPN Video Study Physics
in which the work presented here is embedded (see footnote 1 above).
References
Aldridge, J. M., Fraser, B. J., Taylor, P. C., & Chen, C. (2000). Constructivist learning environments in
a cross-national study in Taiwan and Australia. International Journal of Science Education, 22, 37–55.
Anderson, R. D., & Helms, J. V. (2001). The ideal of standards and the reality of schools: Needed
research. Journal of Research in Science Teaching, 38, 3–16.
Beeth, M. E., & Hewson, P. W. (1999). Learning goals in an exemplary science teacher’s practice:
Cognitive and social factors in teaching for conceptual change. Science Education, 83(6), 738–760.
Beeth, M., Duit, R., Prenzel, C., Ostermeier, R., Tytler, R., & Wickman, P. O. (2003). Quality devel-
opment projects in science education. In: D. Psillos, P. Kariotoglou, V. Tselfes, E. Hatzikraniotis,
G. Fassoulopoulos, & M. Kallery (Eds), Science education research in the knowledge based soci-
ety (pp. 447–457). Dordrecht, The Netherlands: Kluwer.
Biemans, H. J. A., & Simons, P. R.-J. (1999). Computer-assisted instructional strategies for promot-
ing conceptual change. In: W. Schnotz, S. Vosniadou, & M. Carretero, (Eds), New perspectives on
conceptual change (pp. 247–262). Amsterdam: Pergamon.
Biemans, H. J. A., Deel, O. R., & Simons, P. R.-J. (2001). Differences between successful and less
successful students while working with the CONTACT-2 strategy. Learning and Instruction, 11,
265–282.
Cosgrove, M., & Osborne, R. (1985). Lesson frameworks for changing children’s ideas. In:
P. Freyberg (Ed.), The implications of children’s science (pp. 101–111). Auckland: Heinemann.
Ditton, H. (2002). Unterrichtsqualität — Konzeptionen, methodische Überlegungen und
Perspektiven [Teaching quality — Conceptions, considerations on methods and perspectives].
Unterrichtswissenschaft, 30, 197–212.
Driver, R. (1989). Changing conceptions. In: P. Adey, J. Bliss, J. Head, & M. Shayer (Eds),
Adolescent development and school science (pp. 79–104). London: The Falmer Press.
Duit, R. (1999). Conceptual change approaches in science education. In: W. Schnotz, S. Vosniadou, &
M. Carretero (Eds), New perspectives on conceptual change (pp. 263–282). Amsterdam: Pergamon.
Duit, R., & Treagust, D. F. (1998). Learning in science: From behaviourism towards social con-
structivism and beyond. In: B. J. Fraser, & K. G. Tobin (Eds), International handbook of science
education (pp. 3–26). Dordrecht, The Netherlands: Kluwer.
Duit, R., & Treagust, D. F. (2003). Conceptual change: A powerful framework for improving science
teaching and learning. International Journal of Science Education, 25, 671–688.
Duit, R., Euler, M., Friege, G., Komorek, M., & Mikelskis-Seifert, S. (2003). Physik im Kontext: Ein
Programm zur Verbesserung der naturwissenschaftlichen Grundbildung durch Physikunterricht
[Physics in context — A program to improve scientific literacy in physics instruction].
Occasioanal Paper. Kiel, Germany: IPN — Leibniz Institute for Science Education.
Duit, R., Müller, C. T., Tesch, M., & Widodo, A. (2004). A video study on the practice of German
physics instruction. In: D. Zandvliet (Ed.), NARST Conference 2004, Conference Proceedings
(CD Rom). Vancouver: NARST — National Association for Research in Science Teaching.
Duschl, R., & Gitomer, D. (1991). Epistemological perspectives on conceptual change: Implications
for educational practice. Journal of Research in Science Teaching, 28, 839–858.
Fraser, B. (1998). Science learning environments: Assessment, effects and determinants. In: B. J. Fraser,
& K. Tobin (Eds), International handbook of science education (pp. 527–564). Dordrecht, The
Gallagher, J., & Parker, J. (1995). Secondary teacher analysis matrix: Science version, Lansing, MI:
Department of Teacher Education, Michigan State University.
Hanrahan, M. (1998). The effect of learning environment factors on students’ motivation and learn-
ing. International Journal of Science Education, 20(6), 737–753.
Hewson, P. W., Tabachnick, B. R., Zeichner, K. M., & Lemberger, J. (1999). Educating prospective
teachers of biology: Findings, limitations, and recommendations. Science Education, 83, 373–384.
Hodson, D. (1993). Re-thinking old ways: Towards a more critical approach to practical work in
school science. Studies in Science Education, 22, 85–142.
Labudde, P. (2000). Konstruktivismus im Physikunterricht der Sekundarstufe II [Constructivism in
upper secondary physics instruction]. Bern, Switzerland: Paul Haupt Verlag.
Labudde, P., Gerber, B., & Knierim, B. (2003, August). Integrated science in an constructivist ori-
ented approach: Between vision and reality. Paper presented at the Biannual Conference of
ESERA (European Science Education Research Association). Noordwijkerhout, The Netherlands.
Lawson, A. E., Abraham, M. R., & Renner, J. W. (1989). A theory of instruction: Using the learning
cycle to teach science concepts and thinking skills. NARST Monograph Number One. Cincinatti,
OH: University of Cincinatti, National Association for Research in Science Teaching.
Mason, L. (2001). Responses to anomalous data on controversial topics and theory change. Learning
Matthews, M. R. (2000). Appraising constructivism in science and mathematics. In: D. C. Phillips
(Ed.), Constructivism in education (pp. 161–192). Chicago, IL: The National Society for the Study
of Education.
Mayring, P. (2000). Qualitative Inhaltsanalyse — Grundlagen und Techniken [Qualitative content
analysis — Rationale and techniques]. Weinheim, Germany: Deutscher Studien-Verlag.
McComas, W. F. (Ed.). (1998). The nature of science in science education: Rationales and strate-
gies. Dordrecht, The Netherlands: Kluwer.
Müller, C. T. (2004). Subjektive Theorien und handlungsleitende Kognitionen von Lehrern als
Determinanten schulischer Lehr-Lern-Prozesse im Physikunterricht [Teachers’ Subjective
Theories and Cognitions and Teaching and Learning Processes in Physics Instruction]. Studien
zum Physikunterricht, Band 33. Berlin: Logos.
Nunez-Oviedo, M. C., Clement, J., & Rea-Ramirez, M. A. (2002, April). An instructional model
derived from model construction and criticism theory. Paper presented at the National Association
for Research in Science Teaching, New Orleans.
OECD – Organization for Economic Cooperation and Development. (1999). Measuring student
knowledge and skills — A new framework for assessment. Paris: OECD Publications.
Phillips, D. C. (Ed.). (2000). Constructivism in education: Opinions and second opinions on
controversial issues. Chicago, IL: The University of Chicago Press.
Prenzel, M., Duit, R., Euler, M., Lehrke, M., & Seidel, T. (2001). Erhebungs- und Auswertungsverfahren
des DFG-Projekts, Lehr-Lernprozesse im Physikunterricht — eine Videostudie [Research Methods of
the IPN Physics Videostudy]. Kiel, Germany: IPN — Leibniz Institute for Science Education.
Prenzel, M., Seidel, T., Lehrke, M., Rimmele, R., Duit, R., Euler, M., Geiser, H., Hoffmann, L.,
Müller, C., & Widodo, A. (2002). Lehr-Lern-Prozesse im Physikunterricht — eine Videostudie
[Teaching and learning processes in physics instruction — a video study]. Zeitschrift für
Pädagogik, 48(45. Beiheft), 139–156.
Rimmele, R. (2003). Videograph: Multimedia-player for video analysis. Kiel, Germany: IPN —
Leibniz Institute for Science Education (available at: http://www.ipn.uni-kiel.de/projekte/video/
videostu.htm).
Roth, K. J., Druker, S., Kawanaka, T., Okamoto, Y., Trubacova, D., Warvi, D., Rasmussen, D., &
Gallimore, R. (2001, March). Uses of video-based technology and conceptual tools in research:
The case of the TIMSS-R Video Study’. Paper presented at the Annual Meeting of the National
Association for Research in Science Teaching (NARST), St. Louis, MO.
Schnotz, W., Vosniadou, S., & Carretero, M. (Eds). (1999). New perspectives on conceptual change.
Amsterdam: Pergamon.
Seidel, T., Prenzel, M., Duit, R., Euler, M., Geiser, H., Hoffmann, L., Lehrke, M., Müller, C. T., &
Rimmele, R. (2002). Jetzt bitte alle nach vorne schauen! — Lehr-Lernskripts im Physikunterricht
und damit verbundene Bedingungen für individuelle Lernprozesse [Teaching and learning scripts
in physics instruction and conditions for individual learning processes]. Unterrichtswissenschaft,
30(1), 52–77.
Sinatra, G. M., & Pintrich, P. R. (Eds). (2003). Intentional conceptual change. Mahwah, NY:
Erlbaum.
Stigler, J. W., Gonzales, P., Kawanaka, T., Knoll, S., & Serrano, S. (1999). The TIMSS. videotape
classroom study: Methods and findings from an exploratory research project on eight-grade math-
ematics instruction in Germany, Japan and the United States. U.S. Department of Education,
National Centre for Education Statistics (199NCES 99-074). Washington DC: U.S. Government
Printing Office (http//nces.ed.gov/timss).
Taylor, P. C., & Fraser, B. J. (1991, April). CLES: An instrument for assessing constructivist learn-
ing environments. Paper presented at the Annual Meeting of the National Association for Research
in Science Teaching (NARST), Wisconsin.
Taylor, P. C., Fraser, B. J., & Fisher, D. L. (1997). Monitoring constructivist classroom learning envi-
ronments. International Journal of Educational Research, 27, 293–302.
Tenenbaum, G., Naidu, S., Jegede, O., & Austin, J. (2001). Constructivist pedagogy in conventional
on campus and distance learning practice: An exploratory investigation. Learning and Instruction,
11, 87–111.
Watts, M. (1994). Constructivism, re-constructivism and task oriented problem-solving. In:
P. J. Fensham, R. F. Gunstone, & R. T. White (Eds), The content of science: A constructivist
approach to its teaching and learning (pp. 39–58). London: The Falmer Press.
West, L., & Staub, F. C. (2003). Content-focused coaching: Transforming mathematics lesson.
Portsmouth, NH: Heinemann/Pittsburgh, PA: University of Pittsburgh.
Widodo, A. (2004). Constructivist oriented science classrooms: The learning environment and the
teaching and learning process. Frankfurt A.M., Germany: Peter Lang.
COMMENTARY
Chapter 15
First Steps: Scholars’ Promising Movements

Into a Nascent Field of Inquiry
Patricia A. Alexander and Gale M. Sinatra
It is likely that most readers of this volume have had the touching experience of watching
infants as they first struggle to walk unaided. They teeter on uncertain feet and legs and
move ever so hesitantly, but with dogged determination. The frequent stumbles and falls,
which are part and parcel of this experience, can prove quite frustrating to the children and
worrisome to those bearing witness.
Yet, these setbacks or impediments are only temporary and are critical lessons in bal-
ance and mobility for the children. Before they can conquer the art of walking, young chil-
dren must come to understand what their bodies can and cannot do and how those bodies
must work in concert with their physical surroundings. Those first steps are truly develop-
mental milestones that ultimately signal a new-found freedom for toddlers. Once bound to
a limited physical space or dependent on others for mobility, young children find them-
selves able to step out into a world that had been virtually inaccessible to them.
As we read the collection of chapters in this section, we appreciated that we were also
bearing witness to first steps — those uncertain, tentative, but inspiring initial movements
that are a milestone to subsequent development. However, we do not see ourselves in the
role of doting family members who have already mastered this physical feat and are
merely watching the hesitant and unsophisticated movements of those less mature than we.
On the contrary, as we share our reflections and recommendations on the writings by
Entwistle, Murphy, Alexander, Greene, and Edwards, Mason and Gava, Duit, Widodo, and
Wodzinski, and Stathopoulou and Vosniadou (this volume), we see ourselves as literally on
comparable footing. Because the nature and boundaries between epistemology, conceptual
change, and learning are just coming into focus, we are, in reality, partners in this devel-
opment and, thus, trying out our own theoretical and empirical legs. The advantage we
have in this commentary is that we have the luxury of looking across these diverse and
intriguing chapters, able to identify recurring patterns that warrant care and attention if
they are to foster further movement and that hint at the next steps that might be taken.
Specifically, we have organized this analysis into two sections. In the first section, we
outline five themes that emerged from the five chapters — themes that represent areas of
concern for those taking initial steps to define this relatively new field of inquiry. Three of

ISBN: 0-08-045355-4
222 Patricia A. Alexander and Gale M. Sinatra
those themes deal with the clarity, the boundaries and contexts, and the complexity of the
constructs under investigation. What do we mean by epistemological beliefs or conceptual
change, for example, and when do the studies dealing with epistemology actually cross
into the territory associated with ontology? Moreover, how are researchers to be true to the
complex nature of epistemology, conceptual change, or learning without being over-
whelmed by that complexity?
We also explore questions of development and interrelatedness of epistemology, con-
ceptual change, and learning. Here the concern is change over time both within and across
the literatures on epistemology and conceptual change. For instance, how can researchers
address the changing sophistication of individuals’ knowledge and beliefs over time,
especially as they grow in expertise within academic domains? Further, to what degree do
transformations in one of these areas (e.g., epistemic beliefs) contribute to developmen-
tal shifts in the others (e.g., conceptual change)? We will also examine if current devel-
opmental notions, as represented in these chapters, can be judged as indicative of
universal transformations, rather than reflections of Western cultures or postindustrial
societies?
Finally, beyond the questions of how epistemology, conceptual change, and learning are
related to each other, we want to explore the degree to which these phenomena are directly
or indirectly related to other educational and psychological factors, including learner moti-
vation and pedagogical practices. One question to be considered here is the degree to
which formal schooling or the classroom contexts shape students’ epistemic beliefs or their
conceptions or the degree to which students’ own personalities, goals, and motivations
become the defining forces in the formation of such beliefs and conceptions.
In the second section of this chapter, we revisit these five themes of conceptual clarity,
boundaries and contexts, complexity, development, and interrelatedness. However, the
goal in this section is to pose next steps. What can these scholars and others committed to
the nascent field of inquiry (ourselves included) do to ensure that we move more steadily
and assuredly into studies of the interplay of learning, epistemology, and conceptual
change? How can current researchers lay the appropriate groundwork so that others can
follow with greater ease and confidence?
Assessing the First Steps
It is never easy to conduct research on human learning and development. Humans are noto-
riously willful creatures and reluctant participants in psychological probings and prod-
dings. It is even more precarious to attempt studies that go beyond the study of directly
observable and tractable phenomena (e.g., salivating animals) and instead seek to infer
relations (to say nothing of causality) between constructs (e.g., knowledge and beliefs) that
only occasionally rise to the surface of consciousness. This is certainly the reality faced by
the brave authors of these intriguing and provocative chapters. Even as we laud Entwistle,
Murphy and colleagues, Mason and Gava, Duit and colleagues, and Stathopoulou and
Vosniadou (this volume) for their creative and fearless forays into difficult and less-charted
terrain, we want to provide some cautions or warnings that might help them from faltering
or stumbling in their subsequent investigations.
First Steps: Scholars’ Movements Into a Nascent Field of Inquiry 223
Conceptual Clarity
Readers of the aforementioned chapters will be confronted with an array of terminology

central to this emergent and interdisciplinary field of inquiry. Not surprisingly, that lexicon
includes words and phrases about epistemology (e.g., epistemological beliefs, epistemic
beliefs, and personal epistemology) and conceptual change (e.g., concepts, conceptual
knowledge, and knowledge restructuring), as well as learning (e.g., constructivism). As we
have stated, researchers like the scholars contributing to this volume are already pursuing
rather elusive constructs that are notoriously hard to uncover and validate. It would seem
advisable, therefore, for researchers exploring such constructs as knowledge, beliefs, and
learning to take great care to establish the conceptual and operational parameters for their
key constructs. Regrettably, this is not always the case within this growing literature.
Rather, the literature exploring the relations between epistemology, conceptual change,
and learning is often plagued by terminology that is undefined, poorly specified, or vari-
ably defined. This is a broad concern that the authors of these chapters rightfully acknowl-
edge. In certain instances, this conceptual wobbliness seems attributable to the newness of
the research. For example apart from the work of Perry (1970), the psychological research
on epistemology is very much in its infancy, having really come into its own in the last
decade. Consequently, it is not surprising that researchers in this area are still seeking solid
ground with regard to labels and definitions they use.
These conceptual growing pains are quite apparent in discussions over what exactly
researchers are studying when they ask questions about the knowledge and knowing. In
this volume, both Mason and Gava and Stathopoulou and Vosniadou use the terms episte-
mological beliefs and personal epistemology to signify the type of beliefs they are exam-
ining. Even while selecting these particular labels, Stathopoulou and Vosniadou
acknowledge the conceptual unsteadiness they are confronting:
There is a diversity of theoretical and conceptual approaches to the con-

struct of ‘epistemological beliefs’. This is reflected by the fact that the con-
struct is assigned different terms in the literature such as, personal
epistemology, epistemic beliefs, ways of knowing, epistemological per-
spectives, epistemological reflection epistemological thinking, epistemo-
logical theories, epistemological resources etc. This diversity indicates that
what we call ‘epistemological beliefs’ may not be the same in all relevant
studies, or, at least, that the boundaries of the construct may differ. (Hofer
& Pintrich, 1997; Pintrich, 2002) (this volume: p. 145)
Murphy et al. raise the same conceptual issue. Yet, their decision was to draw a sharp
distinction between epistemological beliefs and epistemic beliefs — their preferred term
— and to avoid the notion of personal epistemology entirely. As these authors stated:
For clarity we have taken pains to distinguish between epistemic and epis-
temological beliefs. The former are beliefs about knowledge, whereas the
latter are beliefs about the study of knowledge. Kitchener (2002) has made
a strong argument for maintaining this distinction. Students most likely
have epistemic beliefs, beliefs about knowledge. Many researchers have

epistemic beliefs as well, such as what means of justification are privileged
over others. Finally, those looking at the implicit or explicit assumptions
regarding the study of knowledge are engaging in research on epistemo-
logical beliefs. (this volume: p. 105–106)
Although such conceptual wrangling may be worrisome, we feel that, in the case of
epistemological versus epistemic versus personal epistemological beliefs, it bodes well for
this emergent field. Specifically, it suggests that this literature is maturing to the point of
dealing with complicated but fundamental quandaries that have previously been over-
looked or ignored. Whether this conceptual maturation continues or leads to greater con-
sistency with this psychological research in knowledge and knowing remains to be seen.
In direct contrast to the epistemological/epistemic beliefs debate, some of this concep-
tual teetering may reflect overconfidence or undue comfort with often-used words within
more mature lines of inquiry. Entwistle (this volume) makes this point when he offers an
explicit definition of concepts — a word for which a definition is often assumed.
‘Concept’ is most frequently used to describe a grouping of objects or

behaviours with the same defining features that has become recognised
through research or widespread usage. The same term has also been used in
the literature on conceptual change to indicate individuals’ different ways
of thinking about a particular grouping. Here, for clarity, these individual
variations are referred to as ‘conceptions’. (this volume: p. 124)
Nowhere is this problem of definition through assumption more apparent than in

researchers’ treatment of the three words most at the heart of this program of research
— knowledge, beliefs, and learning. Researchers like Entwistle and Mason and Gava
(this volume) convey a sensitivity to the labels applied to psychological studies of indi-
viduals’ perceptions of knowledge and beliefs. Nonetheless, as the chapter by Murphy
et al. indicates, researchers may be slow to reveal the definitions of knowledge and
beliefs that guide the design and interpretation of their investigations, even if such defi-
nitions are well articulated in their minds. We see the conceptual definition of these con-
structs, along with an explication of learning, to be important to anchor subsequent
investigations, especially given that scholars in this realm are expressly engaged in the
study of epistemic beliefs.
One problematic term, which is at the heart of the study by Duit et al. (this volume), is
constructivism. Despite the fact that Duit et al. portray the notion of constructivism and con-
structivist teaching in rather tidy language, it is far from a tidy construct. As Bereiter (1994),
Phillips (1995) and others have argued, there are many faces to constructivism. Moreover, a
constructivistic orientation toward learning (i.e., knowledge is constructed) should not be
confused with attempts to convert such a learning theory into pedagogical practice. In
essence, if all knowledge is, in actuality, constructed, then the mode of instructional delivery
would be inconsequential. Whether teachers lecture or engage students in active learning,
those students would still form their knowledge via construction. In addition, the effective-
ness of constructivist pedagogy has recently been called into question (Kirschner, Sweller, &
Clark, 2006), further complicating the issue. Thus, adding constructivism to the discussion
of epistemological beliefs and conceptual change may only serve to exacerbate the problem
of conceptual clarity in this nascent literature.
Boundaries and Contexts
Perhaps the unsteadiness we perceive in the terms that mark this emergent field of inquiry
is indicative of an even deeper issue; the nature of the constructs under study and the con-
ceptual boundaries between them. In effect, it is one thing to differentially label the same
construct, but it is another to use the same label to signify varied constructs. Within these
five informative chapters, for example, we see scholars struggling to transfer particular
philosophical constructs (e.g., epistemology) into the realms of psychology or pedagogy.
In so doing, these researchers must confront the possibility that they are now engaged in
the study of fundamentally altered concepts (Alexander, 2006). Part of the struggle in ter-
minology use captured in the epistemological versus epistemic beliefs debate reflects this
awareness. However, what researchers in this realm may be overlooking is that they have
not only stepped across the philosophical threshold into psychology, but may have also
transitioned from epistemology to ontology.
Specifically, when we look at the questions researchers are asking about knowledge,
many of them move beyond questions about the source or justification of knowledge and
more directly target questions about “what is knowledge.” When “nature” is the central
concern, researchers may have crossed the boundary from epistemology (“what is
known”) into ontology (“what is”). This is an argument recently outlined by Murphy et al.:
We would suggest, however, that the psychological literature might benefit

from the stricter definitions of philosophy, separating epistemological
concepts, such as justification, from more ontological constructs, such as
the relations among objects of knowledge or characteristics of knowledge
(e.g., certainty or simplicity). It would also help coordinate the two
disciplines to emphasize the personal aspects of epistemology in the
psychological literature, to clearly identify the differences between
the two fields. (this volume: p. 119)
We concur with the authors of these five chapters that it is important to ascertain what
students, teachers, and researchers believe knowledge and concepts to be. However, as
with Murphy et al. (this volume) and others (Alexander, 2006), we feel it is essential to
accurately mark this as more ontological than epistemological territory. Until we have a
more definitive sense of what researchers and their study participants mean by knowledge,
knowing, or beliefs, then it is more difficult to interpret their responses to questions about
the source, certainty, or justification of knowledge or malleability of beliefs (Alexander &
Dochy, 1995; Sinatra & Kardash, 2004).
In a way, we see this association between ontology and epistemology as analogous to
questions of validity and reliability in educational measurement. Researchers may well be
able to establish that their factor structures or categorizations of questionnaire or interview
data are reliable, as the current researchers have done. However, until these scholars can
demonstrate that the measurements are also psychometrically (i.e., ontologically) valid,
results remain suspect. Thus, until those of us committed to the investigation of epistemo-
logical beliefs, conceptual change, and learning have dealt more directly and clearly with
ontological and epistemological boundaries can we have confidence in the insights and
recommendations forwarded.
This overarching concern over crossing conceptual boundaries is not restricted solely to
the broad epistemological and ontological terrains, but it pertains as well to the nature of
knowledge in general, as opposed to the nature of scientific knowledge in particular.
Notably, in the Mason and Gava, Duit et al., and Stathopoulou and Vosniadou chapters
(and indirectly in the Murphy et al. chapter), researchers are primarily focused on individ-
uals’ knowledge and beliefs within the domain of science. Mason and Gava (this volume)
ask students to process refutational texts about natural selection and biological evolution
and Duit et al. (this volume) study constructivist science teaching, while participants in
Stathopoulou and Vosniadou (this volume) are working within the domain of physics.
Consequently, it is important to ascertain whether these scholars are exploring the nature,
sources, certainty, or justification of knowledge per se or the nature of science or beliefs
about its associated knowledge.
As Sinatra and colleagues have argued (Sinatra, Southerland, McConaughy, &
Demastes, 2003; Southerland, Sinatra, & Mathews, 2001), there is a fine but critical dis-
tinction between views on the nature of science and epistemic beliefs relative to this
domain and general beliefs about knowledge and knowing. Specifically, those holding to
the domain-specific perspective on epistemological beliefs would contend that individu-
als may view scientific knowledge differently than knowledge associated with other
domains (e.g., history or reading). Moreover, tensions between the physical world, which
is at the heart of science, and the nonphysical or metaphysical realms aligned with other
domains (e.g., history or religion, respectively) could well drive different world or onto-
logical views that translate into varied judgments about the source or justification of
knowledge.
What these issues related to conceptual boundaries suggest is that those studying the
nature and interplay of epistemology, conceptual change, and learning must be more defin-
itive about the territory in which they are presently moving if subsequent development is
to be ensured. For example, how does one talk about the major reconstruction of scientific
understandings (Duit et al., this volume), if one holds to the belief that all such knowledge
is personally constructed? How does one reconcile personally constructed knowledge with
the “scientific body of knowledge” or call for the learning of “facts, laws, and theories,” as
Duit et al. have (this volume: p. 199) without sounding oxymoronic.
Complexity Clarity and contextualizing are just two of the themes to arise from the chap-
ters by Entwistle, Murphy et al., Mason and Gava, Duit et al., and Stathopoulou and
Vosniadou. One of the most paradoxical themes pertains to the fundamental complexity of
the core constructs under study. In essence, what these initial explorations have made
apparent is that each of the focal constructs under investigation — epistemic beliefs, con-
ceptual change, and learning — is subtle in form, intangible in nature, and empirically
complicated. The difficulty experienced in constructing a valid and reliable measure of
epistemic beliefs is just one indicator of this problem.
Indeed, the literature is replete with studies that involve the creation or validation of a
questionnaire or interview protocol about epistemic beliefs (or epistemological beliefs or
personal epistemology; Buehl & Alexander, 2005; Hofer & Pintrich, 2002; King &
Kitchener, 1994; Schommer, 1990; Schraw, Bendixen, & Dunkle, 2002; Wood & Kardash,
2002). In the current volume, Mason and Gava opted to measure students’ general beliefs
about knowledge and knowing by using a “reduced and adapted” version of Schommer’s
(1990) often-used Epistemological Questionnaire. Duit and colleagues (this volume) relied
upon the Constructivist-Oriented Science Classrooms (COSC) instrument to categorize
teachers’ instructional behaviors. By comparison, Stathopoulou and Vosniadou (this
volume) used a cultural- and domain-specific measure of epistemic beliefs, the Greek
Epistemological Beliefs Evaluation Instrument for Physics (GEBEP). What these two
studies demonstrate is that variability in the use of domain-general or domain-specific
tools exists, and efforts to craft psychometrically sound measures continue.
But measurement of epistemic beliefs is only one small piece of the complex puzzle in
the literature that considers such beliefs in relation to conceptual change or learning. Here
beliefs about knowledge and knowing must be assessed in relation to other equally com-
plex constructs, most notably conceptual change, as well as characteristics of respondents.
Even though the issue of measure interdependence is not often considered, it seems feasi-
ble to assume that such interdependence warrants reflection. That is to say, there is an
understanding that items appearing on the same questionnaire or test do not function com-
pletely independently. The content or focus of one item can affect thoughts and responses
to subsequent items. Similarly, it is reasonable to assume that the researchers’ decisions to
incorporate multiple measurements tools (e.g., epistemology questionnaire, prior knowl-
edge test, or refutational texts) in the same study can have a cumulative effect not evi-
denced when a single empirical test is used.
There is also the continuing question for researchers as to how to disentangle poten-
tially significant learner characteristics within the study of epistemological or epistemic
beliefs. Trying to understand the relation between school experiences and students’ beliefs
about knowledge and knowing (Entwistle, this volume) or between researchers’ mental
models and the way they approach the study of conceptual change (Murphy et al., this
volume) has the characteristics of the classic chicken–egg problem. What came first and
what affected what? In this fledgling domain of research, we are understandably far from
answering that elemental question.
Development The unsure footing researchers face on their nascent forays into this con-
ceptual terrain is made more difficult by the fact that the road they travel is itself under
construction. That is, the constructs under investigation — epistemic beliefs, conceptual
change, knowledge, and learning — are themselves ever developing in learners. Still more
challenging, these constructs likely develop in concert with one another, affecting how
each develops as the learner matures.
As an example, research suggests that students who view knowledge as fixed and
unchanging are less likely to change their perspective than those who view knowledge as
evolving (Qian & Alvermann, 1995, 2000, Windschitl & Andre, 1998). Developmentally
(through both maturation and educational experience) students become increasingly aware
of their epistemic beliefs. As students acquire the capabilities to be reflective about their
knowledge and beliefs, and teachers and activities bring students’ views to the foreground,
students are afforded the opportunity to expand or restructure their knowledge.
Entwistle describes how Vosniadou (1999) and others have portrayed this interactive
developmental process whereby with
age and with expertise, [learners acquire] a more flexible system, a system
that makes it easier to take different perspectives and different points of
view. (Vosniadou, 1999, pp. 11–12) (this volume: p. 125)
Similarly, Mason and Gava (this volume) illustrate the effects of students’ emerging
metaconceptual awareness. It is this greater developed facility for metaconceptual aware-
ness that affords epistemic belief change, which further bootstraps conceptual knowledge
development. Evidence for this effect can be seen in the Mason and Gava chapter.
Specifically, these authors found that students who viewed knowledge as constructed,
showed greater metacognitive awareness, which likely contributed to their greater degree
of conceptual change.
The view of conceptual development portrayed in these chapters, particularly Entwistle’s
analysis, reflects the complex developmental interplay among these constructs and also
echoes the view of knowledge development portrayed in the Model of Domain Learning or
MDL (Alexander, 1997, 2005; Alexander, Sperl, Buehl, Fives, & Chiu, 2004). The MDL is
a useful framework for understanding how the development of knowledge/expertise interacts
with the development of beliefs about knowledge. As one acquires a richer knowledge base,
and moves toward expertise in a domain, one becomes able to see the complexity of ideas
and understand multiple sides of an issue. This appreciation for the complex and evolving
nature of knowledge promotes opportunities for epistemic change (Sinatra, 2005).
The MDL also emphasizes that development occurs within a domain of knowledge. As
we see in the Duit et al. and Stanthopoulou and Vosniadou chapters (this volume), charac-
teristics of the domain affect how knowledge within that domain develops. For example,
according to Stanthopoulou and Vosniadou’s findings, infants come prepared to acquire
knowledge about the physical world. Physics knowledge is further developed and refined
through experience with everyday objects and events. Thus, beliefs about the nature of
knowledge in the domain of physics are likely strongly held and deeply entrenched.
Overall, these findings highlight the need for researchers to take steps toward understand-
ing knowledge and belief development within different domains of learning.
As we discuss the development of epistemic beliefs, we purposefully avoid the charac-
terization of development as moving from a “naïve” or absolutist view toward a more
“sophisticated” view of knowledge as constructed and interpreted. Stanthopoulou and
Vosniadou (this volume) take up the issue of what constitutes a sophisticated personal epis-
temology, preferring instead the nomenclature of constructivist epistemology.
Some have argued that epistemic development does proceed toward a view of knowledge
as constructed and interpretive (Bendixen & Rule, 2004; Duit et al., this volume). However,
evidence is emerging which suggests that such a view of knowledge may be more reflective
of a Western or academic orientation toward knowledge and knowing. For instance,
Karabenick and Moosa (2005) recently demonstrated that a group of Middle Eastern
(Omani) college students were more likely to believe in simple and certain knowledge and
to regard authorities as the source of knowledge than were their U.S. counterparts. In addi-
tion, Weinstock (2005), in a study of ethnic and gender differences in epistemological under-
standing demonstrated that 7th and 9th grade Bedouin students (particularly boys) were more
likely to be absolutist in orientation than Jewish students of the same grade level, suggesting
epistemological beliefs are both developmental and culturally embedded.
Certainly, it could be argued that the polemics of radio talk shows and 24-hour news
broadcasts in the United States reflect a view of complex issues of politics and morality as
dualistic and simplistic if not dogmatic and absolutist. Thus, the valuing of complex and
nuanced views of knowledge may be unique to those in the academy who may not share
the beliefs about knowledge and knowing prevalent within the broader cultural community
to which they belong. This observation harkens back to Perry’s (1970) conclusions drawn
from the study of Harvard undergraduates.
Interrelatedness
Yet, continued development relative to learning, knowledge, and conceptual change entails
more than simple step-by-step progress in these individual areas. As the first steps in
human locomotion is just a prelude to an array of complex physical routines, so too should
the independent developments in learning, knowledge, and conceptual change foreshadow
coordinated and choreographed study of these particular constructs in conjunction with
significant learner and social/contextual variables. That is, as we come to understand how
epistemic beliefs interact with other learner characteristics in specific situational and
instructional contexts, we can begin to design learning environments and instructional
strategies to support conceptual development/change.
Entwistle (this volume) explains that we are attempting to describe a very complex sys-
tem. Like all complex systems, this involves identifying the important components and
describing how those components interact and affect the overall functioning of the system.
However, as Entwistle points out, there is one important distinguishing aspect of the sys-
tem we hope to describe — the knowledge construction system. That is, it is an intentional
system. This means that while we are trying to understand the interrelatedness of the con-
structs in the learning system, the learner has the capacity to compound this complexity.
Several of the chapters in this section describe this capacity of the learner as engaging
in “metalearning” or “deep learning” (Duit et al., this volume; Entwistle, this volume,
Stanthopoulou & Vosniadou, this volume). For instance, Entwistle explains that learners
are engaging in “metalearning” when they
…think about learning in a more sophisticated way [and] recognize a vari-

ety of different learning processes, and understand that effective learning
depends on finding the most appropriate ways of tackling a particular task
for a specific purpose in a given situation, in a given context. This awareness
can also lead to a more active engagement with learning as students monitor
their process towards understanding. (this volume: p. 130)
The awareness, reflectivity, and self-regulated aspects of learning portrayed in these

chapters is characteristic of intentional learners (Bereiter & Scardamalia, 1989) who are
capable of changing their knowledge and beliefs through the successful orchestration of
the learning situation. Sinatra and Pintrich (2003, p. 6) described this process as intentional
conceptual change or the “goal-directed and conscious initiation and regulation of cogni-
tive, metacognitive, and motivational processes to bring about a change in knowledge.”
As described in these chapters, intentional, reflective learners have many choices to
make when confronted with a new conception. They have the ability to choose which
learning strategies to employ, which knowledge and beliefs to bring to bear, which learn-
ing goals to apply, and ultimately which conception to adopt. When multiple options for
learning are available to the student, Seigler (1996) has argued that learners make adaptive
choices among those options resulting in the more successful strategies prevailing.
Metalearning (i.e., deep learning or intentional learning) involves the opportunity to
make many choices with important consequences for the learner. Learners may make a
controlled intentional analysis of the new conception and chose whether it has adaptive
utility for them. This suggests that, ultimately, it is the learner who must manage the inter-
connections among the components in the complex learning system. Therefore, instruc-
tional strategies that afford students the opportunity to make productive choices are more
likely to facilitate an orchestration of the components of the learning process that results
in knowledge restructuring and long-lasting conceptual change.
Particular pedagogical practices and instructional techniques such as refutational texts
(see Mason & Gava, this volume), as well as argumentation, discussion, and experimenta-
tion, are more likely to give students the opportunity to make choices that support epis-
temic belief and knowledge change (Alexander, Fives Buehl, & Mulhern, 2002; Fives,
Alexander, & Buehl, 2001; Murphy & Alexander, 2004; Sinatra & Kardash, 2004). These
pedagogical approaches allow students to see both sides of an issue, weigh issues and
arguments, examine their own beliefs and knowledge, and actively engage in knowledge
construction and/or restructuring.
The choice to explore and manage the interconnectivity of these constructs is not the sole
purview of the learner. As Duit et al. (this volume), Entwistle (this volume), and Murphy
et al. (this volume) explain, the teacher and the researcher must also examine their views of
epistemic beliefs, conceptual change, and learning. These self-reflections are necessary if
teachers are to orchestrate the components of the learning process in the classroom to facil-
itate knowledge development in their students. They are also requisite for researchers — our-
selves included — so that we may take interconnectivity into account and design research
programs to facilitate the construction of our knowledge base in this domain.
Proposing Next Steps
From the outset of this commentary, we have extolled the intellectual insights and schol-
arly daring of the authors contributing to this section. We have expressed our appreciation
for the critical, groundbreaking steps taken by these contributing authors, even as they
teetered across relatively uncharted terrain at the juncture of epistemology, conceptual
change, and learning. Nonetheless, like proud parents participating in these empirical
developments, we want to see even more movement and are willing to provide whatever
coaxing, enticement, or support it might take to stimulate increased activity in this area.
Toward that end, we visit anew our five initial themes: conceptual clarity, construct bound-
aries, complexity, development, and interrelatedness. We consider several issues that cross
these themes and that seem to merit further analysis or investigation. We do not intend to
be comprehensive or exhaustive in this proposal. Rather, it is our hope that the possibili-
ties we forward might spark the thinking and actions of others who are similarly invested
in the relations between epistemic beliefs, conceptual change, and learning.
Introspection and Explication
The theme of conceptual clarity in particular suggests that subsequent movement in

research on epistemic beliefs, conceptual change, and learning is dependent upon increased
lexical and semantic specificity. As researchers frame their investigations, forge their empir-
ical tools, interact with participants, and interpret their data, they must take certain precau-
tionary steps. First, they should reflect on the meanings they are ascribing to their words,
including such basic concepts as knowledge, knowing, or beliefs. The chapter by Murphy
and colleagues reminds us that we, as educational researchers, may be far quicker to ana-
lyze the meanings that others attach to terms like knowledge or beliefs than we are to con-
sider our own epistemological or ontological frames. Such introspection cannot hurt and
may well give researchers a new perspective on these fundamental constructs and their
study of them.
Second, concurrent with the introspection, researchers should incorporate more explicit
definitions of underlying constructs and concepts in their writings and presentations. It has
been documented in the literature that researchers often fail to provide explicit definitions
and elect instead to define constructs implicitly or referentially (Alexander, Schallert, &
Hare, 1991; Murphy & Alexander, 2000). That is to say, they may explain a particular con-
struct (e.g., teacher quality or reading) by describing some of its outcomes, salient attrib-
utes, or underlying processes (e.g., higher student performance or phonological
awareness). Alternatively, authors may refer to key studies or leading researchers associ-
ated with the target constructs perhaps under the assumption that those references or
names will suffice as grounding for the central constructs.
However, we feel that the nascent, multidimensional field represented in these five
chapters is better served by the systematic and explicit definition of terms, beginning with
words like knowledge, beliefs, concept, conceptual change, or learning upon which this
inquiry is framed. We expect that there will be debates and controversies as researchers
work to conceptually define these core constructs. But, once the dust settles, we believe
that a more solid foundation will have been laid for future decades of theory and research.
Integration and Modeling
Other themes we identified pertained to conceptual boundaries, complexity, and interre-

latedness. Progress in all three of these areas seems predicated on more than semantic
specificity and explication, however. It rests on researchers’ ability to conceptualize how
those constructs operate collectively, as well as how they relate to other variables or func-
tion within varying contexts. One means of concretizing these conceptualizations is
through theoretical models that specify the nature or weight of interactions between and
among variables or that demonstrate how the relations between and among variables
change over time or across contexts.
Recently, a number of scholars including Bendixen and Rule (2004), Dole and Sinatra
(1998), Gill et al. (2004), and Murphy (2005) have attempted to model the process by
which beliefs (or knowledge) undergo transformation. Although these models are still in
development or under test, they share certain correspondences that can be catalysts for
future study. In addition, the intricacy and complexity of the dimensions and processes
portrayed in these models are unavoidable. And yet, none of these models do more than
capture some aspect of knowledge or belief change.
Upon examination, the aforementioned models also suggest that affective concerns may
play more significant roles in the changing of beliefs than in knowledge change. Still, these
proposed models differ markedly in the specific components they contain or exclude and the
paths or causal chains they predict. For instance, across the previously cited models, there is
a distinction in the degree to which doubt, uncertainty, or dissonance is considered an essen-
tial trigger for belief or knowledge change. In general, we look forward to the maturation of
these emergent models and to the focus they bring to central constructs and their interrelations.
Instructional Interventions and Developmental Research
For all the hypotheses and predictions the current chapters and emergent models contribute
to the literature on the nature of knowledge and belief change, there remains a paucity of
intervention and longitudinal studies. Consequently, understandings about relations
between epistemic beliefs, conceptual change, and learning are still based largely on cor-
relational data. What causes what and what happens over time are questions that such cor-
relational data cannot adequately answer.
Without question, therefore, there is a compelling need to expand the empirical base in
this field. That expansion should entail both classroom-based studies that explore the
effects of direct interventions (such as suggested by the Mason and Gava, Duit et al., and
the Stathopoulou and Vosniadou chapters, this volume) and long-term studies that address
developmental questions.
As discussed, however, these intervention and developmental studies cannot be
parochial in perspective, ignoring the possibility of cultural influences on students’ chang-
ing beliefs and knowledge. Instead, it is important that the empirical programs of research
we are envisioning embrace sociocultural factors and take place in varied cultural settings.
On this latter point, the current volume which represents a truly international perspective
on conceptual change and epistemology is to be praised. In our opinion, a similar interna-
tional orientation to classroom and developmental studies of epistemic beliefs, conceptual
change, and learning should be aggressively pursued.
Measurement Tools and Statistical Procedures
Even a cursory examination of the literature in epistemic beliefs, conceptual change, and
learning brings serious measurement and statistical concerns to light — concerns that
must be confronted if future progress is to be made. For example, so much of what we
have come to know about epistemic beliefs rests almost exclusively on self-report data.
It will take a very creative mind to find alternative means of uncovering individuals’
beliefs about knowledge and knowing that move beyond self-reports or at a minimum
corroborate them. Moreover, there has been continuing efforts to construct knowledge
and beliefs measures that result in reliable and valid data. This concern is especially true
for measures of epistemic beliefs that do not always produce data that match hypothe-
sized factor structures or that may compound ontology issues with epistemological ones
(Alexander, 2006).
Combined with the search for more effective and informative measures to epistemic
beliefs, conceptual change, and learning, researchers must continue their quest for sta-
tistical procedures that permit more specific and accurate representation of the inter-
play between these key constructs. Here again, the concern arises that current statistical
procedures simply may not be able to do justice to the complexity or developmental
nature of epistemic beliefs, conceptual change, and learning. Therefore, even as we call
for more enriched and elaborate models of the change process, we acknowledge that
existing models may already over-reach our statistical capabilities (Kulikowich &
Young, 2001).
As a community, we have tended to rely too heavily on data-reduction procedures that
are part of the latent-variable modeling family (e.g., factor analysis or structural equa-
tion modeling) that cannot readily or effectively accommodate the extensive number of
constructs and relations that must be depicted in our theoretical and conceptual models.
As researchers have argued (Alexander, Murphy, & Kulikowich, in press; Kulikowich &
Young, 2001), viable statistical alternatives likely exist. However, those alternatives can
only serve us if our tasks and interventions undergo dramatic transformations as well.
Consequently, future understandings about the relations between epistemic beliefs, con-
ceptual change, and learning will require the direct involvement of those with quantita-
tive and qualitative expertise who can assist in the identification or development of
appropriate statistical tools that effectively complement experimental tasks and inter-
ventions.
Concluding Thoughts
The nature of first steps is such that they represent precarious and rudimentary acts that are
simultaneously significant and foundational; so it is with the findings, implications, and
recommendations outlined by the authors of these preceding chapters. Entwistle, Murphy
and colleagues, Mason and Gava, Duit and colleagues, and Stathopoulou and Vosniadou
are scholars who stand in the vanguard of exciting research that seeks to bring the litera-
tures on epistemic beliefs, conceptual change, and learning into alignment. We recognize
that future progress in this multidimensional arena may not be certain or steady and that
those attempting these movements may occasionally stumble or fall along the way.
Nonetheless, we feel that we are bearing witness to a meaningful event in this nascent field
of inquiry and look forward to whatever unfolds in the next decade of theory and research.
Our underlying hope is that indeed the findings, implications, and recommendations pre-
sented in the aforementioned chapters are truly first steps in what promises to be a long
and fruitful journey.
References
Alexander, P. A. (1997). Mapping the multidimensional nature of domain learning: The interplay of
cognitive, motivational, and strategic forces. In: M. L. Maehr, & P. R. Pintrich (Eds), Advances in
Motivation and Achievement (Vol. 10, pp. 213–250). Greenwich, CT: JAI Press.
Alexander, P. A. (2005). Teaching towards expertise [Special Monograph on Pedagogy — Learning
for teaching]. British Journal of Educational Psychology. Monograph Series II, 3, 29–45.
Alexander, P. A. (2006). What would Dewey say? Channeling Dewey on the issue of the specificity
of epistemic beliefs. Educational Psychology Review, 18(1), 55–65.
Alexander, P. A., & Dochy, F. J. R. C. (1995). Conceptions of knowledge and beliefs: A comparison
across varying cultural and educational communities. American Educational Research Journal,
32, 413–442.
Alexander, P. A., Fives H., Buehl, M. M., & Mulhern, J. (2002). Teaching as persuasion. Teaching
and Teacher Education, 18, 795–813.
Alexander, P. A., Murphy, P. K., & Kulikowich, J. M. Expertise and adult development: New mod-
els and critical implications. In: M. C. Smith, & T. Reio (Eds), Handbook on adult development.
Mahwah, NJ: Lawrence Erlbaum Associates (in press).
Alexander, P. A., Schallert, D. L., & Hare, V. C. (1991). Coming to terms: How researchers in learn-
ing and literacy talk about knowledge. Review of Educational Research, 61, 315–343.
Alexander, P. A., Sperl, C. T., Buehl, M. M., Fives, H., & Chiu, S. (2004). Modeling domain learn-
ing: Profiles from the field of special education. Journal of Educational Psychology, 96, 545–557.
Bendixen, L. D., & Rule, D. C. (2004). An integrative approach to personal epistemology: A guid-
ing model. Educational Psychologist, 39(1), 69–80.
Bereiter, C. (1994). Constructivism, socioculturalism, and Popper’s World 3. Educational
Researcher, 23(7), 21–23.
Bereiter, C., & Scardamalia, M. (1989). Intentional learning as a goal of instruction. In:
L. B. Resnick (Ed.), Knowing, learning and instruction: Essays in honour of Robert Glaser (pp.
361–392). Hillsdale, NJ: Erlbaum.
Buehl, M. M., & Alexander, P. A. (2005). Motivation and performance differences in students’ domain-
specific epistemological beliefs profiles. American Educational Research Journal, 42(4), 697–726.
Dole, J. A., & Sinatra, G. M. (1998). Reconceptualizing change in the cognitive construction of
knowledge. Educational Psychologist, 33(2/3), 109–128.
Fives, H. L., Alexander, P. A., & Buehl, M. M. (2001). Teaching as persuasion: Approaching class-
room discourse as refutational text. In: J. V. Hoffman, D. L. Schallert, C. M. Fairbanks, J. Worthy,
& B. Maloch (Eds), Fiftieth yearbook of the National Reading Conference (pp. 200–212).
Chicago, IL: National Reading Conference.
Gill, M., Ashton, P. T., & Algina, J. (2004). Changing preservice teachers’ epistemological beliefs
about teaching and learning in mathematics: An intervention study. Contemporary Educational
knowledge and knowing and their relation to learning. Review of Educational Research, 67,
88–140.
Hofer, B. K., & Pintrich, P. R. (Eds). (2002). Personal epistemologies: The psychology of beliefs
about knowledge and knowing. Mahwah, NJ: Lawrence Erlbaum.
Karabenic, S. A., & Moosa, S. (2005). Culture and personal epistemology: U.S. and middle eastern
students’ beliefs about scientific knowledge and knowing. Social Psychology of Education, 8,
375–393.
King, P. M., & Kitchener, K. S. (1994). Developing reflective judgment: Understanding and
promoting intellectual growth and critical thinking in adolescents and adults. San Francisco:
Jossey-Bass.
Kirschner, P. A., Sweller, J., & Clark, R. E. (2006). Why minimal guidance during instruction does
not work: An analysis of the failure of constructivist, discovery, problem-based, experiential, and
inquiry-based teaching. Educational Psychologist, 41(2), 75–86.
Kitchener, K. (2002) Skills, tasks, and definitions: Discrepancies in the understanding and data on
the development of folk epistemology. New Ideas in Psychology, 20, 309–328.
Kulikowich, J. M., & Young, M. F. (2001). Locating an ecological psychology methodology for
situated action. Journal of Learning Sciences, 10, 165–202.
Murphy, P. K. (2005). Persuasion in talk and text: Plying psychological tools to philosophical prob-
lems. Richard E. Snow invited address presented at the National Convention of the American
Psychological Association, Washington, DC.
Murphy, P. K., & Alexander, P. A. (2000). A motivated look at motivational terminology [Special
Issue]. Contemporary Educational Psychology, 25, 3—53.
Murphy, P. K., & Alexander, P. A. (2004). Persuasion as a dynamic, multidimensional process: A
viewfinder for individual and intraindividual differences. American Educational Research
Journal, 41, 337–363.
Perry, W. G. (1970). Forms of intellectual and ethical development an the college years: A ccheme.
New York: Holt, Rinehart and Winston.
Phillips, D. C. (1995). The good, the bad, and the ugly: The many faces of constructivism.
Educational Researcher, 24(7), 5–12.
mology. In: B. K. Hofer, & P. R. Pintrich (Eds), Personal epistemology: The psychology of beliefs
Qian, G., & Alvermann, D. (1995). Role of epistemological beliefs and learned helplessness in sec-
ondary school students’ learning science concepts from text. Journal of Educational Psychology,
87(2), 282–292.
Qian, G., & Alvermann, D. E. (2000). The relationship between epistemological beliefs and con-
ceptual change learning. Reading & Writing Quarterly, 16, 59–74.
Schraw, G., Bendixen, L. D., & Dunkle, M. E. (2002). Development and validation of the epistemic
belief inventory (EBI). In: B. K. Hofer, & P. R. Pintrich (Eds), Personal epistemology: The
psychology of beliefs about knowledge and knowing (pp. 261–275). Mahwah, NJ: Lawrence
Erlbaum Associates.
Siegler, R. S. (1996). Emerging minds: The process af change an children’s thinking. New York:
Oxford University Press.
Sinatra, G. M. (2005). The “warming trend” in conceptual change research: The legacy of Paul
Pintrich. Educational Psychologist, 40(2), 107–115.
Sinatra, G. M., & Kardash, C. M. (2004). Teacher candidates’ epistemological beliefs, dispositions,
and views on teaching as persuasion. Contemporary Educational Psychology, 29, 483–498.
Sinatra, G. M., & Pintrich, P. R. (Eds). (2003). Intentional conceptual change. Mahwah, NJ:
Lawrence Erlbaum Associates.
Sinatra, G. M., Southerland, S. A., McConaughy, F., & Demastes, J. (2003). Intentions and beliefs
in students’ understanding and acceptance of biological evolution. Journal of Research in Science
Teaching, 40, 510–528.
Southerland, S., Sinatra, G. M., & Mathews, M. (2001). Beliefs, knowledge, and science education.
Educational Psychology Review, 13(4), 325–351.
Stathopoulou, C., & Vosniadou, S. (this volume) Conceptual change in physics and physics related
epistemological beliefs: A relationship under scrutiny. In: S. Vosniadou, X. Vanvakoussi, &
C. Stathopoulou (Eds), Advances in conceptual change research and the role of epistemological
beliefs. Oxford, UK: Elsevier.
W. Schnotz, S. Vosniadou, & M. Carretero (Eds), New perspectives on conceptual change. Oxford,
UK: Elsevier.
Weinstock, M. (2005). Grade level, gender, and ethnic differences in epistemological understanding
ithin domains. In: F. Hearle, & L. Bendixen (Co-chairs), Current research in children’s personal
epistemology: Implications for developmental aspects of learning and instruction. Symposium
presented at the Meeting of the European Association for Research in Learning and Instruction,
Nicosia, Cyprus.
Windschitl, M., & Andre, T. (1998). Using computer simulations to enhance conceptual change: The
roles of constructivist instruction and student epistemological beliefs. Journal of Research in
Science Teaching, 35, 145–160.
Wood, P., & Kardash, C. (2002). Critical elements in the design and analysis of studies of episte-
mology. In: B. K. Hofer, & P. R. Pintrich (Eds), Personal epistemology: The psychology of beliefs
PART 3:
EXTENDING THE CONCEPTUAL CHANGE

APPROACH TO MATHEMATICS LEARNING
Chapter 16
Extending the Conceptual Change Approach

to Mathematics Learning: An Introduction
Xenia Vamvakoussi
This part Extending the Conceptual Change Approach to Mathematics Learning comprises
four chapters and two commentaries, which attempt to show the relevance of the concep-
tual change approach for the learning and teaching of mathematics. In the first chapter,
Merenluoto and Palonen present the results of interviews with seven professional mathe-
maticians reflecting on their efforts to understand various aspects of the number concept,
from their privileged position as members of the community of mathematicians. The fol-
lowing two chapters by Vamvakoussi and Vosniadou, and Christou, Vosniadou and
Vamvakoussi, attempt to show that the conceptual change approach can make meaningful
predictions regarding students’ difficulties to understand the mathematical concept of
rational number and the use of literal symbols in algebra. In the fourth chapter, Tsamir and
Tirosh describe a number of teaching interventions designed to facilitate students’ under-
standing of certain properties of infinite sets, in light of conceptual change instruction
design principles. The commentators look at the chapters from different points of view.
Brian Greer and Lieven Verschaffel argue in favor of a conceptual change perspective, both
in the context of the historical evolution of mathematics as a discipline and in the context
of mathematics education. On the contrary, Anna Sfard attempts to critically examine the
assumptions and findings of the first four chapters by reinterpreting them through the
lenses of a different theoretical framework, which she describes as communicational.
In their attempt to extend the conceptual change approach in the learning and teaching
of mathematics all the chapters of this part address, either implicitly or explicitly, certain
issues of importance, which we will try to highlight in this short introduction.
Can the Conceptual Change Approach Bring Anything New to

Research on Mathematics Education?
We believe that there is at least one new and important issue that the conceptual change
approach brings to mathematics education and this concerns the role of prior knowledge
in mathematics learning and teaching.

ISBN: 0-08-045355-4
240 Xenia Vamvakoussi
When it comes to mathematics as a school subject, the importance of prior knowledge

in further learning is widely recognized. Indeed, it is reported (e.g., Grossman &
Stodolsky, 1995) that mathematics educators consider their subject to be highly sequen-
tial, more than science or social sciences educators. This implies that students cannot
understand new mathematical content if they have not mastered what has been already
taught — which of course seems as a reasonable thing to believe. However, it reflects a
limited view of the role of prior knowledge in learning: Prior knowledge is considered
to be a base upon which further learning is built and it may constrain learning when it is
either inadequate or even absent. This view is compatible with the belief that learning is
additive and may prevent mathematics teachers, as well as designers of mathematics cur-
ricula, from taking a conceptual restructuring perspective on the processes of mathe-
matics learning.
As Resnick (2006) points out, it has taken some time for mathematics education
researchers to realize that initial mathematical understandings may not always be support-
ive of further learning, and that in some cases they may in fact inhibit further learning. She
makes an example out of the Cognitively Guided Instruction (CGI), a well-documented
approach to mathematics education developed by Carpenter and his colleagues (Carpenter
& Moser, 1984; Carpenter, Fennema, & Peterson, 1987), which encouraged and trained
mathematics teachers to build instruction on children’s intuitive ideas. Although CGI had
strong results for primary-level teachers and their students, its effect was not that strong
for older students, a fact that has been attributed to the lack of an adequate body of research
evidence related to students’ ideas on more advanced mathematical concepts. According to
Resnick, this explanation was based on the implicit assumption that learning of elemen-
tary and more advanced concepts would not be fundamentally different — the possibility
that knowledge based on intuition would become deeply entrenched and stand in the way
of further learning has not been anticipated.
On the other hand, it would be unfair to say that mathematics education researchers
have not noticed the fact that in some cases prior knowledge can inhibit new learning.
Fischbein (1987), for example, was one of the first to notice that intuitive beliefs may be
an important contributor to students’ systematic errors in mathematics, a fact also noted by
Vergnaud (1989) and Sfard (1987). Brousseau’s theory of epistemological obstacles
(1997) shares some similarities with the conceptual change approach. Still, as pointed out
by Greer (2004), such theoretical approaches have not found their way into the actual
mathematics teaching practices.
To summarize, although there have been attempts in mathematics education research to
take a wider perspective on the role of prior knowledge in further learning, it seems that
both research and practice can profit by a well-articulated constructivist framework that
predicts and explains a wide range of difficulties that students may meet and the kinds of
misconceptions that should be expected; also, as Greer and Verschaffel (this volume)
argue, the particular value of the conceptual change framework lies in the fact that it pro-
poses a number of specific and testable principles (see Vosniadou, Ioannides,
Dimitrakopoulou, & Papademetriou, 2001) that can guide instruction and research on
instruction. This particular characteristic is in line with what De Corte (2004) describes as
a major goal of current and future research on the learning and teaching mathematics.
Extending the Conceptual Change Approach to Mathematics Learning 241
Can the Conceptual Change Approach be Applied to Mathematics

Learning?
The attempt to apply the conceptual change theoretical framework in mathematics learn-
ing is relatively recent (Vosniadou & Verschaffel, 2004). The conceptual change frame-
work has its roots in the philosophy and history of science and has been initially applied
to explain students’ misconceptions mainly in learning science. The question whether this
framework can be applied fruitfully in mathematics learning is a plausible one; it might be
the case that patterns of change in the course of mathematics history growth, as well as in
mathematics learning, are not similar to science and science learning.
As Greer and Verschaffel (this volume) argue, although several distinctions can be made
between the characters of the changes that occur in science and in mathematics, there are
several changes in mathematics that merit the title “revolutionary”, in the sense that prior
“background assumptions” (Baltas, this volume) are discarded or that
a significant part of the ‘older’ mathematics will come to be replaced or

dramatically augmented by concepts and techniques that visibly change the
vocabulary and grammar of mathematics. (Dauben, 1992, p. 80, quoted by
Greer & Verschaffel, this volume)
This issue is also discussed in the chapters authored by Merenluoto and Palonen, and
Vamvakoussi and Vosniadou.
Vamvakoussi and Vosniadou (this volume) draw on the argument that, from a historical
point of view, the conceptual change perspective is applicable to the development of math-
ematical concepts rather than theories (see also Vamvakoussi & Vosniadou, 2004). They
refer to the number concept to show that in the process of its development the meaning of
the term “number”, as well as the ontological status of its referents, has undergone con-
siderable changes.
Merenluoto and Palonen (this volume) also refer to the concept of number, to point out
that progress in mathematics is far from being linear and smooth. They focus on the con-
troversies and discussions that lie behind the number concept to highlight the social char-
acter of mathematical knowledge construction.
In educational context, it appears that not only it is expected from students to mas-
ter under instruction parts of the body of mathematical knowledge that took millennia
to develop (Greer, 2006; Greer & Verschaffel, this volume), but also the discussions,
controversies, and shared conventions within the mathematics community are con-
cealed (Merenluoto & Palonen, this volume). Tsamir and Tirosh (this volume) argue
that it may be profitable for students to be informed of mathematicians’ difficulties with
counterintuitive concepts. This is in line with a conceptual change instruction principle,
stating that
it is important in instruction to distinguish new information that is consis-

tent with prior knowledge from new information that runs contrary to prior
knowledge. (Vosniadou et al., 2001, p. 393)
Tsamir and Tirosh presented their students with two excerpts of statements made by
well-known mathematicians who acknowledged the difficulties involved in the compari-
son of infinite sets and stressed that it is legitimate to wonder about the nature of infinity,
in an effective attempt to address students’ entrenched presuppositions about enumeration.
On the other hand, learning within the re-framed conceptual change framework does
not require the replacement of “incorrect” with “correct”’ conceptions, but the ability on
the part of the learner to take different points of view and understand when different con-
ceptions are appropriate depending on the context of use (Vosniadou, this volume).
Merenluoto and Palonen’s (this volume) participants, being expert mathematicians,
describe nicely how their thinking about numbers differs with respect to the context and
how they are able to move flexibly from one context to the other. In educational context,
Tsamir and Tirosh (this volume) aim at supporting their students develop this ability in
relation to the comparison of sets: They report that helping students to become aware of
the different techniques they use to compare finite sets and select the appropriate technique
to use in the case of infinite sets proved to be gainful.
How is Mathematics Learning Different from Science Learning?
As argued in the introduction of this volume, within the re-framed conceptual change
framework, mathematics learning shares many similarities with science learning: As it
is the case that students develop a naïve physics on the basis of everyday experience,
they also develop a naïve mathematics which appears to be neurologically based
(through a long process of evolution), and to consist of certain core principles or pre-
suppositions (such as the presupposition of discreteness in the number concept) that
facilitate some kinds of learning but inhibit others (Dehaene, 1998; Gelman, 2000;
Lipton & Spelke, 2003).
However, the role of empirical facts is not the same in science and mathematics (Tsamir
& Tirosh, this volume). There are arguably more opportunities to have real life experiences
of gravity and motion, than of irrational numbers. As pointed out by Greer (2004) and
Greer and Verschaffel (this volume), most mathematical concepts are introduced in the
course of formal instruction. The question arises, is the conceptual change approach appli-
cable only in the case of elementary concepts that can be directly related to experience?
An immediate answer can be given through the chapters authored by Christou et al.,
Tsamir and Tirosh, Vamvakoussi and Vosniadou, which refer to students at secondary
school and at university level. They show that initial understandings of natural numbers
and enumeration, which are bound to be directly grounded in experience, interfere with the
understanding of more advanced mathematical content, such as rational numbers, literal
symbols in algebra, and the comparison of infinite sets. However, there are also studies
conducted within the conceptual change framework dealing with concepts that are intro-
duced through instruction and are not related — at least not directly — to experience. For
instance, Biza and Zachariades (2006) investigated first-year university mathematics
students’ understanding of the tangent line on a curve. They showed that students generate
synthetic models of the tangent line, reflecting the specific properties of the circle tangent
line that they have been initially introduced to in high school.
Another thing that differentiates mathematics learning from science learning is the
abstract nature of its objects of study and the extensive use of symbols. For a mathemati-
cally versed person, it may be easy to consider 3, –3.14, !, 2/3, and –7i + 6 as objects that
can be called numbers (Greer & Verschaffel, this volume), and to treat them as such. But
for the mathematics novice, the same task is extremely complicated: One has to disregard
superficial differences (like differences in notation) and deeper differences (like differ-
ences in use), to conceptualize 2/3 as an object, rather than as a process. In addition,
Markovitz and Sowder (1991) observe that a major difficulty faced by learners in mathe-
matics is the use of different symbols to represent the same idea and similar-looking sym-
bols to represent different ideas. For instance, 2/4, 3/6, 0.50 all refer to the number “one
half”. On the other hand, one uses the symbol minus (–) to denote an operation (subtrac-
tion) and to signify “negativity” (e.g. Vlassis, 2004). To make meaningful predictions
related to mathematics learning, the conceptual change framework should account for the
difficulties stemming out of these particular characteristics.
In this volume, Christou et al. argue that that prior knowledge of numbers, especially
natural numbers, in the context of arithmetic, interferes with the interpretation of literal
symbols. They present evidence showing that students tend to interpret literal symbols as
standing mainly for natural numbers. They also show that students tend to interpet what
they call the phenomenal sign of an algebraic expression (such as the negative sign in ‘–x’)
on the basis of their knowledge of the actual sign of numbers in the context of arithmetic.
Vamvakoussi and Vosniadou (this volume) point out that interpreting rational number
notation is a major difficulty for students. They present evidence that students tend to inter-
pret different symbolic representations (fractional, decimal) of a rational number to stand
for different numbers, which results in thinking of fractions and decimals as two different,
unrelated sets of numbers. They suggest that students’ initial explanatory frameworks of
number are usually simply enriched with “new kinds of numbers”, i.e. fractions and deci-
mals. They appeal to learning by mere enrichment to explain students’ generation of syn-
thetic models of rational numbers intervals.
Such findings and interpretations are consistent with the conceptual change approach.
Of course, there is much to be done to explore further the advantages and limitations of
this approach to mathematics learning.
What can We Gain from a Conceptual Change Approach to

Mathematics Learning and Teaching?
Tirosh and Tsamir (2004) argue that mathematics education and conceptual change
research both can gain from each other. In this volume, Tsamir and Tirosh systematically
evaluate certain principles for instruction aiming at conceptual change through reflecting
on the design of successful learning environments. They refer to a number of their studies
aiming at teaching effectively the comparison of infinite sets. It appears that the choices
they made, based on their rich experience and their sensitivity about the role of intuitive
beliefs in learning, are highly compatible with the conceptual change instruction design
principles. One could expect that it would be helpful for mathematics educators to be
aware of these principles when starting to design a mathematics learning environment.
In addition, Vosniadou and Vamvakoussi (2006) argue that the conceptual change per-
spective could be helpful in the design of mathematics curricula. In Greer’s (2006) words,
it is essential to take a long-term perspective in the design of mathematics instruction,
which involves
careful sampling from the valid range of examples, the anticipation of

expansions of meaning, the search for bridging devices when such expan-
sions are necessary, and the open discussion with students of why concep-
tual change is necessary, and often difficult, which takes them deep into
historical and philosophical analyses. (Greer, 2006, p. 178)
How is the Conceptual Change Approach Related to Other

Approaches to Learning and Instruction?
The application of the conceptual change approach in mathematics may also reveal the
limitations of this framework — after all, no theory can possibly explain all phenomena of
learning, as Sfard (this volume) suggests. This is why the claim made in the introduction,
namely that the conceptual change approach is complementary and not incompatible with
other approaches, such as the situated cognition approach (Vosniadou, in press; Vosniadou
& Vamvakousi, 2006), has to be further substantiated. This claim has both theoretical and
educational implications. In the case of mathematics learning, it is hard to deny that learn-
ing how to count has a social and communicational character. Before children construct a
principled understanding of counting (Gelman, 2000), they must make several attempts to
assign to the counting ritual the same meaning as their parents do. It takes a lot of effort
and it is presumably an effort initially invested on communication (Sfard & Lavie, 2005).
However, there are differences between the conceptual change approach and the com-
municational framework briefly presented by Sfard (this volume) with respect to their the-
oretical assumptions. In particular, within the communicational framework, learning is
viewed as change in discourse (see, for example, Sfard & Lavie, 2005), whereas from a
conceptual change perspective, learning is accomplished through participation in social
activities and exhibited in changes in discourse, but also involves changes in assumed men-
tal structures. Not committing oneself to account for mental structures can probably result
in frameworks in which less theoretical constructs need to be developed, thus avoiding the
use of inadequately defined terms described by Sfard (this volume). However, we believe
that theoretical frameworks that take into consideration the cognitive component of learn-
ing have greater explanatory power (for a thorough discussion see Vosniadou, in press).
Within the conceptual change framework, the importance of communication is fully
recognized, especially in relation to education. Vosniadou et al. (2001) propose that learn-
ing environments that facilitate group discussion and the verbal expression of ideas make
possible for students to express their internal representations of phenomena and compare
them with those of the others. In a more general fashion, learning environments of this
kind help students increase their metaconceptual awareness of their thinking, which is an
important component of intentional learning (Vosniadou, in press, 2003). In support of this
view, Tsamir and Tirosh (this volume) claim that the implementation of this principle had
positive effect on their students’ learning about the comparison of infinite sets. However,
further research is needed to examine the educational implications of the conceptual
change approach in mathematics learning and teaching.
Acknowledgments
work of the project “Pythagoras — Support of University Research Groups” with 75%
References
Biza, I., & Zachariades, T. (2006). Conceptual change in advanced mathematical thinking: The case
of tangent line. In: J. Novotná, H. Moraová, M. Krátká, & N. Stehlíková (Eds), Proceedings of the
30th conference of the international group for the psychology of mathematics education (Vol. 1,
pp. 168–170). Prague: PME.
Brousseau, G. (1997). Theory of didactical situations in mathematics. Dordrecht, The Netherlands:
Kluwer.
Carpenter, T. P., Fennema, E., & Peterson, P. L. (1987). Cognitively guided instruction: The appli-
cation of cognitive and instructional science to mathematics curriculum development. In:
I. Wirszup, & R. Streit (Eds), Developments in school mathematics education around the world
(pp. 397–417). Reston, VA: National Council of Teachers of Mathematics.
Carpenter, T. P., & Moser, J. M. (1984). The acquisition of addition and subtraction concepts. In:
R. Lesh, & M. Landau (Eds), Acquisition of mathematics: Concepts and processes (pp. 7–44).
New York: Academic Press.
Dauben, J. (1992). Conceptual revolutions and the history of mathematics. In: D. Gillies (Ed.),
Revolutions in mathematics (pp. 15–20). Oxford: Oxford University Press.
De Corte, E. (2004). Mainstreams and perspectives in research on learning (mathematics) from
instruction. Applied Psychology: An International Review, 53(2), 279–310.
Dehaene, S. (1998). The number sense: How the mind creates mathematics. Harmondsworth,
Middlesex, England: Penguin.
Fischbein, E. (1987). Intuition in science and mathematics. Dordrecht, The Netherlands: Kluwer.
Greer, B. (2004). The growth of mathematics through conceptual restructuring. Learning and
Greer, B. (2006). Designing for conceptual change. In: J. Novotná, H. Moraová, M. Krátká, &
N. Stehlíková (Eds), Proceedings of the 30th conference of the international group for the psy-
chology of mathematics education (Vol. 1, pp. 175–178). Prague: PME.
Grossman, P., & Stodolsky, S. S. (1995). Content as ontext: The role of school subjects in second-
ary school teaching. Educational Researcher, 24(2), 5–11.
Lipton, J. S., & Spelke, E. S. (2003). Origins of number sense: Large numbers discrimination in
human infants. Psychological Science, 4, 396–401.
Markovits, Z., & Sowder, J. (1991). Students’ understanding of the relationship between fractions
and decimals. Focus on Learning Problems in Mathematics, 13(1), 3–11.
Resnick, L. B. (2006). The dilemma of mathematical intuition in learning. In: J. Novotná,

H. Moraová, M. Krátká, & N .Stehlíková (Eds), Proceedings of the 30th conference of the inter-
national group for the psychology of mathematics education (Vol. 1, pp. 173–175). Prague: PME.
Sfard, A. (1987). Mathematical practices, anomalies and classroom communication problems. In:
P. Ernest (Ed.), Constructing mathematical lnowledge: Epistemology and mathematics education
(Studies in Mathematics Education Series, Vol. 4). London: The Falmer press.
Sfard, A., & Lavie, I. (2005). Why cannot children see as the same what grown ups cannot see as
different? Early numerical thinking revisited. Cognition and Instruction, 23(2), 237–309.
Tirosh, D., & Tsamir, P. (2004). What can mathematics education gain from the conceptual change
approach? And what can the conceptual change approach gain from its application to mathema-
tics education? Learning and Instruction, 14, 535–540.
Vergnaud, G. (1989). L’obstacle des nombres négatifs et l’introduction à l’algèbre. In: N. Bednarz,
& C. Garnier (Eds), Construction des Savoirs (pp. 76–83). Ottawa: Agence d’ARC.
Verschaffel, L., & Vosniadou, S. (Guest Eds.). (2004). The conceptual change approach to mathe-
matics learning and teaching. Learning and Instruction (Special Issue), 14, 445–548.
Vlassis, J. (2004). Making sense of the minus sign or becoming flexible in ‘negativity’. Learning and
Vosniadou, S. (2003). Exploring the relationships between conceptual change and intentional
learning. In: G. M. Sinatra, & P. R. Pintrich (Eds), Intentional conceptual change (pp. 377–406).
Vosniadou, S. (in press). The cognitive-situative divide and the problem of conceptual change.
Special Issue of Educational Psychologist.
Vosniadou, S., Ioannides, C., Dimitrakopoulou, A., & Papademetriou, E. (2001). Designing learning
environments to promote conceptual change in science. Learning and Instruction, 11, 381–419.
Vosniadou, S., & Vamvakoussi, X. (2006). Examining mathematics learning from a conceptual
change point of view: Implications for the design of learning environments. In: L. Verschaffel,
F. Dochy, M. Boekaerts, & S. Vosniadou, (Eds), Instructional psychology: Past, present and future
trends. Sixteen essays in honour of Erik De Corte (pp. 55–70). Oxford: Elsevier press.
Chapter 17
When We Clashed with the Real Numbers:

Complexity of Conceptual Change in Number
Concept
Kaarina Merenluoto and Tuire Palonen
Introduction
Empirical research on conceptual change in the learning of science has been extensive
over the last two decades (e.g. Carey, 1985; Carey & Spelke, 1994; Chi, Slotta, & de
Leeuw, 1994; Hatano & Inagaki, 1998; Ioannides & Vosniadou, 2002; Karmiloff-
Smith, 1995; Posner, Strike, Hewson, & Gertzog, 1982; Vosniadou, 1994). The main
findings have been that, in situations where learners’ prior knowledge is incompatible
with the scientific concepts accepted and defined by the communities of scientists, stu-
dents are prone to make systematic errors or have misconceptions. Many of these mis-
conceptions are synthetic models (cf. Vosniadou, 1994; 1999), indicating that students
are assimilating new information into prior knowledge instead of making radical
changes. The results of several teaching experiments to facilitate conceptual change
indicate that radical conceptual change is an extremely difficult and complicated
process, and that prior conceptions seem to be quite resistant to teaching efforts (e.g.
Duit, Roth, Komorek, & Wilbers, 2001; Mikkilä-Erdmann, 2002; Trumper, 1997; Wiser
& Amin, 2001). In addition is the social and cultural nature of the learning environ-
ments and communities in this process (Kelly & Green, 1998; Posner et al., 1982;
Strike & Posner, 1982).
Although there has been a great deal of research on conceptual change in the learn-
ing of science during recent decades, there is not much research from this perspective in
mathematics learning. This is because, in traditional contexts, mathematics is considered
as a hierarchical body of knowledge, where new discoveries are not treated as changes
but as enlargements (Boyer, 1959), and in which the previous ones are included as
substructures. Hence, in mathematics culture, changes are not treated as revolutions as
they are in the development of science concepts (cf. Kuhn, 1970). Recent empirical
studies in mathematics learning from a conceptual change perspective (e.g. Lehtinen,

ISBN: 0-08-045355-4
248 Kaarina Merenluoto and Tuire Palonen
Merenluoto, & Kasanen, 1997; Merenluoto & Lehtinen, 2002; Stafylidou & Vosniadou,
2004; Vamvakoussi & Vosniadou, 2004) indicate that the extensions in the number con-
cept actually require a radical revision of prior knowledge structures: a conceptual
change. The natural numbers and their discrete nature have a high intuitive acceptance
attached to them as being self-evident, self-justifiable or self-explanatory. Therefore,
students in the school context seem to have a tendency towards overconfidence
(Fischbein, 1987), and towards relying on the logic of natural numbers in their reason-
ing. This leads to systematic and resistant misconceptions in problems where rational or
real numbers are used (e.g. De Corte & Verschaffel, 1996). Even at the higher levels of
education, the majority of Finnish students was not able to explain how they think about
numbers, and was unaware of the fundamental conceptual differences between the num-
ber domains (Merenluoto & Lehtinen, 2002, 2004).
In order to better understand this difficulty in the school context, we analyze retro-
spective interviews in which mathematicians explain their history in learning real num-
bers as well as their informal, metaphorical descriptions of their thinking about
numbers. We also examine how the shared practices in the mathematics community and
the mathematicians’ involvement in the mathematics culture interact with explanations
of individual learning and thinking about real numbers. By the term school context, we
mean the class community, which does not necessarily provide the respective concep-
tual tools with which to share mathematical understanding. By the term mathematics
culture, we mean the knowledge, a practice of mathematicians, which is mediated by
the historical development of mathematics, as it has been co-constructed over two
thousand years. There, the conceptual tools (like common language, acceptance of con-
ventions, awareness of inconsistencies and ways to deal with them) are developed
through discussions, scientific debate and argumentations. By the term mathematics
community, we mean the society or colloquium in which these jointly developed con-
ceptual tools are used by practicing mathematicians. In the following pages, we first
provide a brief historical background regarding the development of the concept of real
numbers and then explain some of the foundational problems related to these defini-
tions.
Historical Background of Real Numbers
The notion of real numbers is one of the most complex and profound concepts in mathe-
matics, and it has a long and complicated history of development. Behind this long devel-
opment period lays the dichotomy between discrete and continuous quantities that was
faced already in Antiquity. The so-called exhaustion method of Archimedes dealt with
some aspects of this dichotomy. This method was a primitive limiting process to deal with
infinite small quantities, which stated:
If from the greater there be subtracted a magnitude greater than its half, and
from that which is left a magnitude greater than its half, and this process be
repeated continually, there will be left some magnitude which will be less
than the lesser magnitude set out. (Boyer, 1959, p. 33)
Complexity of Conceptual Change in Number Concept 249
The concept of limit was actually embedded in this process, but because of the prob-
lems with infinitesimals — infinitely small quantities — the efforts to use numbers in
order to handle continuity led to paradoxes that disturbed mathematicians for two thousand
years (Boyer, 1959; Dantzig, 1954; Kline, 1980). Finally, in the latter part of the 19th cen-
tury, the continuous and discrete quantities were integrated into the abstract theory of real
numbers. Mathematicians have different kinds of rigorous constructs for these numbers,
the most familiar ones being those where the real numbers are constructed as limits of
Cauchy’s sequence1 or as Dedekind’s cuts2 in the number line.
From the point of view of conceptual change, it is important to notice that the long
development period of the theory of real numbers indicates the radical nature of the
change, which was required when the initial thinking of numbers as discrete objects was
integrated with the knowledge belonging ontologically to another domain of continuous
objects (Boyer, 1959; Kline, 1980; Rudin, 1953). This integration also required a radi-
cal change in the discussion and communication in the socially shared practices of
mathematicians. After this change, mathematical reasoning on continuity was no longer
based on intuitions of physical reality, but on logically consistent arguments. In addition,
during the long period of development, which preceded the formation of this rigorous
theory, mathematicians like Newton tried to justify the concept of limit with the exis-
tence of continuity, for example, using the “ultimate ratios in which the quantities
vanish” (Boyer, 1959, p. 198). Finally, during the 19th century, in the modern theory of
real numbers, the concept of continuity is based on the existence of limit, that is, on the
abstract density of numbers.
The hierarchical system of extensions in the number concept, constructed after the for-
mation of the theory of real numbers at the end of the 19th century, is typical of mathe-
matics. This hierarchical system is a product of successive abstractions, in which the
numbers from the previous domain are used to build the more advanced set of numbers.
Then the more advanced domain creates a new frame of reference, from which mapping
to the previous domain is formed (see Landau, 1960). The numbers in the previous domain
are then treated as a subset of the more advanced ones (Figure 17.1). Thus, for example,
the natural numbers are treated as a subset of rational numbers, and rational numbers as a
subset of real numbers.
In Figure 17.1, the set of real numbers is, for clarification, presented as a set model.
However, for the mathematicians it is more common to use the number line as a represen-
tation of the infinite set of real numbers (the continuum). According to the continuum
hypothesis, only real numbers, including the transcendental numbers (like ! and e), form
the continuum of numbers. This hypothesis, proposed by Cantor, is based on cardinalities
of infinite sets, and it claims that any infinite set of real numbers contains either exactly as
many elements as there are natural numbers or as many as there are real numbers. Thus, it
states that there is no other transfinite number between the cardinality of natural and the
1
A number l is called the limit of an infinite sequence u1, u2, u3,... if for any positive number ! " 0 we can find
a positive number N depending on ! such that !un " l! # ! for all integers n " N.
2
Dedekind’s “cut” means a cut in the number line; thus, for example, when a cut is defined by the number "2#, it
represents all the numbers that are smaller than "#2 on the number line.
Real numbers
Rational numbes
Integers
Natural numbers numbers like π

numbers
....-3, -2, -1,
like 1 , 37 ,
and 2
0, 1, 2,… 4 43
”
no biggest” ”
”
no smallest”
no next”
”
continuum of real numbers”
Figure 17.1: A simplified model of the extensions of number domains from natural num-
bers to real numbers and the changes in the level of abstraction with each enlargement.
cardinality of real numbers. This hypothesis, however, led to serious logical problems
among mathematicians. Finally in 1930, Gödel proved that there will always be statements
within an axiomatic system which can be neither proved nor disproved on the basis of
these axioms. Later, in 1953, Cohen showed the independence of the continuum hypothe-
sis from the other axioms of set theory. This implies that it is impossible to prove that the
continuum hypothesis is either true or false from the usual axioms of set theory. In the
mathematics culture, these cognitive conflicts were finally handled by mutual agreements
and shared discussions which opened up the possibility of choice between different set the-
ories, and a way to avoid disturbing conflicts.
These socially negotiated contracts, rules and examples, which have been constructed
through discussions among mathematicians, are behind the modern concepts of limit and
continuity also used in the school context. According to several studies, these concepts are
difficult also for students today (e.g. Cornu, 1991; Fishbein, Jehiam, & Cohen, 1995;
Merenluoto & Lehtinen, 2002; Sierpinska, 1987; Szydlik, 2000; Williams, 1991). We argue
that in the development of students from novices to experts (Patel & Groen, 1991) there is
a connection between the individual knowledge acquisition processes and participation in
the culture of expertise (Sfard, 1998). The conceptual tools developed, like real numbers,
are jointly formulated during the history of the development of the concept. Through these
kinds of processes, the experiences and practices of a community are transformed into
external and communicable items. In the literature, this has been called reification.
Reification means treating an abstraction or thought as if it were a concrete and real object
(cf. Sfard, 1991, 1994). Reifications represent shared narratives, forms, symbols, common
language, acceptance of conventions, awareness of inconsistencies and ways to deal with
them, all of which support the members’ activity and interaction. The concept of internal-
ization refers, in turn, to the social cultural approach according to which the interaction
between different community members fosters the conceptual growth of the newcomers
(Hakkarainen, Palonen, Paavola, & Lehtinen, 2004; Galperin, 1966; Vygotsky, 1978).
Problems in Bridging Individual and Community-Level Changes
In the school context, students do not have access to the shared narratives of the mathe-
matics culture. The historical conflicts concerning real numbers are usually not taught
explicitly; in fact, it is possible that they are not even mentioned to the students who are
struggling to understand these concepts. It is important to notice that at the upper second-
ary level, the teaching of limit and continuity, the core concepts of real numbers, usually
occurs in a very “novel” domain for the students: they have not discussed concepts, such
as continuity, in mathematics before. What they have experienced about continuity and
limits comes from their everyday experience, where the limit is considered as something
that restricts or stops the continuity. Accordingly, their thinking on continuity is mediated
by metaphors in which continuity is related to the continuity of time, motion or direction.
In fact, the general idea of the continuity of motion is one of the basic features of the phys-
ical world, and even infants grasp it at some intuitive level (e.g. Spelke, 1991). This idea
is vague, dynamically and instinctively understood in the context of motion or time, but
not with numbers (Dantzig, 1954; Cornu, 1991).
Further, everyday thinking of continuity is even strengthened in the school practice
of teaching where the continuity of function is often described as the result of a con-
tinuous motion: “the pencil never leaves the paper”. Thus, students try to use their prior
knowledge and metaphors from their everyday experiences to interpret these mathe-
matical concepts. In our previous survey (Merenluoto & Lehtinen, 2002), less than 5%
of the 538 upper secondary level students showed any indications of a more radical
conceptual change in their explanations of limit and continuity. The majority of stu-
dents described continuity as something that “does not stop” or “what you can draw
without lifting your pen” or is in “one piece”. The limit in students’ explanations rep-
resented something that stops the continuity: “it is the value, where you can not go fur-
ther, a stop-sign” or they described it as approaching something: “…It approaches a
value”, “the graphs approach the point but do not hit it”, “the value the function
approaches but never reaches” (see also Cornu, 1991). The answers of students sug-
gested that they did not recognize the crucial underlying similarity between the density
of numbers on the number line and the continuity of a function based on the concept of
limit.
In the present study, our aim is to show that “doing” mathematics in the school context
is very different from “doing” mathematics within the mathematics community. One of the
important differences between students and mathematicians is that the latter are able to
meaningfully participate in the mathematics community. For this purpose, we analyze ret-
rospective interviews in which mathematicians narrated their learning history of real num-
bers and provided informal, metaphorical descriptions of their thinking about numbers. We
also examined how the shared practices in the community of mathematicians and their
involvement in the culture of mathematics interact with their explanations of individual
learning and thinking of real numbers.
Method
Participants
Ten Finnish university mathematicians, chosen at random, were contacted by phone, and
seven of them agreed to participate in the study. In Finland, most of the mathematicians in
universities are male, and in this group there were six men and one woman. All partici-
pants were working at the same university, except for the female participant. Their expert-
ise in mathematics was in the domains of functional analysis, analytic number theory,
algebraic number theory, applied mathematics, coding theory and automata theory. Five of
them had recently been teaching courses on analysis, where the concept of real numbers
was taught. We refer to the participants using the arbitrary names: Jim, Mathias, Jonas,
Eric, Carl, Thomas and Lea, in order to keep the interviewees’ identity in the strictest con-
fidence. Eric, Thomas, Carl and Jim were professors in mathematics while Jonas, Mathias
and Lea were lecturers.
Procedure
The interviews were conducted by the authors of this chapter. The conditions in which
the individual interviews were held were informal and relaxed; the interview took
place in the interviewees’ offices at the university. In these semistructured interviews,
the mathematicians were asked to explain their personal learning history of real num-
bers, their informal thinking of numbers and the basis of their certainty in using real
numbers. Each interview lasted about 1–1.5 hours, and was tape-recorded and tran-
scribed.
Analysis of Data
The analysis of the transcriptions was based only on what the participants sponta-
neously said during these interviews without making generalizations about their think-
ing. Two main criteria were used in the qualitative content analysis: (1) how the
mathematicians elaborated their individual meta-level thinking about real numbers
(their acquisition of knowledge, conceptual change), and (2) how they spontaneously
described their process of initiation into the mathematics community in relation to
their understanding of real numbers (their introductory experience, their references to
the historical development, the mathematics culture, their descriptions of their indi-
vidual certainty in relation to the logical problems in the history of real numbers). In
addition, a frequency analysis was used to measure the intensity of these expressions
and elaborations, as follows: We calculated (1) how many times the interviewee spon-
taneously referred to the above-mentioned two main criteria during the interview, and
(2) how many words they used in these spontaneous expressions at a time. Thus, for
example, during the interview, Carl spontaneously explained his informal thinking of
numbers 15 times during the interview using altogether 303 words in these expressions
(Table 17.1).
Table 17.1: Intensity of Interviewees’ Expressions and Elaborations.
Frequency of Spontaneous Count of Words in the

Expressions of Elaboration Expressions of Elaboration
Altogether
Number References to Number References to

History History
Eric 18 9 424 490

Thomas 16 8 507 218
Carl 15 5 303 84
Jonas 13 2 265 65
Mathias 8 1 405 13
Jim 9 3 192 107
Lea 5 0 177 0
In addition to the detailed frequency analysis, two independent researchers analyzed the
data and arranged the interviewees in order (from one to seven) according to the above-
mentioned frequency criteria. Cohen’s Kappa was calculated and had a value of .619,
while the Spearman’s rank correlations were .857 for the individual elaboration of think-
ing and .955 for references to mathematics culture.
Results and Discussion
We begin by presenting some of the interviewees’ typical learning experiences, and then
we proceed to describe how they elaborated their initiation into the mathematics commu-
nity in relation to their understanding of real numbers, and building their own personal cer-
tainty in relation to the logical problems in the culture of mathematics. We continue by
describing how the mathematicians described everyday thinking of numbers and their
informal and metaphorical thinking of real numbers. Finally, we present the relation
between the spontaneous references to the culture of mathematics, and the elaboration of
informal thinking of real numbers suggested by the data.
Memories from Typical Learning Experiences
While all the interviewed mathematicians clearly liked mathematics, their spontaneous ref-
erences to interest in and enjoyment of mathematics varied. Lea simply mentioned that
during her school years she understood mathematics and it was “fun”. Jim, on the other
hand, showed his enthusiasm for his experiences and made several references to the beauty
of mathematics. He and Eric both mentioned that mathematics is like music and if taught
like music, it brings joy to large groups of people.
When explaining their learning history, the mathematicians explained how during their
school years, the density of numbers was easily understood intuitively by thinking of the
density of numbers on the number line. Eric and Thomas spontaneously explained their
deliberate strategies to learn calculus. Eric read university-level mathematical books
already in his early teens. Besides thinking formally, he also tried to find informal ways to
explain things to himself. Thomas, in turn, made a visit to the USA as a teenager, where
the concepts of calculus were taught in such a way that he found the derivative and the use
of the rigorous %–" definition3 — fascinating. This discovery paved the way for him “to
understand the mathematical truths”. When looking back on their school years, these
mathematicians did not mention any particular difficulties in learning the numbers during
their school years; instead they excitedly referred to their fondness for mathematics.
Initiation into the Mathematics Culture: A Clash with the Real Numbers
There are two traditional ways to introduce real numbers in university mathematics: the
axiomatic approach and the construction approach. In the axiomatic approach, the axioms
of real numbers are presented to the students and all the following results with real num-
bers are then deducted from these axioms. According to the construction approach, the real
numbers are constructed from rational numbers in vertical abstractions and, as a result, the
real numbers are defined as limits of Cauchy’s sequences or as Dedekind’s cuts.
Thomas spontaneously mentioned at five separate times during the interview that know-
ing the rules of proof gave him a pleasant feeling of confidence. When explaining his
encounter with real numbers at university, he told how the rigorous axiomatic approach
gave him a feeling of comfort. However, in spite of that, he reported how he clashed with
the real numbers. Carl described this experience as follows:
Quite a knock-out it was at the beginning [—], for several weeks I did not
understand a thing.
But for him “this difficulty required a few weeks of thinking” and then came the “flash
of understanding” after which it was easy for him. Lea told how she understood notions
like the “supremum”,4 but in the beginning she did not understand their relation to real
numbers. For Mathias, the construction of real numbers from rational numbers was still
mysterious:
The system of extension; it was such a collision then [—] it was difficult [—
], definitely difficult [—], when it was explained [—] I did not understand
it [—] I was so puzzled..
3
The formal procedure used in dealing with the limit concept. There, for example, the number l is the limit of
f(x) as x approaches x0 if for any positive number ! (however small) we can find some positive number " such
that !f(x) " l! # ! whenever 0# !x " x0! # ".
4
“Supremum” means the “least upper bound” that is the very thing that differentiates the real numbers from the
rational ones; then, for example, "#2 being an irrational number is the “least upper bound” for all the numbers
smaller than "#2. Thus, in the domain of rational numbers, there is no “least upper bound” for respective numbers.
In fact, in his recall of past memories, he mentioned the name “Dedekind” five times,
when speaking of something that was really complicated and difficult:
… the professor explained Dedekind’s cut; and the whole autumn semester
I was wondering, what on earth is this about [—], I did not understand any-
thing there, maybe somebody did, but I did not. It was only later that the
understanding came.
Jim and Jonas, who did not use words like “clash” or “collision”, did, however, refer to
a conceptual leap in their learning of real numbers. Jonas explained that he was very
excited about the axiomatic approach: “I did not know that it was possible to do it that
way!”. But then he added: “But it was so difficult, at first I dif not understand”.
Jim explained how he taught the construction of real numbers to new students: “I really
constructed them from rational numbers” but then he admitted that, pedagogically, it was
not a very good choice because it was too difficult.
Thus, the memories from learning mathematics during the school years were pleasant
ones for the interviewed mathematicians. It was not until they faced the abstract construc-
tion of real numbers during the early stages of their initiation into the mathematics com-
munity, that they experienced an essential change in their “doing” of mathematics.
However, because of their great interest in mathematics, this change, “a clash”, led to a
deeper level of understanding of real numbers.
Formal and Informal Thinking of Real Numbers
Given their wide content knowledge, the mathematicians were able to see the whole pic-
ture of numbers. Therefore, the concept of real numbers for them seemed to represent the
whole construction of the theory of real numbers, such as abstract constructions, dynam-
ics of delta–epsilon processes, Cauchy’s sequences and the logical problems with the con-
tinuum hypothesis. Carl explained that
speaking about numbers and counting, there is no point of bothering your

brains so that all the real numbers are there: in mathematics we strive cog-
nitive economy.
Lea explained how
if some concept has several features, you are aware of them the whole time,
but if in some situation, they are not needed, it is pointless to carry them
around with you all the time.
Matthias vividly explained his thinking of the construction of real numbers as being still
mysterious. He explained that for him real numbers were decimal numbers to begin with,
and only when he is unable to prove something, then “I need to think more deeply about
the constructions”. But in everyday thinking “I do not need the system”.
Thus, they made a clear difference between their formal and informal thinking of num-
bers. All of them referred to their understanding of the formal representations of real num-
bers, but most of them explained that “nobody really thinks that way”. In fact, the thought
of using real numbers in everyday thinking was even slightly amusing for them. The
majority of mathematicians spontaneously and emphatically claimed that they do not use
real numbers in their everyday dealing with numbers. For them, all everyday numbers are
rational numbers, and the real numbers are only for theoretical purposes.
Everyday numbers are [—] always approximations of real numbers, [—] the
real numbers are theoretical objects only for the use of mathematicians. (Eric)
When I think of buying milk… I do not think about the price on the number
line (laughing). (Thomas)
… I do not need irrational numbers when chopping wood at my summer
house (laughing). (Jonas)
When explaining their informal, meta-level thinking of real numbers, the mathemati-
cians used efficient metaphors, based on the underlying general principles in the abstrac-
tion of density. Jim and Jonas referred to the geometrical abstraction of real numbers as
the density of numbers on the number line. Thomas, who had recently been teaching a
course on so-called hyper-real numbers, described his metaphor for these numbers “as
softened atoms padded with little infinitesimals” (referring to the nonstandard analysis).
Eric’s description was the most elaborate:
I think of the real numbers in a naïve way from a topological viewpoint

using the concept of neighbourhood, where it is not so much the question
about the point itself but its neighbourhood– so that every real number
defines a systemic neighbourhood where it can be infinitesimally small and
every number has its own neighbourhood. I like this kind of thinking more
than trying to think about a point.5 You cannot imagine one point, whereas
it is possible to think about its neighbourhoods. Hence one real number is
a systemic neighbourhood. (Eric)
The mathematicians expressed multiple informal representations of these numbers, and

were able to flexibly shift from one representation to the other. In addition, they were
metaconceptually aware of their efficient metaphors in thinking of these numbers. They
were also very aware of the difference between formal and informal representations of
these numbers, and able to flexibly shift between them.
Constructing Independent Personal Certainty in Relation to the Logical Problems in

the Culture of Mathematics
During the interviews, the mathematicians were asked to explain their personal cer-
tainty related to the concept of real numbers. Most of the interviewees then described it in
5
By a point he means the mathematical point, which has no dimensions.
relation to the logical problems in the mathematics culture. Four of the interviewed math-
ematicians clearly described their certainty about real numbers on two different levels. On
the one hand, they were aware of the logical problems and puzzlement in the historical
background, and of the agreements made to deal with these uncertainties within the com-
munity of mathematics. For example, Thomas explained:
…But of course here there are all kinds of logical problems and when you
are aware of them then you know that even thinking about the number line
is not that self-evident. There are all kinds of questions that can be found
like this hypothesis of continuum.
And referring to the agreements to deal with this problem, he added, “you just have to
make a choice of what kind of set theory you want to select”.
On the other hand, they all made a very distinct difference regarding their personal cer-
tainty. For example, Carl explained that:
There are two levels to this certainty: when I think of numbers the way I use
them in everyday life, or as a mathematician, there is no kind of uncertainty.
But if you go into the deep levels of logic, there were great revolutions,
when all these modern paradoxes in mathematics were found.
But then he continued by explaining his individual certainty in relation to the problems
in the culture of mathematics:
But they have no influence on the normal mathematics, although there are
properties which are very surprising, but I do not let those paradoxes
bother me or disturb my good night’s sleep.
Correspondingly, Jonas explained that “this logical part of the theory of real numbers,
the theory is so powerful that
I can not be absolutely sure if there are logical inconsistencies in it… but I
have a very high subjective confidence in it. It is a true theory.
Thus, the mathematicians spontaneously referred to the mathematics culture, especially

when explaining their personal certainty of real numbers in relation to the logical prob-
lems. The mathematicians who made most of the spontaneous references also used more
words in elaborating these explanations (Table 17.1).
According to a frequency analysis we placed interviewed mathematicians in a two-
dimensional field (Figure 17.2). One dimension (the vertical axis) represents the number
of spontaneous references to the culture of mathematics and the other (the horizontal axis)
the frequencies of their elaboration of informal thinking of numbers. There, Lea and
Mathias only referred to their personal (un)certainty in dealing with real numbers (Figure
17.2). Lea also admitted that she had never intentionally thought about the difference
between rational and real numbers, because “they are just numbers”.
Figure 17.2: Interviewed mathematicians in the two-dimensional field of explanations of

their individual thinking and their involvement in the community of mathematics.
Lea and Mathias, who gave rather scanty explanations of their thinking of numbers and
still showed some uncertainties in their own explanations, did not spontaneously refer to
the culture of mathematics during the interview, whereas the others did.
What is also worth noting is that Mathias, Jonas and Jim explained their thinking of real
numbers through the geometric representation of real numbers as points on the number
line. Eric and Thomas were the only ones who also spontaneously referred to their episte-
mology of mathematics as a living and changing system of knowledge, which is not an
empirical science.
Students may think of mathematics as a complete structure where nothing

can be added [—] but in fact, it is a continuously living and changing sys-
tem. Although you may think that today’s mathematics is ready and com-
plete, after a while you find that again something new has been created.
(Eric)
Mathematics is not an empirical science… yes, its conventions make sense,
but they are, however, only conventions and logical selections of what we
need to do… like, which of the set theories is to be chosen if we accept the
continuum hypothesis. (Thomas)
When the mathematicians explained their processes of entering into the community of
mathematicians almost all of them used words like “clash” and “knock-out” in their
descriptions of their initiation into the construction of real numbers. Thus, most of them
remembered how learning to understand the notion of real numbers, especially the con-
struction of those numbers, was something special and required them to adopt a new kind
of abstract thinking. In their individual construction processes, most of them also clearly
defined their individual certainty about these numbers in relation to the logical problems
in the culture of mathematics, of which they were aware.
There were clear differences in the frequencies of spontaneous references to the math-
ematics culture between the interviewed mathematicians (Figure 17.2). Thomas, Eric, Carl
and Jim, who made most of the comments, were all publishing professors in mathematics
with experience of international discussions within mathematics community. Compared to
them, Lea, Matthias and Jonas made clearly fewer references to the mathematics culture.
The fact that the latter were lecturers in mathematics suggests that prospective mathe-
maticians gradually gain access to the culture of mathematics.
Summary and Conclusions
In this chapter we have analyzed retrospective interviews in which mathematicians nar-

rated their learning history of real numbers and elaborated on their informal, metaphorical
thinking of numbers. Our main argument is that prospective mathematicians’ initiation into
the mathematics community involves initiation into the mathematics culture, whereas stu-
dents in the school context are introduced only to pieces of content knowledge. The results
suggest major differences between mathematicians (the experts in the present study) and
students — the novices in our previous study (Merenluoto & Lehtinen, 2002) — a differ-
ence that is of course self-evident. However, there are some important differences that need
to be pointed out in order to understand the problems students have in trying to understand
these difficult concepts.
In the first instance, the experts’ content knowledge is considerably broader than novices’,
which allows them to see the whole picture, to entertain multiple representations, and to shift
flexibly from one to the other. The mathematicians’ informal descriptions of numbers, their
spontaneous references to the number line and efficient use of these metaphors for density of
numbers and for neighborhoods suggest a flexible strategy with active use of the number
line, and conscious attention paid to the underlying fundamental principles of real numbers
explained in the theories of expertise (e.g. Chi, Glaser, & Farr, 1988; Chi, Feltovich, &
Glaser, 1981). What is also related to advanced mathematical thinking was their flexible
changes from one representation to another (e.g. Dreyfus, 1991), and the fact that they paid
primary attention to the relations between mathematical objects (e.g. Hadamard, 1945;
Liljedahl, 2004); all these indicated that they had constructed a systemic environment of
numbers: a reification (cf. Rudin, 1953; Landau, 1960).
The experts were also metaconceptually aware of their thinking and capable of mak-
ing a distinction between their everyday and formal thinking of real numbers. Most of
them spontaneously stated their relation to the mathematics culture. On the other hand,
the novices in the school context gave no indications of this kind of distinction. In fact,
the students did not understand even the basic differences between the different num-
ber domains. It is especially important to pay attention to the difference between stu-
dents’ explanations of limit and continuous objects presented in our previous research
(Merenluoto & Lehtinen, 2002, 2004), and the explanations given by the mathemati-
cians (Figure 17.3).
Figure 17.3: A tentative “graph” of the radical conceptual changes in the extensions of
the number concept and “areas” of students’ and mathematicians’ explanations.
The metaphor is one of the basic mechanisms behind conceptualization (Sfard, 1998),
and different metaphors may lead to different ways of thinking. When examining the dif-
ferences between these metaphors, it is possible to understand why the conceptual change
from discrete numbers to the density of numbers does not usually or easily occur during
school mathematics learning. In the teaching practices in the school context, the problems
inherent in the extensions of its number concept are often implicitly passed over. It is also
possible that many teachers do not understand these concepts very well either. In our
endeavors to facilitate conceptual change in the number concept, we need to give students
some access of the shared practices in the mathematics community. We could, for exam-
ple, explain the efficient metaphors the mathematicians use in explaining their thinking of
real numbers, thus paying attention to the density of numbers as a precondition for exis-
tence of the limit. Further, we could discuss these and present them side by side to discuss
the difference between the everyday metaphors and the metaphors the mathematicians use
(see also Wiser & Amin, 2001).
The results of the present study suggest that for those who do not have problems with
rational numbers, and who want to enter into the mathematics community, the essential
conceptual leap in the extensions of the number concept seems to be in the process of con-
structing real numbers out of rational numbers. It is also important to notice that for the
majority of students, the novices, at the heart of conceptual change is the conceptual dif-
ference between natural numbers compared to rational and real numbers: the difference
between discrete and continuous quantities (e.g. Vamvakoussi & Vosniadou, this volume).
This difference is embedded in the rules of calculation of these numbers, which are dif-
ferent from the rules when working with natural numbers. It is important to notice that for
the majority of students in Finland, even in the upper levels of education, the most prob-
lematic conceptual leap in the constriction of the number concept is the extension from
natural numbers to rational numbers. For these students, understanding rational numbers
seems to be clearly constrained by their foremost practice to try to remember the rules
without understanding (Merenluoto, 2005).
Cultural knowledge can scaffold individuals’ conceptual development and guide new
members, such as novice students, toward a deeper understanding of the difficult abstrac-
tions in mathematics. Learning the language of mathematics and its way to communicate
requires not only a series of individual flashes of insight, but also adopting the socially
negotiated rules and shared representations, descriptions, examples and scripts, all co-con-
structed within the mathematics community. Some of these can also be constructed
through discussions in the school context.
References
Boyer, C. B. (1959). The history of calculus and its conceptual development. New York: Dover
Publications.
Carey, S., & Spelke, E. (1994). Domain-specific knowledge and conceptual change. In:
L. A. Hirsfeld, & S. A. Gelman (Eds), Mapping the mind. Domain specificy in cognition and cul-
ture (pp. 169–200). Cambridge, MA: Cambridge University Press.
Chi, M. T. H., Feltovich, P., & Glaser, R. C. (1981). Categorization and representation of physics
Chi, M. T. H., Glaser, R., & Farr, M. F. (1988). The nature of expertise. Hillsdale, NJ: Lawrence
Erlbaum.
Chi, M. T. H., Slotta, J. D., & de Leeuw, N. (1994). From things to process: A theory of conceptual
change for learning science concepts. Learning and Instruction, 4, 27–43.
Cornu, B. (1991). Limits. In: D. Tall (Ed.), Advanced mathematical thinking (pp. 153–166).
Dordrecht, The Netherlands: Kluwer.
Dantzig, T. (1954). Number, the language of science. New York: The Free Press.
De Corte, E., & Verschaffel, L. (1996). An empirical test of the impact of primitive intuitive models
of operations on solving word problems with a multiplicative structure. Learning and Instruction,
6, 219–242.
Dreyfus, T. (1991). Advanced mathematical thinking processes. In: D. Tall (Ed.), Advanced mathe-
matical thinking (pp. 25–41). Dordrecht, The Netherlands: Kluwer.
Duit, R., Roth, W. M., Komorek, M., & Wilbers, J. (2001). Fostering conceptual change by analo-
gies — between Scylla and Charybdis. Learning and Instruction, 11, 283–304.
Fischbein, E. (1987). Intuition in science and mathematics. An educational approach. Dordrecht,
The Netherlands: D. Reidel.
Fishbein, E., Jehiam, R., & Cohen, D. (1995). The concept of irrational numbers in high-school stu-
dents and prospective teachers. Educational Studies in Mathematics, 29, 29–44.
Galperin, P. Y. (1966). On the notion of internalization. Voprosy Psikhologii, 12, 25–32.
Hadamard, J. (1945). The psychology of invention in the mathematics field. A mathematicians mind.
Princeton, NJ: Princeton University Press.
Hakkarainen, K., Palonen, T., Paavola, S., & Lehtinen, E. (2004). Communities of networked expert-
ise. Professional and educational perspectives. Oxford: Elsevier.
Hatano, G., & Inagaki, K. (1998). Qualitative changes in intuitive biology. European Journal of
Psychology of Education, 12, 111–130.
Ioannides, C., & Vosniadou, S. (2002). The changing meanings of force. Cognitive Science
Quarterly, 2(1), 5–62.
Karmiloff-Smith, A. (1995). Beyond modularity: A developmental perspective on cognitive science.
Cambridge, MA: MIT Press.
Kelly, G. J., & Green, J. (1998). The social nature of knowing: Toward a sociocultural perspective
on conceptual change and knowledge construction. In: B. Guzzetti, & C. Hynd (Eds), Perspectives
on conceptual change. Multiple ways to understand knowing and learning in a complex world (pp.
145–182). Mahwah, NJ: Lawrence Erlbaum Associates.
Kline, M. (1980). Mathematics. The loss of certainty. New York: Oxford University Press.
Kuhn, T. S. (1970). The structure of scientific revolutions. Chicago, IL: University of Chicago Press.
Landau, E. (1960). Foundations of analysis. The arithmetic of whole, rational, irrational and com-
plex numbers. New York: Chelsea Publishing.
Lehtinen, E., Merenluoto, K., & Kasanen, E. (1997). Conceptual change in mathematics: From
rational to (un)real numbers. European Journal of Psychology of Education, 12(2), 131–145.
Liljedahl, P. (2004). Mathematical discovery: Hadamard resurrected. In: M. J. Høines, &
A. B. Fuglestad (Eds), Proceedings of the 28th conference of the international group for the
psychology of mathematics education (Vol. 4, pp. 249–256). Bergen: PME.
Merenluoto, K. (2005). Restrained by beliefs in algorithms and discrete numbers: Student teachers
solution strategies in problem using rational numbers. In: E. Pehkonen (Ed.), Problem solving in
mathematics education. Proceedings of the Promath 2004 meeting (pp. 115–125). Lahti, Finland:
University of Helsinki.
Merenluoto, K., & Lehtinen, E. (2002). Conceptual change in mathematics: Understanding the real
numbers. In: M. Limón, & L. Mason (Eds), Reconsidering conceptual change. Issues in theory
and practice (pp. 233–258). Dordrecht, The Netherlands: Kluwer.
Merenluoto, K., & Lehtinen, E. (2004). Number concept and conceptual change: Towards a systemic
model of the processes of change. Learning and Instruction, 14, 519–536.
Mikkilä-Erdmann, M. (2002). Textbook text as a tool for promoting conceptual change in science.
Doctoral dissertation. University of Turku, Annales Universitateis Turkuensis B 249.
Patel, V. L., & Groen, G. J. (1991). The general and specific nature of medical expertise: A critical
look. In: K. A. Ericsson, & J. Smith (Eds), Toward a general theory of expertise: Prospects and
limits (pp. 93–125). Cambridge: Cambridge University Press.
Posner, G., Strike, K., Hewson, P., & Gertzog, W. (1982). Accommodation of a scientific conception:
Toward a theory of conceptual change. Science Education, 66, 211–227.
Rudin, W. (1953). Principles of mathematical analysis. London: McGraw-Hill.
Sfard, A. (1991). On the dual nature of mathematical conceptions: Reflections on processes and
objects as different sides of the same coin. Educational Studies in Mathematics, 22(1), 1–6.
Sfard, A. (1994). Reification as a birth of a metaphor. For the Learning of Mathematics, 14(1),
44–55.
Sfard, A. (1998). On two metaphors on learning and the dangers of choosing just one. Educational
Researcher, 27(2), 4–13.
Sierpinska, A. (1987). Humanities students and epistemological obstacles related to limits.
Educational Studies in Mathematics, 18, 371–397.
Spelke, S. E. (1991). Physical knowledge in infancy: Reflections on Piaget’s theory. In: S. Carey, &
R. Gelman (Eds), The epigenesis of mind: Essays on biology and cognition (pp. 133–170).
Hillsadale, NJ: Erlbaum.
cal value of fractions. Learning and Instruction, 14, 503–518.
Strike, K., & Posner, G. (1982). Conceptual change in science teaching. European Journal of Science
Education, 4, 231–240.
Szydlik, J. E. (2000). Mathematical beliefs and conceptual understanding of the limit of a function.
Journal for Research in Mathematics Education, 31(3), 258–276.
Trumper, R. (1997). Applying conceptual conflict strategies in the learning of the energy concept.
Research in Science & Technological Education, 15, 5–19.
Vosniadou, S. (1994). Capturing and modelling the process of conceptual change. Learning and
Vosniadou, S. (1999). Conceptual change research: State of art and future directions. In: W. Schnotz,
S. Vosniadou, & M. Carretero (Eds), New perspectives on conceptual change (pp. 3–14). Oxford:
Elsevier.
Vygotsky, L. S. (1978). Mind in society: The development of higher psychological processes.
Cambridge, MA: Harvard University Press.
Williams, S. R. (1991). Models of limit held by college students. Journal for Research in
Mathematics Education, 22(3), 219–236.
Wiser, M., & Amin, T. (2001). “Is heat hot?” Inducing conceptual change by integrating everyday
Chapter 18
How Many Numbers are there in a Rational

Numbers Interval? Constraints, Synthetic
Models and the Effect of the Number Line
Xenia Vamvakoussi and Stella Vosniadou
The Set of Rational Numbers: An Expansion of the

Natural Numbers Set?
A rational number is a number that can be expressed as the ratio of two integers, like 1/2,
!2/3 etc. Alternatively, a rational number can be expressed either as a simple decimal or
as a recurring decimal, like 0.23, !0.4, 0.3454545…, !0.333…, etc. It follows that all
natural numbers are included in the rational numbers set, since 2, for instance, can be rep-
resented as 2/1 or 2.0. The same holds for any negative whole number, since !3, for
instance, can be represented as !3/1 or !3.0. The rational numbers set is closed under
subtraction and division, in the sense that the difference and the quotient of any rational
number are rational numbers as well. This is a property that does not hold within the nat-
ural numbers set: For instance, the outcomes of 3 ! 5 or 3:5 are not natural numbers. The
need for closure under subtraction and division offers a plausible explanation for the need
to expand the set of natural numbers to the set of rational numbers. This expansion can be
described linearly in the following way: The natural numbers set is expanded to the set of
integers to comprise the negative whole numbers. The new set is closed under subtraction,
e.g., 3 ! 5 " !2 is an integer number. The set of integers is then expanded to the set of
rational numbers; the new set is closed under subtraction and division, e.g., 3 ! 5, 3:5 are
rational numbers.
This characterization of the shift from natural to rational numbers may represent a good
way to make a plausible summary of the historical development of the number concept.
However, it does not accurately reflect the historical process. More importantly, it conceals
the fact that the historical development of the rational number concept was far from being
linear and smooth. There were harsh debates in the history of mathematics about the kind
of mathematical constructs that can be considered as numbers and the kinds of operations
that are legitimate (see also Merenluoto & Palonen, this volume). In a more general fash-
ion, the shift from natural to rational numbers involved changes in the status and meaning

ISBN: 0-08-045355-4
266 Xenia Vamvakoussi and Stella Vosniadou
of the term “number” that cannot be accounted for in terms of the mere expansion of the
natural number concept. For example, to accept negative numbers as numbers meant that
the conceptualization of numbers as answers to the questions “How many?” and “How
much?” for discrete and continuous quantities, respectively, had to be radically changed
(Dunmore, 1992).
From a different perspective, the shift from accepting that 3 ! 5 is a legitimate opera-
tion to accepting that 3 ! 5 is a mathematical object can be viewed as an ontological shift
during which a process now becomes a member of the category of objects (Sfard, 1991).
The same holds for the shift in the conceptualization of the ratios of integers from relations
between numbers to numbers.
What we are trying to say here is that the change from natural to rational numbers can-
not be accounted for as a mere expansion of the natural numbers set, in the sense that there
is more involved than just “adding” new numbers to it. Along with the changes mentioned
above, the two sets also have radically different structures: The set of natural numbers is
discrete, i.e., between two successive natural numbers there is no other natural number,
whereas the set of rational numbers is dense, i.e., between any two, non-equal rational
numbers, there are infinitely many numbers.
In this chapter, we investigate secondary and upper secondary students’ understanding
of this particular property of the rational numbers set, namely density. Understanding den-
sity is closely related to the development of the rational number concept. We argue that stu-
dents’ initial explanatory frameworks for number are tied around natural numbers and their
basic properties, such as discreteness. When students are introduced to other kinds of num-
bers, like fractions and decimals, they have to deal with constructs that took millennia to
develop. Although students do not have to recapitulate the historical growth of these math-
ematical ideas, they still face the challenging task to construct meaning for these new
mathematical objects called “numbers” and, in particular, their notation. Moreover, while
moving to a wider explanatory framework for number, students have to realize the confin-
ing role of the presuppositions pertaining to their initial framework, which are bound to
interfere in the interpretation of new information about rational numbers. We argue that the
shift to wider explanatory frameworks for number cannot be accomplished by a mere
enrichment of the initial frameworks for number, but requires conceptual change. That is,
it requires changes in the structure of the number concept, in the fundamental presupposi-
tions within which it is embedded, as well as change in the contexts of its use.
Initial Explanatory Frameworks for Number
It has been argued that even before instruction children form a principled understanding
of number that has its roots in the act of counting (Gelman, 2000; Gelman & Gallistel,
1978). During the first years of instruction, this initial understanding of number supports
children to reason about natural numbers and to learn about their properties, to build strate-
gies in relation to natural numbers operations, etc. For instance, the counting algorithm
supports children to build the successor principle (e.g., that given a specific natural num-
ber, one can always find its next), which in turn may help them to infer that there are infi-
nitely many natural numbers (Hartnett & Gelman, 1998). Addition and subtraction are
How Many Numbers are there in a Rational Numbers Interval? 267
conceptualized in terms of counting [see, for example, Resnick (1989) for a report of
counting-based strategies], multiplication is viewed as repeated addition (e.g., Fischbein,
Deri, Nello, & Marino,, 1985), and numbers can be ordered by means of their position on
the count list.
In terms of the conceptual change theoretical framework that we adopt (Vosniadou, this
volume; Vosniadou & Verschaffel, 2004), this complex network of relations and operations
constitutes young children’s initial explanatory framework of number (see also Smith,
Solomon, & Carey, 2005). The conceptual change framework predicts that when children
will encounter numbers with different properties than natural numbers, their initial
explanatory framework for number will stand in the way of further learning.
Indeed, there is a great deal of evidence showing that students at various levels of
instruction make use of their knowledge of natural number to conceptualize rational num-
bers and make sense of decimal and fraction notation, often resulting in making system-
atic errors in ordering, operations, and notation of rational numbers (e.g., Fischbein et al.,
1985; Moskal & Magone, 2000; Resnick et al., 1989; Stafylidou & Vosniadou, 2004;
Yujing & Yong-Di, 2005). Many researchers attribute these difficulties to the constraints
students’ prior knowledge about natural numbers imposes on the development of the
rational number concept. Such findings and interpretations are compatible with the con-
ceptual change approach and constitute evidence of the confining role of students’ initial
explanatory frameworks for number. In this chapter, we will test the predictive power of
this framework in relation to students’ understanding of the dense structure of the rational
numbers set, which has not been extensively investigated so far, as pointed out also by
Smith et al. (2005).
Understanding the Structure of the Set of Rational Numbers: A

Conceptual Change Approach
To understand the structure of the rational numbers set, students must (a) realize that dis-
creteness is a property of natural numbers which is not preserved in the rational numbers
set, and (b) be aware of the fact that the rational numbers set consists of elements that can
be represented either as fractions or as decimals, yet remain invariant under different sym-
bolic representations.
We assume that the discreteness of numbers is a fundamental presupposition (Vosniadou,
1994, 2001) of the initial explanatory framework of number, within which “orderability” is
not differentiated from “nextness” (Greer, 2004). Research in the area of mathematics edu-
cation has provided evidence that the idea of discreteness of numbers is a barrier to under-
standing the dense structure of the rational and real numbers set for students at various
levels of instruction (Malara, 2001; Merenluoto & Lehtinen, 2002, 2004; Neumann, 1998)
as well as for prospective teachers (Tirosh, Fischbein, Graeber, & Wilson, 1999).
Understanding that a rational number can be represented in multiple ways, but that its
different symbolic representations refer to the same mathematical entity, requires the abil-
ity to move flexibly among different symbolic representations of rational numbers. This
task, although easy for a mathematically versed person, is very challenging for the novices
in mathematics. As Markovits and Sowder (1991) argue, one of the major difficulties faced
by learners in mathematics is the use of different symbols to represent the same ideas. In
the case of rational numbers notation, there is evidence to suggest that students interpret
different symbolic representations to refer to different mathematical entities. This phe-
nomenon has been noted by Khoury and Zazkis (1994) and reported by O’Connor (2001)
as a fact noticed by mathematics teachers.
In our previous work (Vamvakoussi & Vosniadou, 2004), we found evidence that stu-
dents treated different symbolic representations as if they were different numbers in their
attempts to describe the structure of rational numbers intervals. For example, during an
interview, a 9th grader stated that there are infinitely many numbers between 3/8 and 5/8
and she mentioned a number of fractions, all of which were equivalent to 4/8, such as
4.0/8, 8/16, etc., indicating her belief that these symbols represented different numbers.
Another student said that there is no other number between 3/8 and 5/8 and he explained
that 4/8 can be simplified to 1/2, and 1/2 is not between 3/8 and 5/8. This student, although
he “knew” that 1/2 is equivalent to 4/8, failed to assign to it the properties of 4/8. In this
sense, he considered it to be a different mathematical object.
The belief that different symbolic representations stand for different numbers may have
another implication regarding students’ thinking about the structure of the rational num-
bers set. We suggest that students may consider fractions and decimals to be different,
unrelated “sets” of numbers. This assumption is compatible with evidence coming from
research in various domains showing that novices tend to categorize objects on the basis
of superficial characteristics (e.g., Chi, Feltovich, & Glaser, 1981). In the case of rational
numbers, this tendency may be enhanced by the fact that there are considerable differences
between the operations, as well as between the ordering of decimals and fractions.
Thinking of the rational numbers set as consisting of different, unrelated sets of numbers
is bound to constrain students’ understanding of its dense structure. For instance, Neumann
(1998) reports that 7th graders had difficulties accepting that there could be a fraction
between 0.3 and 0.6. In a previous study, Vamvakoussi and Vosniadou (2004) found that
students tend to think differently for fractions than for decimals with respect to their struc-
ture. For instance, a student stated that there is a finite number of numbers between two
given decimals, but infinitely many numbers between two given fractions.
To summarize, children’s initial explanatory frameworks of number are tied around nat-
ural numbers. In elementary school, children are introduced to new “kinds” of numbers,
i.e., fractions and decimals and later on to negative numbers as well. Accepting that these
novel constructs are full-fledged numbers is a major conceptual shift per se (see, for exam-
ple, Sfard, 1991). But let us focus on the expansion of numbers, from a learner’s point of
view. To master the rational number concept (at the level required in school context), stu-
dents must move from their initial to a wider explanatory framework for number, realizing
that certain presuppositions that held before, such as the discreteness of numbers, are only
valid within specific contexts. In developing this wider framework, it does not suffice to
accept that the term “number” refers equally well to natural numbers, fractions, and deci-
mals; one must also realize that fractions and decimals (simple and recurring),1 despite
1
Here and in the following we will use the term “decimal number” with reference only to the decimal numbers
that belong to the rational numbers set.
their differences in notation, ordering, operations, and contexts of use, are alternative rep-
resentations of rational numbers and not different kinds of numbers. This requires the abil-
ity on the part of the learner to take different perspectives on fractions and decimals,
according to the context of use, while being able to think of them as interchangeable rep-
resentations of the same mathematical entities, namely rational numbers. According to the
conceptual change framework, the construction of this wider explanatory framework for
number, accompanied by the deposition of natural number of its privileged position and
the development of different perspectives with respect to rational number notation, cannot
but be difficult, gradual, and time-consuming; moreover, in the process, misconceptions
will appear. In particular, it is predicted that students are bound to generate synthetic mod-
els of the structure of the set of rational numbers, which are caused when learners assimi-
late aspects of the new, incompatible information in their existing knowledge.
While examining the development of the number concept, it is essential to take into con-
sideration the artifacts and tools available in school context to support rational and real num-
ber reasoning. Indeed, there is a general agreement that external representations have the
potential to facilitate students to grasp mathematical ideas. With respect to numbers, students
are presented with external representations, such as Venn diagrams and the number line. In
particular, the number line, being itself continuous, appears to be a good external represen-
tation to convey the idea of density of numbers. However, some mathematics educators argue
that the meaning of a mathematical idea is not necessarily carried by a more concrete repre-
sentation (e.g., Clements & McMillen, 1996). This view has its counterpart in the conceptual
change literature, where there is an ongoing discussion about the effect of external represen-
tations, such as the globe or the map, on students’ reasoning and understanding — in this
case, about the shape of the Earth (Ivarsson, Schoultz, & Säljö, 2002; Schoultz, Säljö, &
Wyndhamn, 2001; Vosniadou, Skopeliti, & Ikospentaki, 2005). According to Vosniadou and
her colleagues, the effect of external representations should be analyzed within a construc-
tivist framework, where the external representations are themselves interpreted on the basis
of students’ prior knowledge. Among the aims of this study was to investigate the effect of
the number line on students’ reasoning about the density of rational numbers.
The Present Study
In the present study we investigated 9th and 11th graders’ understanding of the dense struc-
ture of the rational numbers set and the effect of the number line on their reasoning.
Understanding density is not an explicit goal of the Greek secondary school mathematics cur-
riculum. Still, students by grade 8 have already been taught everything they are supposed to
know about real numbers, including ordering, turning a decimal into a fraction and vice versa.
They also have dealt with tasks that implicitly refer to density. For instance, in the 7th grade
book, students are asked to find a number between two fractions with common denominators
and successive numerators. Eighth graders are explicitly taught how to approximate !"2 with two
rational numbers. Moreover, in grade 10, students are introduced, through Venn diagrams, to the
interrelations between the various subsets of the real numbers. They also review everything they
are supposed to know about real numbers, including properties, operations, and ordering. From
grade 8 and on, students use the real number line to represent real numbers. The mathematics
teachers in the schools where this study was conducted told us that they had mentioned to the
students on several occasions that there are infinitely many numbers in an interval.
We assumed that the idea of discreteness, being a fundamental presupposition of initial
explanatory frameworks of number, is a major barrier on students’ understanding of the
dense structure of the set of rational numbers. We also assumed that new knowledge of
fractions and decimals is used by students in their attempts to describe rational number
intervals, but it is constrained by their belief that different symbolic representations refer
to different numbers, and its implications that were discussed in the introduction.
More specifically, we expected that:
• The performance of 11th graders would be better than 9th graders’. Still, the answer that
there is a finite number of numbers in a rational number interval would appear fre-
quently in both age groups.
• The presence of the number line would have a helpful but limited effect on students’ per-
formance in tasks related to density.
• Students would perform better in forced-choice items, than in open-ended items.
• Understanding the structure of the set of rational numbers would be a slow and gradual
process, and not an “all or nothing” situation: We assumed that there would be inter-
mediate levels of understanding and that students at these levels would form synthetic
models of the structure of rational numbers intervals. We expected these models to
reflect the presupposition of discreteness and the belief that different symbolic repre-
sentations stand for different numbers.
Method
Participants
The participants of this study were 301 students, of which 164 were 9th graders and 137
were 11th graders (approximate age 15 and 17 years, respectively). They came from five
different schools in the Athens area. Almost half of our participants were girls.
Materials
We designed two types of questionnaires: an open-ended (QT1) and a forced-choice one

(QT2). Both types of questionnaires had two parts, each consisting of three questions. One
part made use of the number line while the other did not. Tables 18.1 and 18.2 show the
items in QT1 and QT2, respectively. All questions had a common, general form — indi-
cated as GF on the tables — which focuses on the number of numbers in a given rational
numbers interval.2
2
The number of numbers in an interval is not the only aspect of the property of density; there is also the question
of whether there is a “next number” to a given rational number. However, we chose to focus on this particular
aspect of the property, which we considered to be appropriate for our participants’ age and level of instruction.
Table 18.1: The open-ended questionnaire (QT1).
Items without the number line

GF Are there any numbers that are greater than a and, at the same time, less than b?
If yes, define how many/which these numbers are
If no, explain why
Q1 a " 0.005, b " 0.006
Q2 a " 3/8, b " 5/8
Q3 a " 0.001, b " 0.001
Items with the number line

(Number line present and number a already placed on the number line)
GF Place b on the number line
Are there any numbers that are greater than a and, at the same time, less than b?
If yes, define how many/which these numbers are
If no, explain why
Q4 a " 0.01, b " 0.02
Q5 a " 1/3, b " 2/3
Q6 a " 0.01, b " 0.1
Procedure
The students in each class were equally divided between the QT1 and the QT2 condition.
Half of them received the items with the number line first (NL1), whereas the other half
received first the items without the number line (NL2). In both cases, the first part of the
questionnaire was withdrawn, before the second part was administered. Students had one
school hour (about 45 minutes) to answer all six questions.
Results
Students’ responses to each of the six questions were marked as “Finite,” “Undefined,” and
“Infinite.” “Finite” refers to the response type “There is finite number of numbers,” while
“Infinite” refers to the response type “There are infinitely many numbers.” The answer
“Undefined” appeared only in the open-ended questionnaires when a student selected the
answer “Yes” to claim that there are numbers in a given interval but did not give any fur-
ther information regarding the number of numbers. Table 18.3 shows the percentage of
every type of answer, in the total of the answers given in all six questions, as a function of
grade [9th (GR1), 11th (GR2)] and type of questionnaire (QT1, QT2).
Table 18.2: The forced choice questionnaire (QT2).
Items without the number line

GF How many numbers are there, that are greater than a and, at the same time,
less than b?
i. There is no such number
ii. There are the following numbers:
iii. There are infinitely many decimals/fractions
iv. There are infinitely many numbers: simple decimals, decimals with infinitely many
decimal digits, fractions, square roots
v. None of the above. I believe that …
Q1 a " 0.005, b " 0.006
ii. 0.0051, 0.0052, 0.0053, 0.0054, 0.0055, 0.0056, 0.0057, 0.0058, 0.0059
Q2 a " 3/8, b " 5/8
i*. There is only one number, namely 4/8
ii*. There are infinitely many fractions, all
equivalent to 4/8, e.g., 8/16
Q3 a " 0.001, b " 0.01
ii. a. 0.002, 0.003, 0.004, 0.005, 0.006, 0.007, 0.008, 0.009
ii. b. 0.0011, 0.0012, 0.0013,…, 0.0097, 0.0098, 0.0099
Items with the number line

(Number line present and number a already placed on the number line)
GF Place b on the number line.
How many numbers are there, that are greater than a and, at the same time,
less than b?
i. There is no such number
ii. There are the following numbers:
iii. There are infinitely many decimals (or fractions)
iv. There are infinitely many numbers:
simple decimals, decimals with infinitely many
decimal digits, fractions, square roots
v. None of the above. I believe that …
Q4 a " 0.01, b " 0.02
ii. 0.11, 0.12, 0.13, 0.14, 0.15, 0.16, 0.17, 0.18, 0.19
Q5 a " 1/3, b " 2/3
1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9
ii. #, #, #, #, #, #, #, #, #
3 3 3 3 3 3 3 3 3
Q6 a " 0.01, b " 0.1

ii. a. 0.02, 0.03, 0.04, 0.05, 0.06, 0.07, 0.08, 0.09
ii. b. 0.011, 0.012, 0.013,..., 0.020, 0.021,..., 0.099
Table 18.3: Percentage of answer types, in the total of the answers given in all six ques-
tions, as a function of the type of questionnaire and of grade.
Open-Ended (QT1) Forced-Choice (QT2)
9th grade, 11th grade 9th grade 11th grade

(GR1) N " 498 (GR2), N " 396 (GR1) N " 486 (GR2) N " 426
(83 % 6) (66 % 6) (81 % 6) (71 % 6)
Finite 65% 41.7% 52% 34.5%

Infinite 11.9% 26% 42.7% 61.3%
Undefined 20.4% 28.2% – –
No answer 2.8% 4% 5.3% 4.2%
The Presupposition of Discreteness
As shown in Table 18.3, the answer “Finite” was the dominant answer for 9th graders, in
QT1 and QT2. The same holds for 11th graders, but only in QT1. In QT2, 11th graders gave
the “Infinite” answer more often. Still, about one-third of the answers were “Finite.” This
result shows that the presupposition of discreteness was strong in the 9th graders and
remained strong in the 11th graders too.
Differences by Age
In order to calculate each student’s overall performance in QT2, we scored the “Finite” and
“Infinite” answers as 1 and 2, respectively, and calculated the sum of the scores in all six
questions. In QT1, we scored the “Finite,” “Undefined,” and “Infinite” answers as 1, 2, and
3, respectively, and calculated the sum of the scores in all six questions. We scored the
answer “Undefined” higher than the “Finite” and lower than the “Infinite” answer, respec-
tively, assuming that:
• a student who could not define the number of numbers in an interval, yet took their exis-
tence for granted, should be scored higher than a student who answered that there are no
numbers in the same interval.
• a student who was able to characterize the number of numbers in an interval as “infi-
nitely many” should be scored higher than a student who did not use this expression.
Under both conditions (QT1, QT2), either when a student gave no answer or when he
gave an irrelevant answer, s/he was scored with 0.
We compared 9th and 11th graders’ performance in QT1 by performing a
Mann–Whitney test, which showed that 11th graders’ overall performance was signifi-
cantly better than 9th graders’ (z " !2.862, p $ .05). A second Mann–Whitney test
showed that 11th graders’ overall performance was significantly better than 9th graders’
(z " !2.537, p $ .05) in QT2 as well.
Effects of the Number Line
Examining the influence of the number line, we first tested for possible order effects. We
compared the performance of students who received the items with the number line first,
to the performance of those who received first the items without the number line (NL1 vs.
NL2). We performed four independent Mann–Whitney tests (GRi # QTj, i, j " 1, 2) that
showed no significant order effect. As a result, the order of presentation of the number line
was not taken into consideration in subsequent statistical analyses.
In order to test for possible effects of the presence of the number line on students’ per-
formance, we performed four independent Wilcoxon signed ranks tests (GRi % QTj, i, j "
1, 2). We found no significant difference in 11th graders’ performance in the presence of
the number line either in QT1 or in QT2. However, the presence of the number line
improved significantly the performance of 9th graders, in both QT1 (z " !2.467, p $ .05)
and QT2 (z % !3.395, p " .001).
In order to examine further the effect of the number line on students’ performance, we
separated the students into three categories: those who scored higher in the items with the
number line, those who scored lower in these items, and the remaining students who
scored the same in the items with and without the number line. Table 18.4 shows the num-
ber and percentage of students in each category for each grade and type of questionnaire.
As can be seen in Table 18.4, there are a considerable number of students whose per-
formance was not affected at all by the presence of the number line. Moreover, the num-
ber of 11th graders who scored higher in the questions with the number line is close to the
number of those who scored lower in these questions. Taking a closer look at our data we
find that in QT1, only 3 of the 21 9th graders whose overall performance was better for
questions with the number line had actually moved from “Finite” to “Infinite” answers; in
addition, only one of them gave the “Infinite” answer consistently in all questions with the
number line. In QT2, 20 of the 31 9th graders who performed better for questions with the
number line moved from “Finite” to “Infinite” answers. Only seven of these students gave
Table 18.4: Number and percentage of students in categories formed with respect to their
performance with and without the number line, as a function of grade and type of ques-
tionnaire.
Open-Ended (QT1) Forced-Choice (QT2)
Performance 9th grade 11th grade 9th grade 11th grade

(GR1), (GR2) (GR1), (GR2),
N = 83 N = 66 N = 81 N = 71
Better with the 21 16 31 16

number line (25.3%) (24.2%) (38.3%) (22.5%)
Worse with the 6 20 8 10
number line (7.2%) (30.3%) (9.9%) (14.1%)
No effect 56 30 42 45
(67.5%) (45.5%) (51.8%) (63.4%)
the “Infinite” answer consistently for all questions with the number line. In conclusion, it
appears that the presence of the number line had a limited effect on students’ performance.
Effect of Questionnaire Type
We categorized all students into three groups on the basis of their responses to all six ques-
tions. More specifically, we placed in the first group students who answered only “Finite.”
In the third group, we placed students who answered only “Infinite.” In the second, inter-
mediate group, we placed all remaining students. The first, second, and third group will be
referred to as “Finiteness,” “Mixed,” and “Infinity” category, respectively.
A ! 2 test showed that the effect of the questionnaire type was statistically significant
(df(2), p " .001). More specifically, students who filled in the open-ended questionnaires
were found more often in the category “Finiteness” and less often in the other two cate-
gories, as compared to students who filled in the forced-choice questionnaires.
Intermediate Levels of Understanding
In our attempts to examine intermediate levels of understanding of the structure of the

rational numbers set, we refined our categorization further and identified two subcate-
gories for the category “Finiteness”: Discreteness and Refined Discreteness, both in QT1
and in QT2. In the Discreteness category we placed students who considered the given
numbers to be successive, e.g., 0.005 to be immediately next to 0.006, in all six questions.
In the Refined Discreteness category, we placed students who did not consider the given
numbers to be successive, in at least one out of six questions. These students still answered
that there is a finite number of numbers in the interval, e.g., 0.0051, 0.0052, …, 0.0059
between 0.005 and 0.006.
In the case of forced-choice questionnaires, we were able to define two sub-categories for
the category “Infinity”: Constrained Density and Density. Students who answered that
there are infinitely many numbers of the same symbolic representation in the given inter-
val, for at least one out of the six questions, were placed in the “Constrained Density” cat-
egory. Students in this category were reluctant to accept, for example, that there may be
decimals between two fractions or vice versa. In the “Density” category, we placed the stu-
dents who answered that there are infinitely many numbers, regardless of their symbolic
representation, in all six questions. Almost half of the students of the “Infinity” category
were placed in the “Constrained Density” category.
Table 18.5 summarizes the percentage of students placed in the Discreteness, Refined
Discreteness, Mixed, Constrained Density, and Density categories, for both grades and
types of questionnaire.
As shown in Table 18.5, there were three intermediate levels of understanding, between
the naïve level of Discreteness and the more sophisticated level of Density. Students in the
Refined Discreteness category were constrained by the presupposition of discreteness.
Students in the Constrained Density category had overcome the barrier of discreteness, yet,
they were constrained by their belief that different symbolic representations stand for differ-
ent numbers, in the sense that they tend to group numbers according to their symbolic
representation. Finally, students in the Mixed category, who answered that there are infinitely
Table 18.5: Percentage of students placed in the five categories, as a function of type of
questionnaire and of grade.
Category Open-Ended Questionnaire Forced-Choice

Questionnaire
9th grade 11th grade 9th grade 11th grade

(N " 83) (N " 66) (N " 81) (N " 71)
Discreteness 30% 12.1% 4.9% 4.2%

Refined 16.9% 7.6% 30.9% 16.9%
discreteness
Mixed 43.4% 66.7% 40.7% 42.3%
Constrained 9.6% 13.6% 12.3% 15.5%
density
Density 11.1% 21.1%
many numbers in some, but not all questions, are still constrained by the idea of discreteness
and are also affected by the symbolic representation of numbers. Indeed, more than half
(53.9%) of the students in the “Mixed” category referred consistently to intervals with dif-
ferent structure, according to the symbolic representation of the first and the last number. For
instance, some answered that there are infinitely many numbers between two decimals but a
finite number of numbers between two fractions, or vice versa.
The belief that different symbolic representations refer to different numbers was
expressed also in a more explicit way. More specifically, in the case of forced-choice ques-
tionnaires, 12 out of 81 (15%) 11th graders and 14 out of 71 (19.70%) 9th graders answered
that there are infinitely many fractions between 3/8 and 5/8, all equivalent to 4/8. We also
noted that in the case of open-ended questionnaires, 8% of the students in the Discreteness
category (total 149 students), answered that there is no number between 3/8 and 5/8, proba-
bly because they simplified 4/8 to 1/2 and decided that it is not between 3/8 and 5/8.
Discussion
Our results agree with previous findings showing that the idea of discreteness is a barrier
to the understanding of the dense structure of the rational numbers set (Malara, 2001;
Merenluoto & Lehtinen, 2002, 2004; Neumann, 1998; Tirosh et al., 1999). They also agree
with findings showing that realizing that rational numbers remain invariant under different
symbolic representations is a major conceptual difficulty for students (e.g., Khoury &
Zazkis, 1994). These results are in full agreement with the predictions made within the
conceptual change framework adopted.
Understanding density appears to be a slow and gradual process, in the course of which
students are bound to generate synthetic models of the rational numbers intervals. Indeed,
with the exception of students in the Density category, our participants gave alternative
accounts of the structure of the given intervals. These accounts reflected the enrichment of
an initial explanatory framework of numbers with information about fractions and deci-
mals. However, within this wider explanatory framework, these new “kinds of numbers”
were considered as different, unrelated sorts of numbers; in addition, the presupposition of
discreteness appeared to play a confining role. In this sense, the accounts generated by our
participants can be interpreted as synthetic models of the structure of the rational numbers
intervals. In Table 18.6 there is a more detailed description of the synthetic models gener-
ated by our participants.
Table 18.6: Students’ accounts of the structure of rational numbers intervals: some examples.
Category Description Example
Discreteness Intervals that (0.005–0.006)

preserve the discrete
structure of natural
numbers
The initial numbers (1/3–2/3)
are considered successive
Refined discreteness Intervals that (0.0051, 0.0052,
preserve the discrete 0.0053,…,0.0059,
structure of natural 0.006)
numbers
The initial numbers (3.0/8, 3.1/8,…,
are not considered successive 4.0/8,…, 5/8)
Within the mixed Different structure,
category according to the (Decimal, infinitely
symbolic many decimals,
representation of the decimal) and
first and last numbers (fraction, finite
number of fractions,
fraction). Or vice versa
Intervals that contain (3/8, 4/8, 4.0/8,
“infinitely many” 8/16,…, 5/8)
equivalent numbers
Constrained density Intervals that contain (Decimal, infinitely
infinitely many many decimals,
numbers of the same decimal) or
symbolic representation (fractions, infinitely
many fractions,fractions)
Density Any interval contains
infinitely many numbers,
regardless of their symbolic
representations
Students in the Discreteness category treated the given numbers as if they were succes-
sive, thus preserving the discrete structure of the set of natural numbers.
Students in the Refined Discreteness category also preserved the discrete structure of
natural numbers. However, they made use of new knowledge about rational numbers to
give a more sophisticated answer. For instance, they appealed to the fact that adding one
decimal digit to 0.005, one gets a number that is bigger than 0.005 and still smaller than
0.006, to conclude that there are (a finite number of) numbers between 0.005 and 0.006.
Yet, their commitment to the idea of discreteness prevented them from using this knowl-
edge efficiently and infer that there can be even more numbers in the given interval.
Students who completed the forced-choice questionnaires were placed more often in the
Refined Discreteness category than in the Discreteness category. Apparently, while these
students were facilitated by the presence of more sophisticated answers, they were not able
to overcome the presupposition of discreteness and chose the answer “Infinite.” This
result, along with the finding that students gave more sophisticated answers in QT1, as
compared to QT2, is in accordance with findings from the conceptual change research,
which show that students perform better in forced-choice tasks (Vosniadou, Skopeliti, &
Ikospentaki,, 2004).
Students in the Constrained Density category were able to answer that there are infi-
nitely many numbers in all given intervals, but were reluctant to accept that these numbers
can have various symbolic representations. This type of synthetic model reveals students’
disposition to group numbers according to their symbolic representation, which reflects
the belief that different symbolic representations stand for different numbers.
Among the students in the Mixed Category, those who answered differently, according
to the symbolic representation of the first and last number of the interval, generated syn-
thetic models of the structure of rational numbers intervals, for example discrete structure
between fractions and dense structure between decimals. We suggest that these students
were also constrained by their belief that different symbolic representations refer to dif-
ferent numbers, with different properties. This suggestion is based on findings from our
previous work (Vamvakoussi & Vosniadou, 2004). For example, a 9th grader who partici-
pated in that study claimed that there are infinitely many numbers between 3/8 and 5/8, yet
there is a finite number of numbers between 0.001 and 0.01. When asked to consider both
his answers, he explained that if one turns the same decimals into fractions, one can find
more, infinitely many numbers between them.
The belief that the different symbolic representations of a number refer to different
numbers is explicitly expressed in the case of the students who answered that there are
“infinitely many” numbers between 3/8 and 5/8, all being fractions equivalent to 4/8. We
believe that the same belief drove some students to answer that there is no other number
between these fractions. Based on the findings of a previous study (Vamvakoussi &
Vosniadou, 2004) we suggest that these students converted 4/8 into 1/2, and then assumed
that 1/2 is not between 3/8 and 5/8, based on the difference between the symbolic repre-
sentations. If this is the case, these students, although they may know that 1/2 is equiva-
lent to 4/8, fail to assign to it the properties of 4/8. In this sense, they consider it to be a
different mathematical entity.
Finally, our findings with respect to the number line support the view that the meaning
of a mathematical idea is not necessarily carried by the mere presence of a more concrete
representation (e.g., Clements & McMillen, 1996). Our results indicate that the number
line, despite being continuous, does not facilitate considerably students’ reasoning about
the infinity of numbers in a rational number interval. This is because the number line
improved significantly only the performance of 9th graders, and not of the 11th graders.
Second, even in this case, the 9th graders who scored higher in the presence of the num-
ber line did not answer consistently that there are infinitely many numbers when the num-
ber line was present; in fact, some of them did not change any of their answers from
“Finite” to “Infinite.” Third, the order of appearance of the number did not have a signifi-
cant effect on students’ performance.
It is important to note that the effect of the number line was not always positive, espe-
cially with respect to 11th graders. This result, although not statistically significant, is
interesting and somewhat surprising: Taking a second look at the questions with the num-
ber line we noticed that they might be considered easier than the questions without the
number line. Indeed, the decimals involved in the first case have fewer decimal digits than
those that appear in the second case. Although further research is needed to clarify this
issue, we believe that this result might be a side effect of what Greek students are taught
in upper secondary school about the line in the context of Euclidean geometry. More
specifically, in 10th grade, students are introduced to the notion of a line as a set of points,
which is different from the “holistic” line (Nunez & Lakoff, 1998) and may convey the
idea that lines are discrete in nature. If the number line is interpreted on the basis of stu-
dents’ thinking about the geometrical line, as Vosniadou et al. (2005) would suggest, then
this could account to some extent for the decrease of 11th graders’ performance in the pres-
ence of the number line.
The findings of this study support the hypothesis that understanding the dense structure
of the rational numbers set requires a number of changes in students’ explanatory frame-
works for number. These changes involve objectifying rational numbers, conceptualizing
natural numbers as a subset of rational numbers, and assigning proper meaning to frac-
tional and decimal notation, which involves abandoning the belief that different symbolic
representations refer to different numbers; most importantly, students must stop thinking
about rational numbers in terms of natural numbers and their properties. These changes
cannot be accomplished by mere enrichment of the initial explanatory frameworks of num-
ber. Our view is very similar to the view of Smith et al. (2005), who argue that the shift in
younger children’s thinking about numbers is explained in terms of conceptual change. We
fully agree that
“conceptual changes involve differentiations and coalescences such that the

extension of a concept and its relations to other concepts are qualitatively
different after the change than before it.” (Smith et al., 2005, p. 133)
However, we do not believe that these differentiations and coalescences “commit the child
to concepts that would be incoherent in each side of the divide” (Smith et al., 2005, p. 133).
We want to stress that we do not see the antecedent and subsequent states as “incommensu-
rable,” in the sense that after the change children think of numbers in terms of rational num-
bers. Rather, we argue for the ability on part of the learner to move flexibly between natural
and rational number reasoning, according to the context of use. Learners who have not made
the transition to the wider explanatory framework for number are clearly not in a position to
entertain both perspectives. However, as it has been nicely illustrated in Merenluoto and
Palonen (this volume), mathematically versed persons not only do they show this ability, but
they are also fully aware of their different modes of thinking about numbers.
Conceptual change in the number concept cannot but be gradual and time consuming and
they become even more difficult if the design of instruction builds on the presupposition that
learning is accomplished through additive mechanisms. To give an example, in the Greek 8th
grade mathematics textbook, which is based on a procedure-oriented curriculum, the set of
rational numbers is presented in the following way: “All the numbers that we know, namely,
natural numbers, decimals, and fractions, together with the respective negative numbers, con-
stitute the set of rational numbers.” This “definition” certainly builds on students’ prior
knowledge about numbers, but at the same time, enhances students’ initial tendency to group
numbers on the basis of their symbolic representations and presents the rational numbers set
as consisting of different, unrelated sorts of numbers; in a more general fashion, it presents
the rational numbers set as an expansion of the natural numbers set, in the course of which
“new” numbers are being added to it. The fact that students are explicitly taught how to turn
a decimal into a fraction and vice versa does not necessarily support them in accomplishing
the corresponding conceptual knowledge, as suggested by the results of our study.
The rational number concept is notoriously difficult for students to develop and this fact
has been amply demonstrated in numerous studies. We suggest that the conceptual change
approach could make a useful theoretical tool in the attempt to synthesize such widespread
findings. In a more general fashion, it might be useful to look at mathematics curricula
through the lenses of the conceptual change approach and reconsider the sequence of
appearance of certain mathematical concepts (see also Vosniadou & Vamvakoussi, 2006).
According to Resnick (2006), the answer to the question “Is there a ‘best’ developmental
sequence for teaching mathematical concepts that will maximize positive effects of prior
mathematical learning and minimize interference?” is yet to come. Resnick also points out
that students should not be the only targets of conceptual change interventions — teachers
should also be taken into consideration. The first step towards this direction would proba-
bly be to inform teachers about the issue of conceptual change and stress the fact that an
expansion of a concept from a mathematics point of view may not correspond to an enrich-
ment of the prior knowledge of the learner.
Acknowledgments
The present study was funded through the program EPEAEK II in the framework of the
project “Pythagoras — Support of University Research Groups” with 75% contribution
from European Social Funds and 25% contribution from National Funds.
References
Chi, M. T. H., Feltovich, P. J., & Glaser, R. (1981). Categorization and representation of physics
Clements, D. H., & McMillen, S. (1996). Rethinking concrete manipulatives. Teaching Children
Mathematics, 2(5), 270–279.
Dunmore, C. (1992). Meta-level revolutions in mathematics. In: D. Gillies (Ed.), Revolutions in

mathematics (pp. 209–225). Oxford: Oxford University Press.
Fischbein, E., Deri, M., Nello, M., & Marino, M. (1985). The role of implicit models in solving prob-
lems in multiplication and division. Journal for Research in Mathematics Education, 16, 3–17.
Gelman, R., & Gallistel, C. R. (1978). The child’s understanding of number. Cambridge, MA:
Harvard University Press.
Greer, B. (2004). The growth of mathematics through conceptual restructuring. Learning &
Hartnett, P. M., & Gelman, R. (1998). Early understandings of number: Paths or barriers to the con-
struction of new understandings? Learning and Instruction, 18, 341–374.
Ivarsson, J., Schoultz, J., & Säljö, R. (2002). Map reading versus mind reading: Revisiting children’s
understanding of the shape of the earth. In: ?. Limon, & L. Mason (Eds), Reconsidering concep-
tual change: Issues in theory and practice (pp. 77–99). Dordrecht, The Netherlands: Kluwer.
Khoury, H. A., & Zazkis, R. (1994). On fractions and non-standard representations: Preservice teach-
ers’ concepts. Educational Studies in Mathematics, 27, 191–204.
Malara, N. (2001). From fractions to rational numbers in their structure: Outlines for an innovative
didactical strategy and the question of density. In: J. Novotná (Ed.), Proceedings of the 2nd
Conference of the European Society for Research Mathematics Education II (pp. 35–46). Praga:
Univerzita Karlova v Praze, Pedagogická Faculta.
Markovits, Z., & Sowder, J. (1991). Students’ understanding of the relationship between fractions
and decimals. Focus on Learning Problems in Mathematics, 13(1), 3–11.
Merenluoto, K., & Lehtinen, E. (2002). Conceptual change in mathematics: Understanding the real
numbers. In: M. Limon, & L. Mason (Eds), Reconsidering conceptual change: Issues in theory
and practice (pp. 233–258). Dordrecht, The Netherlands: Kluwer.
Merenluoto, K., & Lehtinen, E. (2004). Number concept and conceptual change: Outlines for new
teaching strategies. Learning and Instruction, 14, 519–534.
Moskal, B. M., & Magone, M. E. (2000). Making sense of what students know: Examining the ref-
erents, relationships and modes students displayed in response to a decimal task. Educational
Studies in Mathematics, 43, 313–335.
Neumann, R. (1998). Students’ ideas on the density of fractions. In: H. G. Weigand, A. Peter-Koop,
N. Neil, K. Reiss, G. Torner, & B. Wollring (Eds), Proceedings of the annual meeting of the
Gesellschaft fur Didaktik der Mathematik (pp. 97–104). Retrieved March 21, 2005 from
http://webdoc.sub.gwdg.de/ebook/e/gdm/1998/
Nunez, R., & Lakoff, G. (1998). What did Weirstrass really define? The cognitive structure of natu-
ral and ?–? continuity. Mathematical Cognition, 4, 85–101.
O’Connor, M. C. (2001). “Can any fraction be turned into a decimal?” A case study of a mathemat-
ical group discussion. Educational Studies in Mathematics, 46, 143–185.
Resnick, L. B. (1989). Developing mathematical knowledge. American Psychologist, 44(2),
162–169.
Resnick, L. B. (2006). The dilemma of mathematical intuition in learning. In: J. Novotná,
H. Moraová, M. Krátká, & N. Stehlíková (Eds), Proceedings of the 30th Conference of the
International Group for the Psychology of Mathematics Education (Vol. 1, pp. 173–175). Prague:
PME.
Resnick, L. B., Nesher, P., Leonard, F., Magone, M., Omanson, S., & Peled, I. (1989). Conceptual
bases of arithmetic errors: The case of decimal fractions. Journal for Research in Mathematics
Schoultz, J., Säljö, R., & Wyndhamn, J. (2001). Heavenly talk: Discourse, artefacts, and children’s
understanding of elementary astronomy. Human Development, 44, 103–118.
Sfard, A. (1991). On the dual nature of mathematical conceptions: Reflections on processes and
objects as different sides of the same coin. Educational Studies in Mathematics, 22, 1–36.
Smith, C. L., Solomon, G. E. A., & Carey, S. (2005). Never getting to zero: Elementary school stu-
dents’ understanding of the infinite divisibility of number and matter. Cognitive Psychology, 51,
101–140.
Stafylidou, S., & Vosniadou, S. (2004). Students’ understanding of the numerical value of fractions:
A conceptual change approach. Learning and Instruction, 14, 503–518.
Tirosh, D., Fischbein, E., Graeber, A. O., & Wilson, J. W. (1999). Prospective elementary teachers’con-
ceptions of rational numbers. Retrieved May 05, 2005 from http://jwilson.coe.uga.edu/Texts.Folder/
Tirosh/Pros.El.Tchrs.html
Vosniadou, S., & Vamvakoussi, X. (2006). Examining mathematics learning from a conceptual
change point of view: Implications for the design of learning environments. In: L. Verschaffel,
F. Dochy, M. Boekaerts, & S. Vosniadou (Eds), Instructional psychology: Past, present and future
trends. Sixteen essays in honour of Erik De Corte (pp. 55–70). Oxford: Elsevier.
Vosniadou, S. (1994). Capturing and modelling the process of conceptual change. Learning and
Vosniadou, S. (2001). On the nature of naïve physics. In: M. Limon, & L. Mason (Eds),
Reconsidering conceptual change: Issues in theory and practice (pp. 61–76). Dordrecht, The
Vosniadou, S., & Verschaffel, L. (2004). Extending the conceptual change approach to mathematics
learning and teaching. Learning and Instruction, 14, 445–451.
Vosniadou, S., Skopeliti, I., & Ikospentaki, K. (2004). Modes of knowing and ways of reasoning in
elementary astronomy. Cognitive Development, 19, 203–222.
Vosniadou, S., Skopeliti, I., & Ikospentaki, K. (2005). Reconsidering the role of artifacts in reason-
ing: Children’s understanding of the globe as a model of the Earth. Learning and Instruction, 15,
333–351.
Yujing, N., & Yong-Di, Z. (2005). Teaching and learning fraction and rational numbers: The origins
and implications of whole number bias. Educational Psychologist, 40(1), 27–52.
Chapter 19
Students’ Interpretations of Literal Symbols

in Algebra
Konstantinos P. Christou, Stella Vosniadou
and Xenia Vamvakoussi
Large-scale research both in Great Britain (Hart, 1981) and in the United States
(Carpenter, Corbitt, Kepner, Lindquist, & Reys, 1981) has shown that students have great
difficulties in understanding algebra. Many students mention specifically the use of literal
symbols as the origin of their difficulties, saying that they understood mathematics until
literal symbols appeared (Sackur, 1995). In the present study, the conceptual change
approach will be used as an explanatory framework for better understanding students’
difficulties in interpreting the use of literal symbols as variables in algebra.
The Conceptual Change Approach
The conceptual change approach focuses on knowledge acquisition in specific domains

and describes learning as a process that sometimes requires the significant reorganization
of existing knowledge structures and not just their enrichment (Vosniadou, 1999;
Vosniadou & Brewer, 1992). According to this approach, by the time formal education
starts, students have already constructed a commonsense understanding of the world based
on their everyday experiences and the influence of the culture. This prior knowledge can
stand in the way of acquiring new information when the new learning content is incom-
patible with what is already known. In these cases, the acquisition of new information
requires conceptual change. Conceptual change is more difficult than learning that can be
accomplished through enrichment. When students use additive mechanisms to assimilate
incompatible information to what they already know (enrichment) they produce a type of
misconception, which can be explained as “synthetic models” (Vosniadou, this volume).
Recently the conceptual change approach has been applied to the learning and teach-
ing of mathematics (see Verschaffel & Vosniadou, 2004) with most of the relevant stud-
ies focusing on the development of the number concept. Prior research has indicated
that students’ principle understanding of numbers is grounded on the natural numbers

ISBN: 0-08-045355-4
284 Konstantinos P. Christou et al.
(Gelman, 2000). Many students even in secondary education tend to project their number
concept grounded on natural numbers onto a non-natural number input. This seems to be
one of the reasons why misconceptions and difficulties appear when numbers other than
natural, such as rational numbers, are introduced in the mathematics curriculum. For
example, many of the errors students make in the case of fractions can be interpreted to
be caused by the application of properties of natural numbers to fractions (e.g., Gelman,
2000; Stafylidou & Vosniadou, 2004). Vamvakoussi and Vosniadou (2004) argue that the
presupposition of discreteness, which is a property that characterizes the natural numbers,
constrains students’ understanding of density, which is a property of rational numbers.
Other research also indicates that prior knowledge of natural numbers hinders students’
understanding of the properties of rational numbers (Resnick et al., 1989). Students’ prior
experience based on calculating only with natural numbers is considered to be responsi-
ble for students’ belief that “multiplication always makes the number larger”. This belief
in turn hinders students’ understanding of calculations when real numbers are involved
(Fischbein, Deri, Nello, & Marino, 1985).
The purpose of the present study is to examine students’ difficulties to interpret the use
of literal symbols in algebra. Literal symbols are used in many ways in algebra: They are
used to stand for mathematical objects such as functions, matrixes, etc., but they are mostly
used to represent the concept of variable. A variable is a mathematical entity that can be
used to represent any number within a range of numbers and can stand on its own right in
the algebraic formal system.
We hypothesized that students’ prior knowledge about the way numbers are used in the
context of arithmetic is likely to affect their interpretation of the use of literal symbols in
algebra. Findings from previous research are consistent with this hypothesis. In the next
session we discuss some of the most important relevant findings.
Research on Students’ Interpretations of the Use of Literal Symbols

in Algebra
Previous studies have demonstrated a series of misconceptions students have in relation to

the use of literal symbols in algebra. For example, students often view literal symbols as
labels for objects, i.e., they think that ‘D’ stands for David, ‘h’ for height, or they believe
that ‘y’ — in the task “add 3 to 5y” — refers to anything with a ‘y’ like a yacht, a yoghurt
or a yam. Alternatively, when students think of literal symbols as numbers they usually
believe that they stand for a specific number only (Collis, 1975; Booth, 1984; Knuth,
Alibali, McNeil, Weinberg, & Stephens, 2005; Kuchemann, 1978, 1981; Stacey &
MacGregor, 1997). These misconceptions appear to be quite strong and difficult to change.
Booth (1982, 1984) designed a teaching experiment specifically to encourage the acquisi-
tion of the interpretation of the literal symbol as generalized number, but found that stu-
dents still faced great difficulties, even when specific instructions were given to them.
Another difficulty students appear to have with the use of literal symbols in algebra
is known as ‘the lack of closure’ error (Collis, 1975), which refer to students’ unwilling-
ness to accept algebraic expressions that contain literal symbols as final answers.
Students’ Interpretations of Literal Symbols in Algebra 285
When presented with tasks such as “write the number which is twice as big as x”, students
are unwilling to accept “2x” as a final answer (Booth, 1984; Firth, 1975), but insist on
replacing it with a specific number. Based on such findings Booth (1984) suggested that
students consider mathematics to be an empirical subject, which requires the production
of numerical answers only.
The ‘concatenation problem’ presents another difficulty that students have with the use
of literal symbols in algebra. In arithmetic, concatenation denotes implicit addition, as
both in place-value numeration, e.g., 27 equals 20 plus 7, and in mixed-numeral notation,
e.g., 2 ½ means 2 plus ½ (Matz, 1980). On the contrary, in algebra, concatenation denotes
multiplication; 2a means 2 times a. When asked to substitute 2 for a in 3a, students tend
to interpret concatenation as it is used in arithmetic responding with 32. It is only when
asked specifically to respond as ‘in algebra’ that some students reply with ‘3 times 2’
(Chalouh & Herscovics, 1988).
Students’ tendency to think that literal symbols can stand for objects, names of objects,
or specific numbers only was originally explained in terms of the Piagetian theory. In other
words, it was argued that the students had not yet reached the stage of formal operations
and, therefore, they were not capable of acquiring the concept of variable (e.g., Collis,
1975; Kuchemann, 1978, 1981). Other researchers noticed that there might be an interac-
tion between students’ knowledge of arithmetic and their attempts to learn new content in
algebra. For example, Booth (1984, 1988) suggested that students’ difficulties in algebra
may be partly due to their problematic knowledge in arithmetic. As the students construct
their algebraic notions on the basis of their experience in arithmetic, erroneous arithmeti-
cal knowledge can be transferred in algebra. On the contrary, Matz (1980) argued that stu-
dents’ difficulties with algebra are not necessarily due to problematic knowledge of
arithmetic, but rather reflect inappropriate use of the properties of arithmetic to interpret a
new field in mathematics.
A similar, albeit more elaborated account than that of Matz’s was offered by Kieran
(1992). According to Kieran, most of the problems students have in their introduction to
algebra arise because of the shift to a set of conventions different from those used in
arithmetic. For example, in arithmetic, letters can be used as labels: ‘m’ can be used to
denote meters, monkeys, etc., and 12 m can mean 12 meters, that is, 12 times 1 meter. But
in algebra, 12 m can mean 12 times the number of meters. Algebra uses many of the
symbols used in arithmetic, such as the equal sign or the addition and subtraction signs,
but in different ways. For example, the equal sign is used in arithmetic as the ‘enter com-
mand’ for the result of a calculation to be announced. Kieran (1981) argues that the belief
of novices in algebra that the equal sign is a “do something signal” rather than a symbol
of the equivalence between the left and right sides of an equation is demonstrated by their
initial reluctance to accept statements such as 4 ! 3 " 6 ! 1.
Both Matz and Kieran suggest that students’ prior experience with arithmetic and the
fact that symbols have different roles in arithmetic compared to algebra can explain some
of students’ difficulties in algebra. This view is highly compatible with the conceptual
change approach. The conceptual change approach provides an explanatory framework
that can account for the previous findings and also make new and meaningful predictions
regarding students’ difficulties with the use of literal symbols in algebra.
The Conceptual Change Approach and the Use of Literal Symbols

in Algebra
When students are introduced to algebra, they face the difficult task of (a) assigning mean-
ing to new symbols and (b) assigning new meaning to old symbols, which were used in
the context of arithmetic. According to Resnick (1987), when students connect an alge-
braic expression1 with the ‘world of numbers’, they give a referential meaning to the
algebraic expression which can affect their performance in various mathematical tasks.
We suggest that a number of misconceptions that students have with the use of literal
symbols could be explained to result from the inappropriate transfer of prior knowledge
about numbers in the context of arithmetic, for the interpretation of literal symbols in alge-
bra. Because students’ explanatory frameworks of number are initially tied around natural
numbers (Gelman & Gallistel, 1978; Gelman, 2000), we will start by making explicit, in
Table 19.1, some of the important differences between the use of natural numbers in the
context of arithmetic and the use of literal symbols as variables in the context of algebra.
Table 19.1: Differences between the natural numbers in the context of arithmetic and
literal symbols in the context of algebra.
Natural Numbers Literal symbols as

in arithmetic variables in algebra
Form 1, 2, 3, … a, b, x, y, …
Sign Actual sign (positive) Phenomenal sign (positive
or negative)
Symbolic Representation Each number in the natural Each literal symbol corre-
number set has a unique sponds to a range of real
symbolic representation — numbers — different
different symbols correspond symbols could stand for
to different numbers. the same number.
Ordering —Density Natural numbers can be ordered Literal symbols in algebra
by means of their position on cannot be ordered by
the count list. There is always a means of their position on
successor or a preceding the alphabet. There is no
number. There are no numbers such thing as a successor
between two subsequent or preceding literal
numbers. symbols.
Relationship to the unit The unit is the smallest There is no smallest
natural number. number that can be
substituted for a variable,
unless otherwise specified.
1
An algebraic expression is a mathematical object that contains one or more variables together with other
symbols, such as numbers or signs.
As we can see in Table 19.1, natural numbers are expressed as a series of digits
(1, 2, 37, etc.), whereas variables are expressed as letters of the alphabet (a, b, x, y, etc.).
Natural numbers are positive numbers and as such they have no sign attached to them. On
the other hand, variables may have a phenomenal sign, which is the intuitively obvious
sign that the variable appears to have as a superficial characteristic of its form.
Mathematically speaking, the variable does not have a specific sign of its own. The values
of a variable are determined when specific numbers are substituted for the literal symbol.
Variables can stand for either positive or negative numbers independently of the phenom-
enal sign that is attached to them. For example, the variable which is represented by literal
symbol ‘x’ can stand for positive and also for negative numbers; ‘#x’ can stand for posi-
tive or negative numbers as well [this happens because #(#3)"3].
In the natural number set, every natural number has a unique symbolic representation
and different symbols represent different numbers. For example, the symbol of the natural
number ‘2’ stands only for the number ‘2’. On the other hand, in the algebraic notation, a
literal symbol could stand for a range of real numbers. For example, the literal symbol ‘x’
could stand for any real number such as 2, 4.555, ½, etc., and different literal symbols
could stand for the same number. The arithmetical value of a literal symbol is determined
by the number that it represents and for this reason literal symbols cannot be ordered with-
out reference to the numbers they stand for. For the same reason, a literal symbol does not
have a lowest arithmetical value unless the range of numbers that it represents has a lower
limit. Because literal symbols can stand for a range of real numbers they do not have any
special relation with the unit.
The conceptual change perspective predicts that incompatibility between the use of lit-
eral symbols in algebra and students’ prior knowledge about numbers, in particular natu-
ral numbers, could result in errors. These errors could be explained to originate in students’
tendency to use their prior experience with numbers in the context of arithmetic to inter-
pret literal symbols in algebra.
Findings of prior studies on students’ difficulties with the use of literal symbols in alge-
bra are consistent with this view. For example, some students believe that when the literal
symbol changes, then the value that it represents also changes (Booth, 1984; Kuchemann,
1981; Wagner, 1981). These students explain that ‘x ! y ! z’ can never equal ‘x ! p ! z’
because ‘different literal symbols must correspond to different numbers’. They are unwill-
ing to accept that different symbols could stand for the same value. However, this belief is
applicable to natural numbers, where each number has a unique symbolic representation
and where different symbols stand for different numbers.
There is also evidence that some students associate literal symbols with natural num-
bers, in the sense that they respond as if there is a correspondence between the linear order-
ing of the alphabet and that of the natural numbers system (Booth, 1984; Stacey &
MacGregor, 1997; Wagner, 1981). For example, they tend to assign the numerical value 8
to the literal symbol ‘h’ (used to represent a boy’s height), because ‘h’ is the eighth letter
in the alphabet. Or they say that 10 ! h " 18, and 10 ! h " R, because the tenth letter
after ‘h’ in the alphabet is ‘R’.
In this study, we hypothesized that prior experience with numbers, in particular with
natural numbers, would result in a strong tendency on the part of the students to interpret
literal symbols as standing mostly for natural numbers. We also hypothesized that it would
be difficult for students to understand that variables have a phenomenal sign, which is not
the actual sign of the values they represent. This hypothesis is based on the fact that in the
context of arithmetic, no sign implies positive value. This is a characteristic of natural
numbers that holds for all positive numbers. When students are introduced to negative
numbers they learn that the presence of the negative sign means ‘negative value’. We thus
predicted that students would tend to interpret ‘x’ to stand for positive numbers and ‘#x’
to stand for negative numbers, a phenomenon also noted by researchers such as Chiarugi,
Fracassina, and Furinghetti (1990) and Vlassis (2004). These hypotheses were investigated
in a series of empirical studies.
Previous work by Christou and Vosniadou (2005) investigated some of the above-men-
tioned hypotheses. They gave 8th- and 9th-grade students a questionnaire (Questionnaire
A, QR/A), which asked them to write down the numbers they thought could be assigned
to algebraic expressions such as ‘a’, ‘#b’, ‘4g’, ‘a/b’, ‘d+d+d’, etc. The results showed
that only about one-third of the students gave the mathematically correct response, namely
that ‘all numbers can be assigned to each algebraic expression’. When asked, for example,
to write down numbers they thought could be assigned to the algebraic expression ‘a’, 66%
of the students responded only with natural numbers. Natural numbers were mostly used
in the remaining questions as well. In most of their responses students substituted only
natural numbers for the literal symbols of the given algebraic expression and maintained
the form of the algebraic expression: fraction-like in the case of ‘a/b’, multiples of 4 in the
case of ‘4g’, natural numbers larger than 3 in the case of ‘k + 3’, etc. When asked to write
down numbers that can be assigned to ‘#b’, 72% of the students responded only with neg-
ative whole numbers (#1, #2, #3, etc.).
Again, we interpreted these responses to reflect students’ tendency to substitute only
natural numbers for the literal symbol ‘b’ and to maintain the phenomenal negative sign of
the given algebraic expression ‘#b’. Very few students answered the question by provid-
ing numbers other than natural numbers, such as decimals, fractions, negatives, or real
numbers. Students in this questionnaire were affected by the phenomenal sign of the alge-
braic expressions in the sense that they maintained it when they substituted numbers for
the literal symbols. The majority of the students assigned only positive numbers to the
positive-like algebraic expressions and negative numbers to the negative-like algebraic
expression ‘#b’. It could, therefore, be objected that the students responded with the first
numbers that came to their mind, in full knowledge that their answer was correct, since
all values can be assigned to any algebraic expression. However, these responses differ
in important ways from the responses expected from a mathematically sophisticated
participant.
In order to further explore this possible hypothesis, we designed a second open-ended
questionnaire (Questionnaire B, QR/B) in which the students were asked to write down
numbers that they thought could not be assigned to a set of algebraic expressions. The set
of algebraic expressions used was the same as the one used in QR/A. Unlike QR/A, in
QR/B there is only one correct response — namely, that “all numbers can be assigned to
each algebraic expression” or that “there are no such numbers that cannot be assigned to
each algebraic expression”.
The results obtained in QR/B showed that again only about one-third of the students
gave this mathematically correct response. About half of the students said that negative
whole numbers (#1, #2, #3, etc.) could not be assigned to the algebraic expression ‘a’
and that natural numbers (1, 2, 3, etc.) could not be assigned to ‘#b’. Similarly in the
remaining algebraic expressions, ‘4g’, ‘a/b’, ‘d!d!d’, and ‘k ! 3’, the students tended to
respond by replacing literal symbols only with negative whole numbers, while maintain-
ing the form of the algebraic expression.
For example, the majority of the students gave numbers such as (#1) ! (#1) ! (#1),
or (#2) ! (#2) ! (#2), as numbers that could not be assigned to the algebraic expres-
sion ‘d!d!d’, and numbers such as (#2)/(#3), and (#3)/(#4) as numbers that could not
be assigned to the fraction-like algebraic expression ‘a/b’. We suggest that these students
interpreted the literal symbols to stand only for natural numbers and thus they thought that
the additive inverses of natural numbers (the negative whole numbers) could not be sub-
stituted for the literal symbols. In QR/B students appeared again to be sensitive to the phe-
nomenal sign of the algebraic expressions which they seem to have interpreted as the
actual sign of the values that they represented. The majority of the students responded to
the questions by using numbers with the opposite sign of the phenomenal sign of the
algebraic expressions used. They said that negative numbers could not be assigned to
the positive-like algebraic expressions and that positive numbers could not be assigned
to the negative-like algebraic expression ‘#b’.
In order to further examine students’ tendencies to maintain or change the phenomenal
sign of the given algebraic expression as a function of the questionnaire type, students’
responses were subjected to a one-way ANOVA. Responses that maintained the phenom-
enal sign were marked as 1, responses that changed the phenomenal sign were marked as
2, and mathematically correct responses were marked as 3. The results showed main
effects for questionnaire type [F(1, 281) " 6.126, p $ 0.05], which were due to the fact
that students maintained the phenomenal sign in QR/A but changed it in QR/B. We inter-
preted students’ sensitivity to the phenomenal sign of the algebraic expression to be intri-
cately related to their belief that literal symbols in algebra stand only for natural numbers.
Students think of ‘#7x’, for example, as always negative and ‘7x’ as always positive
because they tend to think of the literal symbol ‘x’ as only standing for natural numbers.
In another ANOVA we examined the effect of students’ tendency to substitute only nat-
ural numbers vs non-natural numbers for the literal symbols themselves, independently of
the sign of the algebraic expression, in the two questionnaires. Responses that substituted
literal symbols only with natural numbers were marked as 1, responses that used non-
natural numbers were marked as 2, and mathematically correct responses were marked as
3. The results showed no significant differences between the two questionnaires. In both
questionnaires, students tended to substitute mostly natural numbers for the literal symbols
themselves and appeared unwilling to also present any fractions, decimal numbers, or real
numbers under any condition. This finding was consistent with the theoretical hypothesis
of the research, namely, that students tend to consider literal symbols in algebra to stand
for natural numbers only.
A possible criticism of our experiment could be that the students provided natural num-
bers not because they thought that these are the only ‘correct’ substitutions for the literal
symbols, but because these are the most common numbers, used both at school and in
everyday situations. In school mathematics, natural numbers are used in most of the pro
blems students are asked to solve and the solution to these problems, most of the time,
involves natural numbers only. Thus, it could be argued that students responded with nat-
ural numbers because they thought that this was what they should do and not because they
did not know that it is possible to substitute the literal symbols with non-natural numbers.
In order to further explore this possibility we designed another study in which we used
a forced-choice questionnaire. The advantage of a forced-choice questionnaire, in com-
parison to the open-ended ones used earlier, is that it presents students with specific alter-
natives that can include both natural and non-natural numbers. It can thus provide a more
rigorous test of the hypothesis that students interpret literal symbols in algebra to stand
only for natural numbers.
The Present Study
In this study, we constructed a forced-choice questionnaire (Questionnaire C, QR/C) that

presented students with a set of specific alternatives for the same algebraic expressions
used in Questionnaires A and B described earlier. These alternatives included both natural
and non-natural numbers such as negative integers, positive and negative fractions, and
positive and negative decimals. The correct response, namely that ‘all numbers can be
assigned’, was one of the alternatives. The students were asked to choose the alternatives
that they thought could not be assigned to the given algebraic expression. We used the neg-
ative substitution form (could not be assigned) because it is only in this condition that we
can say with certainty that only the mathematically correct response applies. In the posi-
tive substitution condition all responses can be considered to be technically correct.
If students indeed interpret literal symbols to stand only for natural numbers, they
should exclude some numbers from the given set, such as fractions, decimals, etc., depend-
ing on the form of the given algebraic expression. For example, given the algebraic expres-
sion ‘a/b’, they should think that only positive fractions could be assigned to it, and thus
that whole numbers or even decimal numbers could not be assigned. Alternatively, we
would expect that in the case of ‘#b’, they would tend to exclude all the positive numbers
of the given set of alternatives.
Method
Participants
Thirty-four children participated in this study. There were 8th and 9th graders (mean age
14.5 years old) from two middle-class high schools in Athens. All of them completed the
forced-choice questionnaire (QR/C).
Materials
QR/C consisted of the following six algebraic expressions: Q1: a, Q2: #b, Q3: 4g, Q4:
a/b, Q5: d!d!d, and Q6: k ! 3. For each algebraic expression, the students were asked
to choose from a given set of alternative numbers those they thought could not be assigned
to them. The set of alternatives consisted of 11 number choices, which included positive
and negative fractions, positive and negative decimals, positive and negative integers. The
twelfth alternative was always the correct response, namely, that all numbers can be
assigned to each algebraic expression. An example is shown in Table 19.2.
Procedure
The following instructions were given to the students: “In algebra, we use literal symbols
(such as a, b, x, y, etc.) mostly to stand for numbers. In this questionnaire we use such sym-
bols. Read the following questions carefully. If you think there are some numbers among
the given alternatives that cannot be assigned to the given algebraic expressions, please
place a circle around them. You may choose more than one numbers if you wish”. Students
completed the questionnaire in the presence of one of the experimenters and their mathe-
matics teacher in their classroom.
Results
Based on the findings from our previous studies, we created three main categories of
responses namely ‘NN/1’, ‘NN/2’, and ‘Phenomenal sign’. The category ‘NN/1’ attempted
to capture all responses that reflected the belief that literal symbols stand for natural num-
bers only (NN belief). For example, students who chose all numbers from the given alter-
natives except the positive fractions for the algebraic expression ‘a/b’, would be placed in
this category.
The category ‘NN/2’ captured responses that included some but not all alternatives pre-
dicted by the NN belief. For example, in the case of ‘a/b’, students could choose all the
integers of the given set of alternatives. These responses would be placed in this category.
In the ‘Phenomenal sign’ category we placed student’s responses that included all numbers
with the opposite sign from the phenomenal one. For example, in this category we placed
Table 19.2: An example of the way in which questions were posed in the forced choice
questionnaire.
Are there some numbers among the following alternatives that you think cannot be
assigned to 4g?
2
a) 6 e) 6.74 i) # %
3
5
b) 2 f) % j) 8
7
c) #0.25 g) 8 k) 2.333
d) –3 h) 4 l) No, all numbers can be
assigned to it.
students’ responses only with negative numbers in the case of ‘a/b’, or only with positive
numbers in the case of ‘#b’.
There were three additional categories of responses: ‘Non-systematic’, ‘No response’,
and ‘Correct’. The ‘Non-systematic’ category was used for all non-systematic responses.
The ‘No response’ category included null responses, and the ‘Correct’ category repre-
sented the correct alternative.
All responses were categorized by one of the experimenters, and a second rater scored
half of the responses using the same criteria. The agreement of the categorization was
96%. All disagreements were discussed until consensus was achieved. Tables 19.3, and
19.4 in more detail, presented the frequencies and the percentages of each category of
responses.
Table 19.3 presents the total percentage of students’ responses to all the questions for
each response category. Only 18.6% of the students’ responses represented the ‘correct
response despite the fact that it was an explicit alternative in all questions. One-third of stu-
dents’ responses (30.3%) were affected by the NN belief in the strict (22%, NN/1) or in a
more differentiated way (8.3%, NN/2). One-fourth of students’ responses (25.4%) were
affected by the phenomenal sign of the algebraic expressions. There was a large percent-
age of responses in the non-systematic category (16.1%) that could be explained by the
complexity and counterintuitiveness of the questions in QR/C, the fact that they were
expressed in the negative form, and, finally, the forced choice nature of the questionnaire.
Previous studies have also shown that non-systematic responses appear more frequently in
forced-choice questionnaires (see Vosniadou, Skopeliti, & Ikospentaki, 2004) than in
open-ended ones.
Table 19.4 presents in greater detail the frequencies and percentages for students’
responses in each category for the 6 algebraic expressions, together with examples of the
type of numbers/responses for each category. In the case of the algebraic expression ‘a’,
only about one-third of the students (29.4%) responded by selecting the correct response,
namely that ‘all values can be assigned to it’. The majority of the responses reflected the
belief that the literal symbol ‘a’ stands only for positive numbers (38.2%). Another 20.5%
of the responses indicated that numbers other than natural numbers could not be assigned
to ‘a’. In the remaining algebraic expressions, such as ‘4g’, ‘k ! 3’, or ‘a/b’, students’
responses appeared to be slightly different. The majority of the students were affected
Table 19.3: Percentages of students’ responses to the combined questions.
Questionnaire C (choose, from the given set of numbers, those that you think
cannot be assigned to the given algebraic expressions)
Categories Correct NN/1 NN/2 Phenomenal Non- No

Sign systematic Response
Mean 18.6% 22% 8.3% 25.4% 16.1% 9.3%

percentage 30.3%
more by the NN belief than by the phenomenal sign. For example, in the case of ‘a/b’,
32.3% of the responses were placed in the category NN/1, and 11.7% in the category
‘Phenomenal sign’. In the remaining algebraic expressions the results were similar with
the exception of ‘#b’. In this case, the results were quite different because of the pres-
ence of the negative sign. About half of the students’ responses (52.9%) indicated that stu-
dents interpreted this expression as standing for negative numbers only.
Table 19.4: Frequencies, percentages, and examples of students’ responses to each

question.
Questionnaire C (choose, from the given set of numbers, those that you think
cannot be assigned to the given algebraic expressions)
Questions Correct NN/1 NN/2 Phenomenal Non- No
Sign systematic Response
Q1: a All but Decimals, Negatives

NN fractions
10 7 3 13 1 #
(29.4%) (20.5%) (8.8%) (38.2%) (2.9%)
Q2: #b All but Natural Positives
negative numbers
wholes
6 2 2 18 5 1
(17.6%) (5.9%) (5.9%) (52.9%) (14.7%) (2.9%)
Q3: 4g All but Negatives
NN
7 9 # 5 10 3
(20.5%) (26.5%) (14.7%) (29.4%) (8.8%)
a
Q4: % All but Integers Negatives
b
positive
fractions
6 11 2 4 5 6
(17.6%) (32.3%) (5.9%) (11.7%) (14.7%) (17.6%)
Q5: d!d!d All but Fractions Negatives
NN
4 10 3 6 6 5
(11.7%) (29.4%) (8.8%) (17.6%) (17.6%) (14.7%)
Q6: ! + 3 All but Fractions Negatives
NN
5 6 7 6 6 4
(14.7%) (17.6%) (20.5%) (17.6%) (17.6%) (11.7%)
Discussion
The results from the present study are consistent with the findings from our previous study
(Christou & Vosniadou, 2005), and support the hypothesis of the conceptual change
approach that students’ interpretation of the use of literal symbols in algebra is strongly
influenced by their experience with numbers, in particular natural numbers, in the context
of arithmetic. The conclusion was based on two sources of evidence: first, students tended
to substitute only natural numbers for the literal symbols of the algebraic expressions, and
second, students interpreted the phenomenal sign of the algebraic expressions as the sign
of the numbers that they represent. This was the case, despite the fact that students were
taught in school that each literal symbol corresponds to any real number.
Students interpreted algebraic expressions such as ‘k ! 3’ or ‘d ! d ! d’ to stand
mostly for positive numbers and believed that negative numbers cannot be assigned to
them. The greatest influence of the phenomenal sign appeared in the case of the negative-
like algebraic expression ‘#b’, which the majority of the students interpreted as standing
for negative numbers only.
This finding is consistent with results from Vlassis’ (2004) research with polynomials,
where students appeared to consider the minus sign at the beginning of a polynomial as the
sign of a negative number. We interpret students’ belief that the phenomenal sign of the
algebraic expression is the actual one of the numbers that it represents to originate from
prior experience with arithmetic. In the context of arithmetic, numbers with “no sign”
means numbers with “positive value”, whereas numbers with “negative sign” means num-
bers with “negative value”. The transfer of this knowledge in the area of algebra causes a
misconception, which is strong even in the case of the older students and constrains the
acquisition of more advanced mathematical concepts such as, for example, the absolute
value of a number. For any real number a, the absolute value of a, denoted |a|, is always a
positive number, so |a| is equal to a, if a & 0 or to #a, if a $ 0. As the students are affected
by the phenomenal sign of the algebraic expressions they do not think of ‘–a’ as a symbol
that could possibly stand for a positive number when ‘a’ stands for a negative one
(see Chiarugi et al., 1990). In order for students to understand that the phenomenal sign of
an algebraic expression is not the sign of the numbers that it represents, they need to
reorganize their prior knowledge about numbers as shaped in the context of arithmetic.
Furthermore, students’ prior experience with numbers in the context of arithmetic con-
strains their understanding of the generalized nature of a literal symbol, i.e., as a variable
that stands for any real number. The present findings agree with the previous research
which shows that the initial understanding of number as natural number may hinder
the acquisition of more advanced mathematical concepts, as in the case of fractions,
rational numbers, or algebraic rules, etc. (Gelman, 2000; Resnick et al., 1989; Stafylidou
& Vosniadou, 2004; Vamvakoussi & Vosniadou, 2004).
Resnick (1987) has argued that algebraic expressions can take their meaning from their
position in the formal system of algebra. Over and above, there is also a referential mean-
ing, which algebraic expressions take either from the situations in which relations among
quantities and actions upon quantities play a role or from its connection with the ‘world of
numbers’. She noted that student’s capability to assign a referential meaning to the alge-
braic expressions affects their performance in algebraic transformation tasks. Focusing on
the referential meaning that can be assigned to the algebraic expressions with its linkage
to the world of numbers, we found that many students tend to think that literal symbols
represent only natural numbers and as a consequence they have a very restricted range of
numbers from which the algebraic expressions take their referential meaning. This affects
students’ performance in many mathematical tasks such as, for example, when students
have to estimate the sign or the value of an algebraic expression in situations where the
monotony of a function is tested, or in the case of a radicand.
We believe that we have provided some evidence that the conceptual change theoreti-
cal framework can help us systematize students’ errors in interpreting the use of literal
symbols in algebra. The results of the present study can also provide useful information
for the design of curricula and for instruction. It is important for teachers of algebra as well
as for the curriculum designers to be familiar with students’ beliefs and the possible rea-
sons for their mistakes when they use literal symbols in algebra, as well as in other
domains for example physics, chemistry, etc. (see, e.g., Heck, 2001; Sherin, 2001). Greer
(1994, 2005) has suggested various devices for expanding arithmetic operations beyond
natural numbers. He proposed to give students mathematical tasks, which use non-natural
numbers as factors, such as the equation 2.67x2 – 3.86x – 12.23 = 0, as this could help them
extend their conceptual fields beyond the natural numbers. Some researchers investigate
the implications of introducing algebraic thinking in elementary school (Carraher,
Schliemann, & Brizuela, 2001). Of course, more detailed empirical research is needed to
further investigate students’ difficulties and the effect of specific instructional innovations
before introducing them in schools.
Acknowledgments
The research was supported by a grant from the Program ‘‘Pythagoras — Support of
University Research Groups” (EPEAEK II), 75% from European Social Funds and 25%
from National Funds.
References
Booth, L. R. (1982). Developing a teaching module in beginning algebra, Proceedings of the 6th
conference of the international group for the psychology of mathematics education, Antwerp
(pp. 280–285).
Booth, L. R. (1984). Algebra: Children’s strategies and errors. Windsor, Berkshire: NFER-Nelson.
Booth, L. R. (1988). Children’s difficulties in beginning algebra. In: The ideas of algebra, K-I2, 1988
NCTM Yearbook (pp. 20–32). Reston, Virginia: National Council of Teachers of Mathematics.
Carpenter, T. P., Corbitt, M. K., Kepner, H. S. Jr., Lindquist, M. M., & Reys, R. E. (1981). Results
from the second mathematics assessment of the national assessment of educational progress.
Reston, VA: National Council of Teachers of Mathematics.
Carraher, D., Schliemann, A., & Brizuela, B. (2001). Can young students operate on unknowns?
In: M. van den Heuvel-Panhuizen (Ed.), Proceedings of the 25th conference of the international
group for the psychology of mathematics education, Utrecht University, Utrecht (Vol. 1,
pp. 130–140).
Chalouh, L., & Herscovics, N. (1988). Teaching algebraic expressions in a meaningful way. In: A.
F. Coxford (Eds), The Ideas of Algebra, K-12 (1988 Yearbook) (pp. 33–42). Reston, VA: National
Council of Teachers of Mathematics.
Chiarugi, I., Fracassina, G., & Furinghetti, F. (1990). Learning difficulties behind the notion of
absolute value. Proceedings of the 13th conference of the International Group for the Psychology
of Mathematics Education, PME, Mexico (Vol 3, pp. 231–238).
Christou, K. P., & Vosniadou, S. (2005). How students interpret literal symbols in algebra: A con-
ceptual change approach. In: B. G. Bara, L. Barsalou, & M. Bucciarelli (Eds), Proceedings of the
cognitive science 2005 conference Stressa, Italy (pp. 453–458).
Collis, K. F. (1975). The development of formal reasoning. Newcastle, Australia: University of
Newcastle.
Firth, D. E. (1975). A Study of Rule Dependence in Elementary Algebra. Unpublished master’s the-
sis. University of Nottingham, England.
Fischbein, E., Deri, M., Nello, M., & Marino, M. (1985). The role of implicit models in solving prob-
lems in multiplication and division. Journal of Research in Mathematics Education, 16, 3–17.
Gelman, R., & Gallistel, C. R. (1978). The child’s understanding of number. Cambridge, MA:
Harvard University Press.
Greer, B. (1994). Extending the meaning of multiplication and division. In: G. Harel, & J. Confrey
(Eds), The development of multiplicative reasoning in the learning of mathematics (pp. 61–85).
Albany, NY: SUNY Press.
Greer, B. (2005). Designing for conceptual change. In: J. Novotná, H. Moraová, M. Krátká, & N.
Stehlíková (Eds), Proceedings of the 30th conference of the International Group for the
Psychology of Mathematics Education, PME, Prague. (Vol. 1, pp. 171–172).
Hart, K. M. (1981). Children’s Understanding of Mathematics: 11–16. London: John Murray.
Heck, A. (2001). Variables in computer algebra, mathematics, and science. International Journal of
Computers in Mathematics Education, 8(3), 195–221.
Kieran, C. (1981). Concepts associated with the equal symbol. Educational Studies in Mathematics,
12, 317–326.
Kieran, C. (1992). The learning and teaching of school algebra. In: D. Grouws (Ed.), Handbook of
research on mathematics teaching and learning (pp. 390–419). New York: Macmillan.
Knuth, E. J., Alibali, M. W., McNeil, N. M., Weinberg, A., & Stephens, A. C. (2005). Middleschool
understanding of core algebraic concepts: Equivalence and variable. International Reviews on
Mathematical Education, 37(1), 68–76.
Kuchemann, D. (1978). Children’s understanding of numerical variables. Mathematics in School,
7(4), 23–26.
Kuchemann, D. (1981). Algebra. In: K. Hart (Ed.), Children’s understanding of mathematics: 11–16
(pp. 102–119). London: John Murray.
Matz, M. (1980). Building a metaphoric theory of mathematical thought. Journal of Mathematical
Behavior, 3(1), 93–166.
Resnick, L. B. (1987). Understanding algebra. In: J. A. Sloboda, & D. Rogers (Eds), Cognitive
process in mathematics (pp. 169–203). Oxford: Clarendon Press.
Resnick, L. B., Nesher, P., Leonard, F., Magone, M., Omanson, S., & Peled, I. (1989). Conceptual
bases of arithmetic errors: The case of decimal fractions. Journal for Research in Mathematics
Sackur, C. (1995). Blind Calculators in algebra: Write false interviews. In: E. Cohors-Fresenborh
(Ed.), Proceedings of the first European Research Conference on Mathematics Education
(ERCME ’95) (pp. 82–85). Osnabruck, Germany.
Sherin, B. L. (2001). How students understands physics equations. Cognition and Instruction, 19(4),
476–541.
Stacey, K., & MacGregor, M. (1997). Ideas about symbolism that students bring to algebra. The
Mathematics Teacher, 90(2), 110–113.
Stafylidou, S., & Vosniadou, S. (2004). Students’ understanding of the numerical value of fractions:
A conceptual change approach. Learning and Instruction, 14, 503–518.
Verschaffel, L., & Vosniadou, S. (Guest Eds). (2004). The conceptual change approach to mathe-
matics learning and teaching. Special Issue of Learning and Instruction, 14.
Vlassis, J. (2004). Making sense of the minus sign or becoming flexible in ‘negativity’. Learning and
W. Schnotz, S. Vosniadou, & M. Carretero (Eds), New Perspectives on Conceptual Change
(pp. 3–13). Amsterdam: Elsevier.
Vosniadou, S., Skopeliti, I., & Ikospentaki, K. (2004). Modes of knowing and ways of reasoning in
elementary astronomy. Cognitive Development, 19, 203–222.
Wagner, S. (1981). Conservation of equation and function under transformations of variables.
Journal for Research in Mathematics Education, 12, 107–118.
Chapter 20
Teaching for Conceptual Change: The Case of

Infinite Sets
Pessia Tsamir and Dina Tirosh
Introduction
The conceptual change approach has been widely used to interpret students’ solutions to
science problems in a series of developmental studies referring to science education (e.g.,
Carey, 1985; Posner, Strike, Hewson, & Gertzog, 1982; Vosniadou & Brewer, 1992; Wiser
& Amin, 2001). This approach was mainly used to explain knowledge acquisition in spe-
cific domains, while describing the importance of reorganization of existing knowledge
structures in some processes of learning.
Vosniadou, Ioannides, Dimitrakopoulou, and Papademetriou (2001) argued that intui-
tive knowledge of the physical world is necessary for functioning in the world, and that
learners tend to develop various beliefs and presuppositions related to scientific topics.
However, scientific explanations of the physical world often run counter to fundamental
principles of intuitive knowledge, which are confirmed by our everyday experience.
Consequently, in the process of learning, new information interferes with prior knowledge,
resulting in the construction of synthetic models (or misconceptions), and this shows that
knowledge acquisition is a gradual process during which existing knowledge structures are
revised slowly.
Similarly, when studying mathematics, in the course of accumulating mathematical
knowledge, the student goes through successive processes of generalization, while also
experiencing the extension of various mathematical systems. For instance, the concept of
number, a central concept in mathematics, is introduced very early on in primary school
via the system of natural numbers. Then, gradual transitions occur beginning with integers,
through rational numbers, irrational numbers and real numbers, and concluding with the
system of complex numbers presented in the upper grades of high school. The move from
one number system to a wider one preserves some numerical characteristics, adds some
others, while yet others are lost. For example, the transition from natural numbers to inte-
gers enables one to solve a problem like 5–7 (closure under subtraction). Yet, at the same

ISBN: 0-08-045355-4
300 Pessia Tsamir and Dina Tirosh
time it becomes impossible to generalize that subtraction “always makes smaller” and the
system no longer has a smallest number.
Awareness of the changes caused by the enlargement of mathematical systems and the abil-
ity to identify the variant and invariant elements under a specific transition are important
factors in the growth of mathematical knowledge. However, research findings clearly indicate
that learners tend to attribute all properties of a specific domain of numbers to a more general
one (e.g., regarding rational numbers: Greer, 1992; Hart, 1981; Klein & Tirosh, 1997; regard-
ing decimals: Moloney & Stacey, 1996; Putt, 1995; regarding negative numbers: Hefendehl,
1991; Streefland, 1996; regarding irrational numbers: Fischbein, Jehiam, & Cohen, 1995).
This general description suggests that the conceptual change framework, much like
other theories that relate to intuitive knowledge (e.g., Fischbein, 1987; Tall & Vinner,
1981), may be suitable for analyzing the development of students’ mathematical know-
ledge. Indeed, in recent years, several researchers have attempted to explore the promises
of this theoretical framework to mathematics learning and teaching. A special issue
devoted to the conceptual change approach to mathematics learning and teaching was
recently published in Learning and Instruction (Vosniadou & Verschaffel, 2004). Most
studies explored the conceptual changes involved in the transition from one number sys-
tem to a wider one (e.g., Merenluoto & Lehtinen, 2004; Stafylidou & Vosniadou, 2004;
Vlassis, 2004; Vamvakoussi & Vosniadou, 2004; see also, Hartnett & Gelman, 1998). In
this chapter we discuss the applicability of the conceptual change approach to the learning
and teaching of the Cantorian Set Theory. We focus on one major aspect of this theory: the
equivalency of infinite sets. The terms “comparing infinite sets,”1 “comparing infinite
quantities,” and “determining the equivalency of infinite sets” are used interchangeably to
account for the comparison of the cardinalities of these sets.
The Cantorian Set Theory is the most commonly used theory of infinity today. Yet stu-
dents face great difficulties in acquiring various properties of the equivalency of infinite
sets (Borasi, 1985; Duval, 1983; Fischbein, Tirosh, & Hess, 1979; Lakoff & Nunez, 2000;
Tall, 1980, 1992, 2001; Tirosh, 1991; Tsamir, 1999). It was reported that when asked to
compare the numbers of elements in two infinite sets students at different grade levels used
methods that are adequate only for the comparison of the number of elements in finite sets.
For example, students expect that the number of elements in a set, which is the union of
two distinct, non empty sets, is larger than that of the number of elements in each of these
sets. This, however, is true for finite sets, but not for infinite ones. It seems evident from
the related research findings and also from the historical development of the Cantorian Set
Theory that the acquisition of various aspects of the theory in general and the equivalency
of infinite sets, in particular, necessitates radical reconstruction.
In the course of the last twenty years we designed and evaluated several methods of
teaching the Cantorian Set Theory (Tirosh, 1991; Tirosh, Fischbein, & Dor, 1985; Tirosh
& Tsamir, 1996; Tsamir, 1999, 2003a; Tsamir & Tirosh, 1999). These instructional prac-
tices were inspired by Fischbein’s (1987) theory of intuition in mathematics and science.
The major principles that guided the development of these instructional practices (as
described, for instance, in Tirosh, 1991) were: identifying the intuitive criteria students use
1
The term “comparing infinite sets” is an abbreviation to “comparing the number of elements in two infinite sets”
Teaching for Conceptual Change 301
to compare infinite quantities; raising students’ awareness of inconsistencies in their own

thinking; discussing the origins of students’ intuitions about infinity; progressing
from finite to infinite sets; stressing that it is legitimate to wonder about infinity;
emphasizing the relativity of mathematics; and strengthening students’ confidence in the
new definitions.
We evaluated the impact of traditional courses with little or no emphasis on students’
intuitive tendencies to overgeneralize from finite to infinite sets, and of courses that were
developed in line with the principles listed above, on high-school students and on
prospective mathematics teachers’ intuitive and formal knowledge of Cantorian Set
Theory. Our findings in the different studies indicate that instruction that implemented
these principles led to the promotion of the reconstruction of knowledge structures
(Tirosh, 1991; Tsamir, 1999). These interventions promoted learners’ awareness of the
differences between finite and infinite systems, and of the contradictions that result from
interchangeably applying different criteria when comparing infinite sets. Looking at these
instructional interventions through different lenses could provide additional insights into
their pros and cons.
In this chapter we show that the instructional design principles deriving from the con-
ceptual change approach (as presented by Vosniadou et al., 2001) offer a valuable frame-
work for analyzing and reflecting on instructional interventions in mathematics. More
specifically, we focus on the mathematical notion of equivalency of infinite sets, using the
instructional design recommendations of the conceptual change approach to analyze and
reflect on related learning environments.
Using the Conceptual Change Approach to Reflect on Learning

Environments: The Case of Infinite Sets
In 2001, Vosniadou, Ioannides, Dimitrakopoulou and Papademetriou described eight rec-

ommendations for designing learning environments to promote conceptual change in sci-
ence. In the following section we show that these principles could be applied for reflecting
on learning environments that aimed at promoting learners’ grasp of the equivalency of
infinite sets. For this purpose we revisit examples from instructional interventions that we
developed, reflecting on them in light of Vosniadou et al.’s recommendations. Each of the
following subsections is dedicated to one of the recommendations.
Taking into Consideration Students’ Prior Knowledge
The realization that students do not come to school as empty vessels but
have representations, beliefs and presuppositions about the way the physi-
cal world operates that are difficult to change has important implications for
the design of science instruction. (Vosniadou et al., 2001, p. 392)
Vosniadou et al. emphasize the need, when designing instruction, to consider the ways
in which students see the physical world. They mention two sources where this informa-
tion might be found: the accumulated data on students’ conceptions and ways of thinking,
published in science education literature, and the examination of students’ ideas prior to
instruction.
This recommendation indicates the importance of a body of knowledge regarding stu-
dents’ conceptions, and their ways of thinking about concepts whose development could
necessitate radical reconstruction. In the case of equivalency of infinite sets such a body
of knowledge exists. Research in the field of mathematics education indicates that the
transition from the comparison of finite sets to the comparison of infinite sets (i.e., the
comparison of the number of elements in these sets) is problematic for many students
(e.g., Borasi, 1985; Duval, 1983; Falk, Gassner, Ben Zoor, & Ben Simon, 1986; Fischbein
et al., 1979; Martin & Wheeler, 1987; Sierpinska, 1987; Tall, 1980). When asked to com-
pare the number of elements in two infinite sets, students used different methods in their
comparisons, inevitably leading to contradictions. For instance, when asked to compare
the number of elements in the set of natural numbers and the set of positive, even
numbers, students used inclusion, 1:1 correspondence, and single infinity as their bases
for comparison. The use of inclusion (i.e., justifying the claim that two sets consist of dif-
ferent numbers of elements by stating that a proper subset has less elements than the set
itself) led them to conclude that the infinite sets have an unequal number of elements.
When using 1:1 correspondence (i.e., justifying the claim that two sets have the same
number of elements by pairing each element of one set with a unique element of the other
with no “leftovers”), or single infinity (i.e., all infinite sets have the same number of ele-
ments — infinity) they concluded that the number of elements was equal. Students often
accepted these two contradictory solutions to the same problem as both valid (e.g., Tirosh
& Tsamir, 1996).
In our studies, we found that consulting the research in mathematics education on learn-
ers’ conceptions of the equivalency of infinite sets is imperative for instruction related to
this complex concept. Our decision to start from reflecting on strategies that are often
implicitly used to compare the number of elements in finite sets, and only then to move to
infinite sets was based on the consistent observation, in various studies, that students tend
to apply methods applicable only for finite sets, to infinite ones. Some examples will be
presented later on.
Consulting the accumulated data on students’ conceptions and ways of thinking, as
described in the mathematics education literature, is one component of prior knowledge
that Vosniadou et al. (2001) recommended to consider when designing learning environ-
ment. The other component of prior knowledge that Vosniadou et al. (2001) discussed —
the examination of the ideas held by the students prior to instruction — was also found to
be important. For example, in a pretest conducted with 11th graders enrolled in the same
school, significant differences were found between the methods that students from two
classes used for comparing infinite sets. In one class, almost all students argued that “all
infinite sets have the same number of elements — infinity,” while in another class most
students tended to use inclusion considerations. Our search for an explanation of this unex-
pected diversity led us to examine their curriculum. It was found that at the time of the
pretest, the first class was learning about limits, addressing infinity as “a single sized
entity,” while the others were not. Such differences were taken into account when design-
ing the learning environments for each of these classes (Tsamir, 1990).
Facilitating Metaconceptual Awareness
To help students increase their metaconceptual awareness it is necessary to

create learning environments that make it possible for them to express their
representations and beliefs. This can be done in environments that facilitate
group discussion and the verbal expression of ideas. It is also important to
create learning environments that make it possible for students to express
their internal representations of phenomena, to compare them with those of
others. (Vosniadou et al., 2001, p. 392)
Vosniadou et al. provide an example of the design and implementation of a learning

environment to teach mechanics to 5th graders. The class was divided into small groups in
which the students were first asked to write individual answers in their notebooks, then to
discuss the written answers with their group peers, and finally, to agree on a “group
response” to be presented in a class discussion. During class discussion, students were
invited to challenge the answers of their own and other groups’ representatives. The
authors commented that there were often unresolved disagreements and an experiment was
conducted in the class, using everyday materials, to provide an objective answer to the
problem being investigated.
Similar instructional sequences were included in our intervention studies. Students
were often first asked to individually describe in writing the methods they used to deter-
mine whether two sets have the same number of elements. Then, they were asked to dis-
cuss their solutions in small groups in an attempt to convince their peers, before reporting
on their solutions to the entire class. During the class discussion, various opinions were
raised and discussed.
For example, in the following activity, we aimed at promoting prospective teachers’
awareness of the different criteria that they used for the comparison of finite sets. The
development of this activity was based on our findings, which students tend to argue, that
counting is the only strategy for comparing finite sets (the activity is also described in
Tsamir, 1999).
Comparing finite sets. The activity consisted of three stages (Appendix A). Stage 1
aimed at exposing the prospective teachers’ tendency to claim that counting the number of
elements in each finite set and comparing the two resulting numbers is the only method for
comparing sets. For this purpose two disjoint, finite sets were introduced, both with small
numbers of elements. Stage 2 aimed at raising the prospective teachers’ awareness of vari-
ous methods that they actually use for the comparison of the number of elements in two
finite sets. Here, five problems were presented, each intended to elicit a different method for
the comparison of finite sets: Problem 1 — 1:1 correspondence; Problem 2 — counting;
Problem 3 — inclusion; Problem 4 — intervals (comparing distances between “consecutive
elements” in the two sets) and 1:1 correspondence; and Problem 5 — intervals. Stage 3 had
several aims, including: (a) to practice reflection upon the methods used to compare the
number of elements in the sets; (b) to conclude that comparisons are possible even when the
elements cannot be counted; (c) to realize that counting, even when applicable, is not always
the best method for comparing; and (d) to realize that counting is actually creating 1:1 cor-
respondence between a set and a specific subset of the natural numbers.
In our study, we addressed mathematics prospective teachers for secondary school, and
indeed, all participants individually solved the comparison in Stage 1 by counting. When
asked to reach a generalized conclusion, during small-group discussions and later on in
class discussion, they responded that counting is the only way to compare finite sets.
Like in Stage 1, the problems in Stage 2 were first solved individually. These problems
could be solved by various methods. The participants did in fact use all four available
methods to solve the problems. They listed these methods in response to the first assign-
ment in the reflection-stage (Stage 3) (counting, mainly in Problem 2; 1:1 correspondence,
mainly in Problem 1; inclusion, mainly in Problem 3; and intervals, mainly in Problems 4
and 5). When all the methods used were presented in a single table, two findings emerged:
(a) for each comparison, each prospective teacher used a single method; (b) 1:1 corre-
spondence was used by several prospective teachers to solve some problems.
The class discussion emphasized that: (1) the number of elements in two finite sets could
be compared even when counting was impossible (see Problem 1 and Problem 5). (2) Even
where counting was applicable, it was not necessarily the preferable method for attaining
the solution (see Problems 3 and 4). (3) Counting is basically pairing, that is, creating 1:1
correspondence between the elements of a given set and a subset of the natural numbers.
The combination of individual work, of small group work, and of class discussions was
crucial in raising students’ awareness of the various explicit as well as implicit methods
they themselves used to compare finite sets.
Addressing Students’ Entrenched Presuppositions
It is important in instruction to distinguish new information that is consis-

tent with prior knowledge from new information that runs contrary to prior
knowledge. (Vosniadou et al., 2001, p. 393)
Vosniadou et al. further emphasize that when new information is consistent with prior
knowledge, it is most likely to be understood even if it is presented as a fact without any
further explication. However, when the new information runs contrary to the existing con-
ceptual structures, simply presenting the new information as a fact may not be adequate.
Vosniadou et al. noted that in science instruction counterintuitive information is often
introduced as a fact.
In the case of the equivalency of infinite sets, some of the criteria for comparing the
number of elements in two sets could be used only for finite sets (i.e., counting), other
methods could, in principle, be used for both finite and infinite sets (e.g., 1:1 correspon-
dence and inclusion). However, some of the latter methods are considered inadequate for
comparing the number of elements in two infinite sets, in the framework of Cantorian set
theory (e.g., inclusion). In fact, here only one criterion is applicable to both finite and infi-
nite sets, namely, 1:1 correspondence. Actually, Cantor used this criterion for defining the
equivalency of two sets, either finite or infinite.
We have noted before that students regarded counting as the criterion for comparing
sets, and that they had only finite sets in mind when expressing this view. However, when
extending the system, relating both to finite and infinite sets, obviously, this criterion can-
not be used for the new types of sets. In our instruction we led the students to acknowl-
edge the limitations of the “counting” criterion. The participants themselves noted that this
criterion could not be used for comparing the number of elements in the new type of sets.
This raised their awareness of the need to choose another, more suitable criterion. Thus,
they became aware of the need to re-examine each of the criteria that had already been dis-
cussed with reference to finite sets.
In addressing students’ entrenched presuppositions, we stressed that it is both natural
and legitimate to wonder about the nature and characteristics of infinite sets. For this
purpose we extensively related to statements of mathematicians who acknowledged
the difficulties involved in this topic. For instance, we presented the students with the
following two excerpts: the first showing that even Galileo acknowledged the puzzling
aspect of infinity, and the second, by Hann, discussing the difficulties related to
the equivalency between a set (natural numbers) and its proper subset (positive, even
numbers).
Here a difficulty presents itself which appears to me insoluble. Since it is

clear that we may have one line segment longer than another, each con-
taining an infinite number of points, we are forced to admit that, within one
and the same class, we may have something greater than infinity, because
the infinity of points in the long line segment is greater than the infinity of
points in the short line segment. This assigning to an infinite quantity a
value greater than infinity is quite beyond my comprehension. (Galileo,
1954/1638, p. 31)
An infinite set is equivalent to at least one of its proper subsets. If we look

for examples of enumerable infinite sets we arrive immediately at highly
surprising results. The set of all positive even numbers is an enumerable
infinite set and has the same cardinal number as the set of all the natural
numbers, though we would be inclined to think that there are fewer even
numbers than natural numbers. (Hann, 1956, p. 1604)
Motivation for Conceptual Change
Students often do not see the reason to change their beliefs and presuppo-
sition because they provide good explanations of their everyday experi-
ences, function adequately in the everyday world, and are tied to years of
confirmation. (Vosniadou et al., 2001, p. 393)
Vosniadou et al. (2001) relate to two major aspects in the task of persuading students to
invest the substantial effort required to re-examine their initial explanations, and to become
literate. The first aspect is creating an environment that will motivate the desired changes
and the second is to ensure that experiences embedded in the environment are selected
carefully so that they are theoretically relevant. Vosniadou et al. further explain that
“theoretically relevant” means addressing
the underlying presuppositions and beliefs that constrain students’ repre-

sentations and influence the way they interpret scientific information and
show them that these presuppositions are not adequate to explain the known
empirical facts. (Vosniadou et al., 2001, p. 393)
We designed various activities aimed at increasing students’ motivation and interest in

the comparison of sets. In these activities students experienced surprise when realizing that
they were unaware of several adequate and inadequate criteria that they had been using in
various situations. Here, we describe one such activity.
Inclusions? 1:1 Correspondence? Intervals? The activity unfolded in three stages

(Appendix B). Stage 1 aimed at (a) increasing prospective teachers’ awareness of the
validity of all four methods (counting, 1:1 correspondence, inclusion, and intervals) for
comparing finite sets and (b) increasing prospective teachers’ awareness of the extent to
which each of the methods is applicable in a given problem. Stage 2 aimed at (a) dis-
cussing the role of consistency in mathematical systems and (b) demonstrating that 1:1
correspondence is the only method that is always applicable. At Stage 3 the students were
encouraged to draw the final conclusions for finite sets, namely that all four criteria can be
used to compare the number of elements in two finite sets but that 1:1 correspondences is
the most general criterion.
Here, much like in the activity previously described under Stages 1 and 2, the participants
individually solved the problems. It was interesting for them to discover that the comparison
of the number of elements in two finite sets can be conducted by using counting, inclusion,
intervals, and 1:1 correspondence without violating the consistency of the theory.
The students were quite surprised to realize that 1:1 correspondence is a method that
enabled them to compare the number of elements in various sets, in all these problems.
This, in itself, created some motivation to examine the usefulness of the various methods
that they intuitively used to compare the number of elements in infinite sets.
Cognitive Conflict
Conceptual change is a difficult process that involves the reorganization of

not just one misconception or one belief, but an interrelated system of
beliefs and presupposition that takes a long time to be accomplished. Thus,
conceptual change cannot but be a gradual affair, which requires the use of
many different instructional interventions of which cognitive conflict is
only one. (Vosniadou et al., 2001, p. 394)
Vosniadou et al. (2001) claimed that cognitive conflict should be used in science
instruction with caution. They showed examples of ways of using cognitive conflict, where
the conflict was triggered by making children realize that their explanatory framework
could not explain some empirical results. In the case of comparing infinite sets, we exten-
sively used the cognitive conflict approach to increase students’ realization that “some-
thing is wrong” and in need of reconstruction. We also used other approaches, such as
analogy (Tsamir, 2003).
We shall here present two related activities; one was used with 10th graders and the
other with prospective teachers.
Even numbers and multiples of four: Inclusion vs. single infinity. The activity was
used with 10th graders. It aimed at encouraging participants to discuss the consequences
of relying only on intuitive criteria when comparing the number of elements in two infi-
nite sets. In this activity the class was divided into teams of four; each student was first
asked to provide his/her own intuitive answer to the question:
Two sets are given: N = {2, 4, 6, 8, 10, …}, M = {4, 8, 12, 16, 20, ….}
Is the number of elements in set M equal to the number of elements in set N?
Explain your answer.
Each team was then instructed to discuss its answers and to reach mutual agreement
about the correct response.
Notably, when applying the question “Are these two equal” to the context of infinite
sets it has no formal meaning for the students, because the term “equal” was not yet
defined in the context of infinite sets. However, all the students provided intuitive answers
to this question, based on the meaning that they attributed to the term “equal”.
We found that in each team, some students claimed that set M has less elements,
because it is included in set N, while other students in the same team claimed that both sets
are infinite and therefore they both have an infinite number of elements. The participating
students noted that both answers seemed reasonable and that they were unable to decide
which of them was correct. They further expressed the feeling that something was wrong.
Natural numbers and multiples of 100: Inclusion, 1:1 correspondence, and single
infinity. This activity was used with prospective teachers. It aimed at promoting their
awareness that the application of different methods when comparing infinite sets leads to
contradictory answers. Participants were first asked to individually solve several compar-
isons of infinite sets tasks. For instance,
Given B = {1, 2, 3, 4, 5, 6, 7, …}, P = {100, 200, 300, 400, 500, 600, 700, …}
The number of elements in sets P and S is equal/not equal.
Explain …
We chose the tasks so that they would trigger the use of three different criteria (inclu-
sion, 1:1 correspondence, and single infinity). Indeed, most students used one method for
each problem; however, they commonly used more than one method in the comparisons
conducted in the different problems.
The prospective teachers were then asked, once more, to individually solve the same
problems. This time, however, the instruction was “Try to apply all the three methods to
each problem.” Here is an example:
Given B = {1, 2, 3, 4, 5, 6, 7, …}, P = {100, 200, 300, 400, 500, 600, 700, …}
Is ‘1:1’ correspondence applicable? Yes/No
If your answer is Yes — Use this method to solve the problem
Is the number of elements in set B equal to the number of elements in set P?
Yes/No
Is ‘inclusion’ applicable? Yes/No

Yes/No
Is ‘all infinities are equal’ applicable? Yes/No
Yes/No
The participants, while performing this activity, realized that the application of the
three methods led to contradictory solutions. This created a state of confusion and raised
questions (For example: Why did it work with finite sets? What should be done in the case
of infinite sets?). Participants were then asked to reflect in groups on their responses. The
specific guiding questions were:
1. Is it OK to alternatively use these methods for the comparison of infinite sets?
Why?
2. In your opinion, which (if any) of the various methods for comparing infinite sets is
preferable? Why?
In the class discussion, after each group presented its approach, the participants reached
the conclusion that at least two of the above-mentioned methods are valid for the compar-
ison of infinite sets, but only if used exclusively to compare all pairs of infinite sets.
Otherwise, essential consistency is violated. Consequently, the choice of a method for the
comparison of all infinite sets should be made in advance, and this method must then be
used exclusively.
A question that naturally arose was: Is there a preferable method? that is, one that pro-
vides more conclusive answers. Participants found that: (1) all infinities are equal, that is,
all sets should have the same number of elements, and therefore there is no reason for com-
parison; (2) inclusion was only occasionally applicable; (3) 1:1 correspondence was found
to enable the comparison of infinite sets both when such a relationship existed (indicating
“equality”) and when it could be proved that there is no such correspondence (indicating
“inequality”). Moreover, this method was also applicable for comparing finite sets. Thus,
1:1 correspondence appeared to be the most applicable method both for finite sets and for
infinite sets.
This assignment, before formally studying the theorems of the Cantorian Set Theory,
pointed to the possibility of using various methods when comparing given sets as a prop-
erty lost in the transition from finite to infinite sets; another property that was lost in this
transition is: “The whole is always greater than its part.” The possible equivalency of a set
and its proper subset is a characteristic gained in this transition.
Providing Models and External Representations
Models and external representations can be used to clarify aspects of a sci-

entific explanation that are not apparent when the explanation is given in a
linguistic or mathematical ways. The visual qualities of a model are useful
in making an explanation better understood and more easily memorized.
(Vosniadou et al., 2001, p. 394)
Vosniadou et al. showed how in the learning environments that they created to teach
mechanics to 5th grade students they used measurements of forces, symbolic represen-
tations of force and energy, and a friction model to clarify aspects of the scientific expla-
nations that were not apparent when the explanation was given in a linguistic or
mathematical way.
In the case of comparison of infinite sets, we presented 16–18-year-old students and
prospective teachers with different representations of problems that required the compari-
son of the number of elements in pairs of infinite sets. For example, we used four repre-
sentations of the task asking to compare the number of natural numbers with that of the
natural numbers larger than 2, namely, {1, 2, 3, 4, 5, 6, 7, 8, 9, 10, …} and {3, 4, 5, 6, 7,
8, 9, 10, …}.
Horizontal representation — the sets are presented beside each other.

A = {1, 2, 3, 4, 5, …}, B = {3, 4, 5, 6, 7, …}
Vertical representation — the sets are presented one below the other.
A = {1, 2, 3, 4, 5, …}
B = {3, 4, 5, 6, 7, …}
Explicit representation — the sets are presented one below the other so that it is visually
easy to match each element of one set with one element of the other set. Here the pairs are
1↔1+2, 2↔2+2, 3↔3+2, 4↔4+2, 5↔5+2 etc.
A = {1 , 2 , 3 , 4 , 5 , …}
B = {1+2, 2+2, 3+2, 4+2, 5+2, …}
Geometrical representation — the elements of the given sets are related to geometrical fig-
ures that trigger the matching of each element of one set with one element of the other set
(Figure 20.1). For example, participants were asked to consider the two sets of numbers,
and the set of trapezoids:
The upper base of each trapezoid is shorter by 2 cm than the bottom base. The first
upper base is 1 cm long and each following segment is 1 cm longer than the previous one.
The first bottom base is 3 cm long and each following bottom base segment is 1 cm longer
than the previous one. Set A presents the numbers that express the lengths (in cm) of the
upper bases; and set B presents the numbers that express the lengths (in cm) of the bottom
bases in the trapezoids.
A = {1, 2, 3, 4, 5, 6, 7, 8, 9, 10, …}
B = {3, 4, 5, 6, 7, 8, 9, 10, 11, 12, …}
1cm 2cm 3cm 4cm
{ 3cm
,
4cm
,
5cm
,
6cm
, ... }
Figure 20.1: Vertical and Graphical Representations.
For each representation, students were asked to judge whether the numbers of elements
in Sets A and B were equal, and to explain their answers.
It was found that the horizontal representation encouraged more “natural numbers”
responses, usually justified by inclusion arguments, to the effect, for instance, that “set A
consists of all the elements that are included in set B in addition to two extra elements.
Thus, there are two more elements in set A than in set B.” The other representations evoked
“the same number of elements” responses, accompanied by various types of justifications.
The vertical representation triggered “single infinity” justifications, claiming that “all infi-
nite sets have the same number of elements, so the presented sets are also equal.” The
explicit and geometrical representations triggered one-to-one correspondence considera-
tions, pointing to a way to pair matching elements. Hence, students reached contradictory
responses to the different representations of the same task; for instance, “equal number of
elements” to the geometrical and explicit representations, and “unequal number of ele-
ments” to the horizontal representation.
We would like to emphasize that two representations, the explicit and the geometrical,
led to responses consistent with Cantorian Set Theory, in terms of both the judgment and
the method used. These representations were used to design anchoring tasks in instruc-
tional interventions (e.g., Tsamir, 2003b; Tsamir & Tirosh, 1999).
Breadth of Coverage of the Curriculum
It may be more profitable to design curricula that focus on the deep explo-
ration and understanding of a few, key concepts in one subject matter area
than curricula that cover a great deal of material in a superficial way.
(Vosniadou et al., 2001, p. 391)
Vosniadou et al. suggest focusing on a limited number of key concepts and giving
students enough time and adequate learning environments to achieve a qualitative under-
standing of these concepts. Clearly, the concepts: “equal,” “unequal,” and “equivalent”
are key concepts in mathematics. They are widely used in various mathematical domains
including arithmetic, algebra, trigonometry, analysis, and geometry. In fact, it is hard to
imagine a mathematical domain where these notions play no significant part. Hence,
devoting time and attention to designing learning environments that address these
notions in various ways, and evaluating them, should be a major aim of mathematics
learning and teaching.
Order of Acquisition of the Concept Involved
The relational structure in which the concepts of a domain are acquired

needs to be taken into consideration. (Vosniadou et al., 2001, p. 391)
Vosniadou et al. explained that in the design and implementation of learning environ-
ments a great deal of attention should be paid to the sequence in which the concepts were
introduced. This should be done to avoid the formation of new misconceptions and to over-
come existing ones. In designing instructional interventions Vosniadou et al. (2001)
attempted to consider possible synthetic models that could be constructed by children on

the basis of their prior knowledge, and to use this analysis as basic criteria for the design
of the order of the concepts to be taught.
When considering the sequencing of instruction in the case of the equivalency of infi-
nite sets, clearly, one needs to think about the order of the presentation of the relevant
concepts (e.g., set, finite sets, infinite sets, equal, equivalent, inclusion, 1:1 correspon-
dence, cardinality). However, there are other mathematical concepts that could influence
a student’s reactions to the comparison of finite and infinite sets tasks, with which the
student might (or might not) be familiar (e.g., function, series, sequences, limits).
Decisions regarding the sequence of instruction should take account of the concepts that
are embedded in the specific domain and of the concepts embedded in additional
domains.
The sequences that we developed took account of both these types of concepts. For
example, the instructional unit for 10th graders was developed for students who were not
familiar with the concept of limit. The instructional sequences designed for prospective
teachers were sensitive to possible impacts of the participants’ familiarity with limits. The
two sequences were similar with regard to the order of presentation of the set-theory
embedded notions: from finite sets to infinite sets with attention to comparison methods
that could be applied to both domains.
Summing Up and Looking Ahead
In this chapter we have shown that the conceptual change-based recommendations for
designing science learning environments could be valuable for mathematics education as
well. The case that we discussed, the equivalency of infinite sets, could be regarded as a
touchstone for examining the applicability of these recommendations. The related diffi-
culties are well documented from both the phylogenetic and the ontogenetic viewpoints.
In our case, Vosniadou et al.’s (2001) recommendations were used for reflection on instruc-
tional interventions that were formulated without reference to these principles. We have
shown a way in which the given recommendations can serve as a tool for analyzing the
environments that we developed. However, it seems that these recommendations could and
should be used for designing learning environments for mathematics instruction that are
apt to involve conceptual change. The eight recommendations, their discussion, illustra-
tions, and labeling could assist in designing and formulating instruction in mathematics
education.
However, the impact of several inherent differences between mathematics and science
education should be carefully considered when using Vosniadou et al.’s (2001) recommen-
dations for designing learning environments for science instruction in mathematics instruc-
tion. Issues to be considered include the role of empirical facts in each of these domains,
the consequences of confronting one specific misconception in each domain, and the rela-
tive nature of mathematics. Studies should be conducted to examine the impact of mathe-
matics learning environments that are designed according to the recommendations of
Vosniadou et al. (2001), regarding students and teachers’ mathematical conceptions and
ways of thinking.
References
Borasi, R. (1985). Errors in the enumeration of infinite sets. Focus on Learning Problems in
Mathematics, 7, 77–88.
Duval, R. (1983). L’obstacle du dedoublement des objects mathematiques. Educational Studies in
Falk, R., Gassner, D., Ben Zoor, F., & Ben Simon, K. (1986). How do children cope with the infin-
ity of numbers? In: The Proceedings of the 10th conference of the international group for the psy-
chology of mathematics education (pp. 7–12). London, UK: University of London Institute of
Education.
Fischbein, E. (1987). Intuition in science and mathematics: An educational approach. Dodrecht, The
Netherlands: Reidel.
Fischbein, E., Jehiam, R., & Cohen, D. (1995). The concept of irrational numbers in high-school stu-
dents and prospective teachers. Educational Studies in Mathematics, 29, 29–44.
Fischbein, E., Tirosh, D., & Hess, P. (1979). The intuition of infinity. Educational Studies in
Galileo, G. (1954/1638). Dialogues concerning the new sciences. New York: Dover.
Greer, B. (1992). Multiplication and division as models of situations. In: D. Grouws (Ed.), Handbook
of research on mathematics teaching and learning (pp. 276–295). New York: Macmillan.
Hann, H. (1956). Infinity. In: J. R. Newman (Ed.), The world of mathematics (Vol. 3,
pp. 1593–1611). New York: Simon & Schuster.
Hart, K. (Ed.). (1981). Children’s understanding of mathematics. London: Murray.
Hartnett, P., & Gelman, R. (1998). Early understanding of numbers: Paths or barriers to the con-
struction of new understand? Learning and Instruction, 8, 341–374.
Hefendehl, H. L. (1991). Negative numbers: Obstacles in their evolution from intuitive to intellec-
tual constructs. For the Learning of Mathematics, 11, 26–32.
Klein, R., & Tirosh, D. (1997). Teachers’ pedagogical content knowledge of multiplication and divi-
sion of rational numbers. In: E. Pehkonen (Ed.), Proceedings of the 21st conference of the inter-
national group for the psychology of mathematics education (Vol. 3, pp. 144–151). Lahti, Finland.
Lakoff, G., & Nunez, R. (2000). Where mathematics comes from. NY: Basic Books.
Martin, W. G., & Wheeler, M. M. (1987). Infinity concepts among preservice elementary school teach-
ers. In: J.G. Bergeron, N. Herscovics, & C. Kieran (Eds), Proceedings of the 11th conference of the
international group for the psychology of mathematics education (pp. 362–368). Paris, France.
Moloney, K., & Stacey, K. (1996). Understanding decimals. Australian Mathematics Teacher, 52,
4–8.
Posner, G. J., Strike, K. A., Hewson, P. W., & Gertzog, W. A. (1982). Accomodation of a scientific
conception: Toward a theory of conceptual change. Science Education, 66, 211–227.
Putt, I. J. (1995). Preservice teachers ordering of decimal numbers: When more is smaller and less
is larger! Focus on Learning Problems in Mathematics, 17, 1–15.
Sierpinska, A. (1987). Humanities students and epistemological obstacles related to limits.
Educational Studies in Mathematics, 18, 371–387.
cal value of fractions. Learning and Instruction, 14, 503–518.
Streefland, L. (1996). Negative numbers: Reflection of a learning researcher. Journal of
Mathematical Behavior, 15, 57–77.
Tall, D. (1980). The notion of infinite measuring numbers and its relevance in the intuition of infin-
ity. Educational Studies in Mathematics, 11, 271–284.
Tall, D. (1992). The transition to advanced mathematical thinking: Functions, limits, infinity and
proofs. In: D. Grouws (Ed.), Handbook of research on mathematics teaching and learning
(pp. 495–511). New York: Macmillan.
Tall, D. (2001). Natural and formal infinities. Educational Studies in Mathematics, 48, 199–238.
Tall, D., & Vinner, S. (1981). Concept image and concept definition in mathematics, with special
reference to limits and continuity. Educational Studies in Mathematics, 12, 151–169.
Tirosh, D. (1991). The role of students’ intuitions of infinity in teaching the Cantorian theory. In: D.
Tall (Ed.), Advanced mathematical thinking (pp. 199–214). Dordrecht, The Netherlands: Kluwer.
Tirosh, D., & Tsamir, P. (1996). The role of representations in students’ intuitive thinking about infin-
ity. International Journal of Mathematics Education in Science and Technology, 27, 33–40.
Tirosh, D., Fischbein, E., & Dor, E. (1985). The teaching of infinity. In: L. Streefland (Ed.),
Proceedings of the 9th conference of the international group for the psychology of mathematics
education (pp. 501–506). Utrecht, The Netherlands.
Tsamir, P. (1990).
[Students’ inconsistent ideas about actual infinity]. Unpublished thesis for the Master’s degree. Tel
Aviv University, Tel Aviv. [In Hebrew].
Tsamir, P. (1999). The transition from comparison of finite to the comparison of infinite sets:
Teaching prospective teachers. Educational Studies in Mathematics, 38, 209–234.
Tsamir, P. (2003a). From “easy” to “difficult” or vice versa: The case of infinite sets. Focus on
Learning Problems in Mathematics, 25, 1–16.
Tsamir, P. (2003b). Primary intuitions and instruction: The case of actual infinity. Research in
Collegiate Mathematics Education, 12, 79–96.
Tsamir, P., & Tirosh, D. (1999). Consistency and representations: The case of actual infinity. Journal
for Research in Mathematics Education, 30, 213–219.
Vlassis, (2004) Making sense of the minus sign or becoming flexible in negativity. Learning and
Vosniadou, S., & Verschaffel, L. (Guest Eds.). (2004). The conceptual change approach to mathe-
matics learning and teaching. Learning and Instruction, 14 (Special Issue).
Vosniadou, S., Ioannides, C., Dimitrakopoulou, A., & Papademetriou, E. (2001). Designing learning
environments to promote conceptual change in science. Learning and Instruction, 11, 381–419.
Wiser, M., & Amin, T. (2001). “Is heat hot?” inducing conceptual change by integrating everyday
Appendix: A
Stage 1: Counting
Here are two sets A and B: A={1, t, !}, B={7, w}

Is the number of elements in sets A and B equal? Yes/No How did you reach this con-
clusion? _________________
Complete the following generalization: In order to compare the number of elements in
any two finite sets one could ________________
Stage 2: Various Methods
Problem 1
At a dance party all the students danced in couples, a boy and a girl in each couple.
No pupils were left without a partner. Z = {The boys} W = {The girls}
Is the number of elements in set Z equal to the number of elements in set W?
Yes/No
How did you reach this conclusion? _________
Problem 2
Given the sets: X = {1,2,3,4,5,6,7,8,9,10,11,12}, Y = {a,b,c,d,e}
Is the number of elements in set X equal to the number of elements in set Y?
Yes/No
Problem 3
Given the sets: Y = {a,b,c,d,e,f}, V = {a,b,c,}
Is the number of elements in set Y equal to the number of elements in set V?
Yes/No
Problem 4
Dan was ill. The doctor prescribed one green tablet every 3 hours for the first week.
Then, in the second week he was ordered to take a red capsule every 3 hours.
G = {The green tablets}, R = {The red capsules}
Is the number of elements in set G equal to the number of elements in set R?
Yes/No
Problem 5
Along the new promenade, a lane of trees was planted; a tree every 200 m. Every 400 m a
street light was placed adjacent to a tree. The first and the last trees had a street light next
to them. L = {The street lights}, T = {The trees}
Is the number of elements in set L equal to the number of elements in set s?
Yes/ No
Stage 3: Reflection
Let’s reflect —
1. Which methods did you use in order to compare the above sets?
2. Which method did you use in each problem?
3. Could the number of elements be compared even when counting was impossible?
4. In cases that counting was applicable, was it always the preferable method?
5. What is the meaning of ‘counting’?
Appendix: B
Stage 1: The Four Criteria: The Finite Case
Participants were asked to try to use all four methods to compare the finite sets in each of
the problems that were given in Stage 2 of the activity, which is described in Section 2.2.
One example follows:
Sample Problem (Problem 1)
At a dance party all the students danced in couples, a boy and a girl
in each couple. No pupils were left without a partner.
Z = {The boys} W = {The girls}
Is the number of elements in set Z equal to the number in set W? Yes / No
Try to apply all the four methods to each problem.

In order to answer this question —
Is ‘1:1’ correspondence applicable? Yes/No . If your answer is Yes — Use this method to
solve the problem-
Is the number of elements in set Z equal to number in set W? Yes/No
Is ‘inclusion’ applicable? Yes/No. If your answer is Yes — Use this method to solve the
problem-
Is ‘intervals’ applicable? Yes/No. If your answer is Yes— Use this method to solve the
problem-
Is ‘counting’ applicable? Yes/No. If your answer is Yes — Use this method to solve the
problem-
Stage 2: The Four Criteria in the Finite Case
At Stage 2 the participants were asked to reflect on their actions during Stage 1, by filling
in a summary table, referring both to the problems and to the criteria.
Let’s reflect —
1. List the methods applicable to each problem.
Problem No. 1 2 3 4 5
Boys — 1 to 12 — a to f — tablets — lights —
Girls a to e a to e capsules trees
a. Counting
b. Inclusion
c. Intervals
d. Pairing — 1:1 correspondence
2. Is it OK to alternatively use these methods for the comparison of finite sets? Why?
3. Is there, after all, a preferable method?
4. How do the methods work?
5. Does the “counting” method enable to compare the numbers of elements in two given
sets when there is no way to count?
6. Does the “inclusion” method enable to compare the numbers of elements in two given
sets when there is no inclusion relationship between the sets?
7. Does the “intervals” method enable to compare the numbers of elements in two given
sets when there is no way to predict the intervals?
8. Does the “1:1 correspondence” method enable to compare the numbers of elements in
two given sets when there is no way to pair?
Stage 3: Summary — Finite Sets
This was the final activity relating to the comparison of finite sets. Here students were
expected to express awareness of the various methods that can be used for the comparison
of finite sets, and of the advantage of 1:1 correspondence.
Final conclusion for finite sets:
Complete the following:
In order to compare the number of elements in any two finite sets you can
COMMENTARIES
Chapter 21
Nurturing Conceptual Change in

Mathematics Education
Brian Greer and Lieven Verschaffel
Introduction
Conceptual change lies at the heart of mathematics as a human creation. What is now con-
sidered the body of mathematical knowledge that a student may be expected to understand
after schooling is the achievement of the collective intellectual activity of mankind over
millennia. Consequently, as Sinclair (1990) pointed out, the challenge of mathematics edu-
cation is that we expect children to master in a few years complexities that historically took
millennia to evolve through conceptual change.
Galileo (as noted by Tsamir & Tirosh, this volume), despite being a genius untram-
meled by conventional thinking, could not comprehend how the number of points in a
shorter line could be equal to the number in a longer line (though he did have the consid-
erable insight that there was something puzzling that he did not understand). Yet a student
of today is expected to understand the resolution of this apparent paradox. In order to
achieve this miracle through education, it is essential to know how to nurture conceptual
change in today’s students.
Conceptual Change Theory Applied to Mathematics
Conceptual change theory (CCT), developed by Vosniadou and her colleagues and
presented in this volume, originated within the domain of science education. Authors of
chapters in Part 3 argue that its principles are equally applicable to mathematics education,
despite certain differences such as
the role of empirical facts in each of these domains, the consequences of

confronting one specific misconception in each domain, and the relative
nature of mathematics. (Tsamir & Tirosh, this volume)

ISBN: 0-08-045355-4
320 Brian Greer and Lieven Verschaffel
In particular, philosophers and historians of science and mathematics have tended to

emphasize a distinction between the characters of the conceptual changes that occur in
these two fields (see, especially, Gillies, 1992b). From examples such as the replacement
of the Ptolemaic system by the Copernican, it is argued that the changes whereby old
theories are replaced by new ones in science are revolutionary in the sense that the old one
is displaced. By contrast, when conceptual change occurs in mathematics, the old structure
is retained as a substructure of the new, the obvious example being that of number
(e.g., Lakoff & Nunez, 2000).
However, the answer to the question “Are there revolutions in mathematics?” (Gillies,
1992a) depends on the definition of “revolution”. Dunmore (1992) argued that there have
been changes in mathematics at the meta-level whereby earlier views have indeed been
displaced. For example, in the long struggle to establish the ontological status of negative
integers, the doctrine that they are “impossible” was rejected. Likewise, the belief in the
uniqueness of Euclidean geometry as the description of space has been irreversibly dis-
carded. From this perspective, there are conceptual changes in mathematics that truly merit
the title “revolutionary”. Dauben (1992) argued along the same lines, citing as four exam-
ples (a) the discovery of incommensurable magnitudes/irrational numbers, (b) Cantor’s
transfinite set theory, (c) new standards of rigour for calculus introduced by Cauchy,
(d) Robinson’s non-standard analysis, created in the 20th century. He concluded that
when a true revolution has taken place, a significant part of the ‘older’
mathematics will come to be replaced or dramatically augmented by con-
cepts and techniques that visibly change the vocabulary and grammar of
mathematics. (Dauben, 1992, p. 80)
There are already theories of conceptual change in mathematics education, of course —

every theory of cognitive development is a theory of conceptual change, in some
sense — and at some point a systematic review of how they relate to CCT would be
valuable. Piagetian theory, in particular, has several core constructs that are highly
compatible with CCT, notably those of equilibration and the complementary processes
of assimilation/accommodation. However, Piaget was essentially an epistemologist and
the weakness of his theory, from our perspective, is that he did not take sufficient account
of the fact that cognitive development in mathematics takes place in historical/cultural
circumstances, and, in particular, almost entirely under instruction.
One of Piaget’s most trenchant critics, Freudenthal, was originally a mathematician
who turned his attention to mathematics education, and was the leading founder of
arguably the single most important theoretical/applied centre within mathematics educa-
tion, whose work continues at the Freudenthal Institute. Freudenthal analysed in consider-
able detail the historical development of mathematics which, as he said, “grows by its
self-organizing momentum” (1990, p. 15) through “the interplay between form and con-
tent” (1990, p. 12) from roots in practical experience and problem solving. One major
thread of his philosophy of mathematics education, that is highly relevant to CCT, is the
need for children to retrace the steps taken historically in the construction of mathematics:
I asked the question of whether the learner should repeat the learning
process of mankind. Of course not. Throughout the ages history has, as it
Nurturing Conceptual Change in Mathematics Education 321
were, corrected itself, by avoiding blind alleys, by cutting short numerous

circuitous paths, by rearranging the road-system itself. We know nearly
nothing about how thinking develops in individuals, but we can learn a
great deal from the development of mankind. Children should repeat the
learning process of mankind, not as it factually took place but rather as it
would have done if people in the past had known a bit more of what we
know now. (Freudenthal, 1990, p. 48)
A second major theoretical perspective from within mathematics education is that of

Fischbein, originally a psychologist, and his followers (including Tsamir and Tirosh). In
particular, Fischbein (1987) analysed the difficulties in overcoming intuitions in the face
of situations requiring conceptual change, and abundantly illustrated them with historical
examples. He pointed out that:
From the educational point of view there is an important problem to be con-

sidered by curricula writers and by teachers. A certain interpretation of a
concept or an operation may be initially very useful in the teaching process
as a result of its intuitive qualities (concreteness, behavioral meaning etc.).
But as a result of the primacy effect that first model may become so rigidly
attached to the respective concept that it may become impossible to get rid
of it later on. The initial model may become an obstacle which can hinder
the passage to a higher-order interpretation — more general and more
abstract — of the same concept. (Fischbein, 1987, p. 198)
This analysis clearly bears on several examples in the chapters of Part 3.

While CCT shares characteristics with these other theories, its particular value, in our
opinion, is that it proposes a number of specific and testable principles that can guide
instruction and research on instruction, as illustrated, in particular, by Tsamir and Tirosh
(this volume).
In the rest of this chapter, we consider a number of key themes prompted by the chap-
ters in this part of the volume.
Nobody Learns Alone

There is no universal, “natural” course of development of understanding in mathematics.
Otherwise, presumably, Galileo would have understood Cantor’s theory of countable infin-
ity! How, and what, a student learns (apart from learning that takes place in environments
other than school) is dependent on his/her instruction. As a simple and clear example,
Tsamir and Tirosh (this volume) found that students who were learning about limits,
addressing infinity as “a single sized entity”, were liable to say that two infinite sets had
the same cardinality on the grounds that all infinite sets have the same number of elements,
whereas students in another class mainly relied on considerations of inclusion of one set
within the other.
For ascertaining research (i.e., research that probes students’ existing understanding) in
CCT, accordingly, attention should be given, as far as possible, to the students’ instructional
history. Details of the curriculum provide some information, but there is often a large gap
between the intended curriculum and the implemented curriculum. Moreover, as pointed
out by Merenluoto and Palonen (this volume) teachers’ own understanding of the relevant
mathematics may be fragile.
In order to develop CCT fully, therefore, it is necessary to conduct studies within
classes to document the nature of the teaching and attempt to relate it to the development
of students’ conceptions. The most direct way of doing this, of course, is through inter-
vention studies in which the researchers, either personally or in collaboration with teach-
ers, design and carry out the teaching. An example is provided by Tsamir and Tirosh (this
volume), who retrospectively evaluated the principles of CCT in relation to their investi-
gations. They recommend that:
Studies should be conducted to examine the impact of mathematics learn-

ing environments that are designed according to the recommendations of
Vosniadou et al. (2001), regarding students’ and teachers’ mathematical
conceptions and ways of thinking. (this volume)
Learning from History
While avoiding a simplistic interpretation (for example, through concentrating on

vignettes and individuals rather than conceptual shifts in communities of practice), there
are many powerful ways in which the historical record can inform attempts to improve
instruction through understanding of cognitive obstacles and the conceptual changes
needed to overcome them (e.g., Fischbein, 1987; Kaput, 1994; and see Greer, 2004). Most
obviously, the tenacity of resistance to conceptual change is easy to illustrate, the struggle
to accept negative integers being perhaps the best example (Fischbein, 1987, Chapter 8).
As late as 1831 an eminent mathematician could write that “3–8 is an impossibility, it
requires you to take from 3 more than there is in 3, which is absurd” (De Morgan, 1910
[originally 1831], pp. 103–104). The initial use of the negative sign was extremely
restricted in arithmetic, to the subtraction of a smaller natural number from a larger, with
well-known results. In recent years — exemplifying a case where research has impacted
practice — considerable attention has been given to the variety of situations modelled by
subtraction, thus addressing the effects of the limited intuitive “take away” model
(Fischbein, Deri, Nello, & Marino, 1985). As Christou et al. (this volume) and Vlassis
(2004) show, the extension to manipulation of algebraic expressions brings multiple new
challenges. (For a fascinating historical example, see De Morgan [1910, originally 1831,
Ch. 9].)
The Role of Representational Tools
Throughout the historical record can be traced the central role of physical devices and cog-
nitive tools, including language, representations, and notations, in extending the domain of
mathematical entities (think of Cartesian geometry, notational advances in arithmetic and

algebra [e.g., for exponentiation, see Greer, 2004], Argand diagrams for the representation
of complex numbers and operations on them, the whole range of new representational
means provided by computers [Kaput, 1992], etc.).
A problem with representations, in general, is that they are transparent to those who
understand them and opaque to those who do not. When the former group includes the
teacher and the latter the students, the dangers for the teaching/learning process are obvi-
ous, as constructivists pointed out. A further problem is that students may focus on super-
ficial aspects (sometimes unwittingly abetted by teachers seeking “short-cuts”).
To the mathematician’s eye, the number line is a way par excellence to represent num-
bers — starting with the natural numbers, adding rationals, negative numbers, and finally
the continuum of the real numbers (see, for example, the case of Eric, described by
Merenluoto and Paronen [this volume], describing his imagery of the real numbers in
terms of neighbourhoods). However, Vamvakoussi and Vosniadou (this volume) found
that the presence of a number line in a test had a limited effect on students’ performance
on tasks related to density, and that this effect did not last when the number line was
removed.
Language
The great mathematician, Poincaré, defined mathematics as “the art of giving the same
name to different things”. Often an interesting question to ask is: “Why are X and Y both
called A?” — for example, why are 3, −3.14, !, and −7i ! 6 all called numbers? In the
development of mathematics, the bestowing of a name reifies the integration of previously
disparate entities, thus creating mathematical objects that can be conceptually manipu-
lated. An excellent historical example is the definition of group (through axioms) by
Cayley in 1854 (see Freudenthal, 1990, p. 31).
Likewise, we may ask, “Why are various mathematical objects all called tangents?”
(and what have they to do with the trigonometrical usage of the same word — and what,
in turn, does that have to do with tangent as a function of a real or complex variable?). This
particular means of growth in mathematics reflects some notion of the essence of a con-
cept, with inessential elements (such as that a tangent only cuts a curve once and it divides
the plane in two parts, one of which contains the whole curve) being discarded. As pointed
out throughout the chapters, and as central to CCT, problems arise when students attempt
to maintain such elements when the concept is extended to a wider domain in which they
no longer apply.
The examples above illustrate that, despite frequent assertions to the contrary, mathe-
matical language, notation, and representations are inherently ambiguous. This ambiguity
is an advantage to those who are familiar with it, can effortlessly disambiguate through
contextual cues, and understand the conceptual linkages underlying the ambiguity, but for
the neophyte it can be a major source of confusion. A clear case in point is the extension
of use of the minus sign (Christou et al., this volume; Vlassis, 2004). Another is the poly-
semous usage of literals in algebra (Christou et al., this volume).
The Weakness of Formality
Freudenthal commented, “Traditionally, mathematics is taught as a ready made subject”

and added, “Only a small minority learn mathematics in this way” (1990, pp. 47–48).
Unfortunate results can occur when mathematicians exert undue influence over the cur-
riculum, as in the time of the “New Math”. As Freudenthal commented: “New Math’s
wrong perspective was to replace the learner’s insight with the adult mathematician’s”
(1990, p. 112). From the perspective of CCT, mathematicians are prone to see mathemat-
ics as a complete structure that can be systematically and formally transmitted to the
learner, forgetting their own conceptual struggles (which may have been minor, in any case
— see Merenluoto & Paroten, this volume).
For these reasons, it is appropriate to be suspicious of the rather fashionable character-
ization of mathematics education as the enculturation of students into the mathematical
community. As Merenluoto and Paronen point out,
‘doing’ mathematics in the school context is very different from ‘doing’

mathematics within the mathematics community. (this volume)
Their final recommendation is
to facilitate conceptual change in the number concept, we need to give the

students some access to the shared practices in the community of mathe-
maticians. (this volume)
While we agree with that, up to a point, it needs to be clarified how much, and of what
kind, “some” is, and, further, if reference to “the community of mathematicians” implies
research mathematicians then that is problematic, given how esoteric the practices of that
group have become (Davis & Hersh, 1981, p. 39 et seq.).
A particular feature of (many) formal mathematicians that has considerable bearing on
CCT is their belief in the efficacy of definitions (and other formal apparatus such as
axioms). Yet there are abundant examples in mathematics education where a learner cor-
rectly repeats a definition but then responds to a question in a way that contradicts his/her
own definition just stated. For example, a student may give a correct definition of a quadri-
lateral but immediately say that a concave quadrilateral (or a square) is not a quadrilateral.
A useful theoretical construct here is the contrast between “concept definition” and “con-
cept image” (Vinner, 1982, 1991). As Thorndike (1922) emphasized long ago, children
construct concept images and properties inductively from the sample of instances that they
encounter. For example, long experience with whole-number multiplication and division
create a strong and persistent intuition that multiplication always makes bigger and divi-
sion always makes smaller, which is no longer true when the operations are extended to
apply to rational numbers, with well-documented results (Greer, 1994). Another term that
emphasizes that people do not generally act in accordance with definitions is “prototype”.
Research in psychology (e.g., Rosch, 1978) has shown that people think of some birds —
for example, a robin — as more “bird-like” than others, such as a penguin. In mathemat-
ics, the circle is the “paradigmatic intuitive model” (Fischbein, 1987, Chapter 13) of a
curve. Fischbein cites data from Vinner (1982) who found that being able to state the def-
inition for a tangent correctly was far from a guarantee that the student could escape the
influence of the prototype or concept image. A particularly useful tool to test the solidity
or fragility of definition-based understanding is to present a singular case. For example,
Vinner (1982) found that most students could not comprehend the notion of a tangent to a
point on a straight line segment. One suggested remedy for this general pattern of results
is to “let the student into the secret” about how definitions work in the community of math-
ematicians, which is different in systematic ways from their use in everyday parlance.
The Pedagogical Dilemma
Fischbein and colleagues identified “a fundamental didactical dilemma”, pointing out that
if one tries to avoid building the ideas related to arithmetical operations on

a foundation that is behaviorally and intuitively meaningful, one certainly
will violate the most elementary principles of psychology and didactics.
(Fischbein et al., 1985, p. 15)
As a consequence:
The initial didactical models seem to become so deeply rooted in the

learner’s mind that they continue to exert an unconscious control over men-
tal behavior even after the learner has acquired formal mathematical
notions that are solid and correct. (Fischbein et al., 1985, p. 15)
The dilemma may be unavoidable, but prevalent patterns of curriculum design exacer-
bate it. A common guiding principle, consistent with the perspective of the person who
already knows the system, is to build up mathematics systematically — in the case of arith-
metic, for example, first come operations with the natural numbers, then rationals, then
integers, then real numbers. As a consequence the intuitive models can become very firmly
established. Bearing in mind that the student will see many examples of numbers other
than natural numbers in use both in and out of school, we propose that it would make sense
to deliberately and pre-emptively introduce a wider variety of numbers earlier than cur-
rently is typical.
A related point is the predominant selection of “easy” numbers that one can see in
almost any textbook, based on the well-intentioned but misguided (in our view) assump-
tion that this will make it easier for the student to understand — which may well be true
in the short term, but not the in the long term (see next section). Have you ever, for exam-
ple, seen a quadratic expression like this: 2.67x2 – 3.86x ! 12.23? Yet, with a calculator
there is no reason why a student should not be able to solve it. By planfully varying the
examples, it should be possible to avoid the establishment of expectations such as that
when a general formula for a function, such as f(x) " ax2, is presented, the parameter “a”
ranges only over the natural numbers.
The Long-Term Perspective
CCT clearly implies that curricular design must be long term. For example, attention must
be given to the series of conceptual restructurings that take place as numbers, and opera-
tions on them, are progressively extended. Yet, as Merenluoto and Paronen point out,
in … teaching practices the problems inherent in the extensions of the num-

ber concept are often implicitly passed over. (this volume)
Recent treatments of algebra (e.g., Kaput, 1999; and see Verschaffel, Greer, &
Torbeyns, 2006) have emphasized that, rather than conceiving of a transition from arith-
metic to algebra, the two strands should be intertwined from an early age (see Verschaffel,
Greer, & De Corte, in press). A particularly clear justification for this stance comes from
recent work on young children’s restricted interpretation of the equals sign whereby they
think of it solely as an instruction to carry out computation on the expression on the left,
resulting in a single number on the right. As a result, researchers recommend early expo-
sure to arithmetic equations such as 7 " 3 ! 4 to promote the broader interpretation of the
equals sign as indicating equivalence, an interpretation essential for algebra (e.g.,
Carpenter, Franke, & Levi, 2003; Seo & Ginsburg, 2003).
An extremely strong example of the building of a case for long-term curricular design
— with painstaking analysis of the historical record — was provided by Kaput who
explained:
I look closely at the origins of the major underlying ideas of calculus for
clues regarding how calculus might be regarded as a web of ideas that
should be approached gradually, from elementary school onward in a lon-
gitudinally coherent school mathematics curriculum. (Kaput, 1994, p. 78)
Revolutions in Mathematics Education?
“Are there revolutions in mathematics?” asked Gillies (1992a). We may similarly ask: “Are
there revolutions in mathematics education?” Lerman (2000) referred to “the social turn in
mathematics” (and a turn is a revolution!). As regards the research community, the work
in this part of the book suggests the onset of a conceptual shift away from the view of
mathematics learning as additive to the view that it is characterized by conceptual restruc-
turings. Since affective and metacognitive factors are also given prominence (Merenluoto
& Lehtinen, 2004) and the historical record is mined, this shift is in alignment with the
“second wave” (De Corte, Greer, & Verschaffel, 1996) of what has been called the
Cognitive Revolution (Gardner, 1985), during which the importance of historical, cultural,
social, and effective factors became recognized (Greer & Verschaffel, 1990).
However, there are only limited signs that this shift is impacting school mathematics,
and there are powerful forces that resist change and militate against long-term strategic
planning — most noticeably, but by no means exclusively, in the United States. These
forces include the pressure to maximize scores on the next test, which is a powerful motive
for not thinking about the longer-term consequences of short-term pedagogical expedi-
ency. Moreover, the positivistic style of research now being demanded in the United States
favours short-term local interventions, the results of which can be quantified, with little
hope that the findings will accumulate into any comprehensive curricular blueprint (see
Verschaffel, Greer, & De Corte, 2007).
So, much remains to be done, in terms of exposing the problems caused by insufficient
attention to conceptual change, exploring ways to facilitate it, and trying to ensure that
they enter the classroom.
References
Carpenter, T. P., Franke, M., & Levi, L. (2003). Thinking mathematically: Integrating arithmetic and
algebra in elementary school. Portsmouth, NH: Heinemann.
Dauben, J. (1992). Conceptual revolutions and the history of mathematics. In: D. Gillies (Ed.),
Revolutions in mathematics (pp. 15–20). Oxford: Oxford University Press.
Davis, P. J., & Hersh, R. (1981). The mathematical experience. Brighton,UK: Harvester.
De Corte, E., Greer, B., & Verschaffel, L. (1996). Mathematics teaching and learning. In: D. Berliner
& R. Calfee (Eds.), Handbook of Education Psychology (pp. 491-549). New York: Macmillan
De Morgan, A. (1910). Study and difficulties of mathematics. Chicago, IL: University of Chicago Press.
Dunmore, C. (1992). Meta-level revolutions in mathematics. In: D. Gillies (Ed.), Revolutions in
mathematics (pp. 209–225). Oxford: Oxford University Press.
Fischbein, E. (1987). Intuition in science and mathematics. Dordrecht, The Netherlands: Reidel.
Fischbein, E., Deri, M., Nello, M. S., & Marino, M. S. (1985). The role of implicit models in solv-
ing verbal problems in multiplication and division. Journal for Research in Mathematics
Freudenthal, H. (1990). Revisiting mathematics education: China lectures. Dordrecht, The
Gardner, H. (1985). The mind’s new science. New York: Basic Books.
Gillies, D. (1992a). Introduction. In: D. Gillies (Ed.), Revolutions in mathematics (pp. 1–14).
Oxford: Oxford University Press.
Gillies, D. (Ed.) (1992b). Revolutions in mathematics. Oxford: Oxford University Press.
Greer, B. (1994). Extending the meaning of multiplication and division. In: G. Harel, & J. Confrey
(Eds), The development of multiplicative reasoning in the learning of mathematics (pp. 61–85).
Albany, NY: Suny Press.
Greer, B. (2004). The growth of mathematics through conceptual restructuring. Learning and
Greer, B., & Verschaffel, L. (Eds). (1990). Mathematics education as a proving ground for informa-
tion-processing theories. Special issue of International Journal of Educational Research, 14(1).
Kaput, J. (1994). Democratizing access to calculus: New routes to old roots. In: A. H. Schoenfeld
(Ed.), Mathematical thinking and problem solving (pp. 77–156). Hillsdale, NJ: Lawrence Erlbaum
Associates.
Kaput, J. (1999). Teaching and learning a new algebra. In: E. Fennema, & T. Romberg (Eds),
Mathematics classrooms that promote understanding (pp. 133–155). Mahwah, NJ: Lawrence
Erlbaum Associates.
Kaput, J. J. (1992). Technology and mathematics education. In: D. A. Grouws (Ed.), Handbook of
research on mathematics teaching and learning (pp. 515–556). New York: Macmillan.
Lakoff, G., & Nunez, R. E. (2000). Where mathematics comes from: How the embodied mind brings
mathematics into being. New York: Basic Books.
Lerman, S. (2000). The social turn in mathematics education research. In: J. Boaler (Ed.), Multiple
perspectives on mathematics teaching and learning (pp. 19–44). Westport, CT: Ablex.
Rosch, E. (1978). Principles of categorization. In: E. Rosch, & B. Lloyd (Eds), Cognition and cate-
gorization (pp. 27–48). Hillsdale, NJ: Lawrence Erlbaum Associates.
Seo, K.-H., & Ginsburg, H. P. (2003). “You’ve got to carefully read the math sentence …”:
Classroom context and children’s interpretations of the equals sign. In: A. J. Baroody, &
A. Dowker (Eds), The development of arithmetic concepts and skills (pp. 161–188). Mahwah,
NJ: Lawrence Erlbaum Associates.
Sinclair, H. (1990). Learning: The interactive recreation of knowledge. In: L. P. Steffe, & T. Wood
(Eds), Transforming children’s mathematics education: international perspectives (pp. 19–29).
Hillsdale, NJ: Lawrence Erlbaum Associates.
Thorndike, E. L. (1922). The psychology of arithmetic. New York: Macmillan.
Verschaffel, L., Greer, B., & De Corte, E. (2007). Whole number concepts and operations. In: F. K.
Lester (Ed.), Second handbook of research on mathematics teaching and learning (pp. 557–628).
Greenwich, CT: Information Age Publishing.
Verschaffel, L., Greer, B., & Torbeyns, J. (2006). Numerical thinking. In: A. Gutierrez, & P. Boero
(Eds), Handbook of research on the psychology of mathematics education — Past, present, and
future (pp. 51–82). Rotterdam/Taipei: Sense Publishers.
Vinner, S. (1982). Conflicts between definitions and intuitions — the case of the tangent. In:
A. Vermandel (Ed.), Proceedings of the sixth international conference for the psychology of math-
ematics education (pp. 24–28). Antwerp, Belgium: Universitaire Instelling.
Vinner, S. (1991). The role of definitions in the teaching and learning of mathematics. In: D. Tall
(Ed.), Advanced mathematical thinking (pp. 65–81). Dordrecht, The Netherlands: Kluwer.
Vlassis, J. (2004). Making sense of the minus sign or becoming flexible in “negativity”. Learning
Vosniadou, S., Ionannides, C., Dimitrakopoulou, A., & Papademetriou, E. (2001). Designing learn-
ing environments to promote conceptual change in science. Learning and Instruction, 11,
381–419.
Chapter 22
Reconceptualizing Conceptual Change

Anna Sfard
This commentary is an exercise in reinterpretation: I will be looking at the four studies

reported in this part of the volume and asking myself what would have changed — for
better or for worse — had the researchers used somewhat different theoretical lenses. An
attempt at theoretical modification is a useful thing to do. Recurring acts of reinterpreta-
tion keep us aware of the fact that different theories may have different relative advan-
tages, whereas no one theory can “do it all.” While undertaking the task of theory
modification, I will be recapitulating my own past experience. A quarter of a century ago,
when I was beginning my career as an explorer of mathematical thinking, my research
worldview was closer to the one adopted by the authors of this book than it is today. If I
am now in a somewhat different place, it is exactly because of my habit to question my
own work in the never-ending pursuit of answers to puzzles which the current approach
left open.
All four chapters discussed in this commentary focus on the same type of phenomena,
ask the same questions, and, in their attempt to answer, use the same types of analysis. The
focus is on the regularly observed, systematically non-standard (“misconceived”) ways in
which mathematics learners solve certain types of problems. Thus, three of the four teams
— Vamvakoussi and Vosniadou, Merenluoto and Palonen, and Tsamir & Tirosh — show
that different kinds of numbers — rational, real, infinite — when first introduced, are
treated as if they possessed all the properties of the previously learned numbers. The fourth
team, Christou, Vosniadou, and Vamvakoussi, analyzes a similar phenomenon occurring in
the transition from numbers to algebraic variables. All the authors agree that the student’s
non-standard responses are indicative of the fact that she has reached a point in her mathe-
matical development where learning cannot occur by a straightforward accrual. To be
effective, the learning requires a change in the former knowledge.
The existence of such critical developmental junctures has always been stirring much
interest among students of human cognition. While in agreement on many related issues,
such as the location of the “singular points” on the developmental continuum, the diffi-
culties with which these transitions face the learner, and the behavioral indices of these
difficulties, researchers have been divided on the question of how to conceptualize what
they were able to observe. Consequently, they differed on two basic issues: What is it that
changes? and What is the mechanism of the change? For Piaget (1950), the object of

ISBN: 0-08-045355-4
330 Anna Sfard
change was the cognitive scheme sitting in the head of the learner, and the critical change
was one that required the accommodation of a scheme rather than a straightforward assim-
ilation of a new item into the existing mental structure. Vygotsky (1987) preferred to speak
about the change of concepts, and viewed the critical transitions as the developmental
junctures where everyday “complexes” reincarnate into scientific concepts. Although the
adherents of the conceptual change (CC) approach adopt the Vygotskian unit of analysis,
their tendency to talk in terms of models (e.g., synthetic) and of assimilating aspects of the
new, incompatible information in their existing knowledge (e.g., Vamvakoussi &
Vosniadou, this volume) makes their approach closer to that of Piaget than that of
Vygotsky. The theoretical consistency of the four chapters is remarkable and not any less
impressive is the authors’ explicitly stated belief in the power of CC to unify the existing
research in mathematics education under a common theoretical framework. The contribu-
tors are also in the full agreement that this theoretization can then be used to predict
students’ difficulties in mathematics learning and inform mathematics teaching
(e.g., Vamvakoussi & Vosniadou, this volume).
The authors’ enthusiasm is contagious. Indeed, there is much appeal in the CC frame-
work: one cannot but admire its elegance, its unifying power, and its applicability as a tool
for informing educational practice.1 This said, CC, like any other theory, has aspects that
may benefit from additional refinement, and the operationality of its vocabulary is one of
them. The authors of the chapters contained in this volume speak about concepts, concep-
tions, misconceptions, knowledge structures, information, models (e.g., synthetic),
explanatory frameworks, conceptual change, etc., but none of these terms is explicitly
defined. While reading the chapters I felt the need for some guidance about criteria used
in identifying concepts or conceptions and in telling the conceptual change from any other.
I also wished I had been given some operational explanation about the differences between
misconceptions, synthetic models, and the models that should not be called synthetic.
Finally, I wished to be told where the concepts, conceptions, or synthetic models are sited
and how they can be retrieved from the data. What follows is an attempt to provide what
is missing.
While trying to operationalize the term concept it is natural to turn to the work
of Vygotsky (1987), whose life project was devoted to the investigation of conceptual
development, and who defined his key term, concepts, as referring to words together
with their meaning. This description, to count as fully operational, requires that the word
meaning be provided with an operational definition of its own. An inspiration for such
definition may be found in Wittgenstein’s (1953) famous statement that meaning is the
use of a word in language. When combined together, these two descriptions, Vygotskian
and Wittgenteinian, define the term concept as referring to words together with their
discursive uses.
At this point, I could stop my attempts to clarify CC to myself. My aim seems to have
been attained: the Vygotsky/Wittgenstein definition introduced above presents the key
1
CC’s long history both in science education and mathematics education research (its roots go back at least to
Fischbein’s theory of intuitive-tacit models; see Fischbein, 1987, 2001) is an apt, if somewhat indirect, evidence
of its attractiveness.
Reconceptualizing Conceptual Change 331
notion concept in terms of features that are accessible to public inspection. As a result, we
now have a pretty good idea about what we should look at while trying to pinpoint peo-
ple’s concepts. And yet, since theories are well-organized, tightly interconnected systems
of concepts, operationalization of one of these concepts is likely to have an effect on all
the others. In the rest of this commentary I argue that this, indeed, is the case. More speci-
fically, I show that the introduction of Vygotsky/Wittgenstein’s definition makes a differ-
ence in the unit and methods of analysis, in the resulting interpretations of data, and in
practical implications.
Unit of Analysis
Vygotsky/Wittgenstein’s definition suggests, among others, that in order to get a sense
of, say, a person’s concept of number, one needs to consider the totality of this
person’s discursive activities in which the term number may appear. The combined
Vygotskian/Wittgensteinian rendition of the term concept thus makes it clear that
in research on development of numerical thinking, nothing less than the entire discourse
on numbers must constitute the unit of analysis. Since discourse is the activity of
communicating, this latter fact justifies describing the resulting research framework as
communicational.2
The new unit of analysis is, of course, much broader than the former one, the concept.
In fact, the term numerical discourse (or discourse on numbers) is the communicational
counterpart of the term knowledge of numbers.3 Almost any tenet of conceptual
change theory may now be translated into a claim on discourse. Thus, for example, the CC
statement
[S]tudents are assimilating new information into prior knowledge instead of

making radical changes. (Merenluoto & Palonen, this volume)
corresponds to the communicational claim “Students are incorporating new numerical

words and symbols into their current numerical discourse without introducing any changes
in this discourse.” Similarly, the CC assertion
[M]isconceptions … are caused when learners assimilate aspects of the

new, incompatible information in their existing knowledge. (Vamvakoussi
& Vosniadou, this volume)
may be translated into the communicational statement “Many non-standard narratives pro-
duced by students result from their attempts to apply former discursive routines to new,
incompatibly defined words and symbols.”
2
For detailed presentation of the communicational framework see, for example, Sfard and Lavie (2005).
3
This, replacement is, indeed, necessitated by several theoretical developments, and the one due to Michael
Foucault (1972) is probably most prominent among them.
332 Anna Sfard
Methods of Analysis
One may wonder whether these latter reformulations are any more than an idle semantic
game. I thus now wish to claim that they do have far-reaching consequences. Among oth-
ers, they are bound to inform methods of study.
To begin with, the term discourse, unlike knowledge, is fully operational. There are pub-
lic criteria for distinguishing discourses and thus also for detecting changes induced by
processes of learning. Numerical discourse, like any other, is recognizable by its four char-
acteristics: (1) its keywords (e.g., number words, words that refer to quantitative compar-
isons) and their routine uses, (2) its visual mediators (e.g., numerals, symbols for operations
and for numerical relations) and their routine uses, (3) its routines (e.g., computational algo-
rithms, heuristics for numerical estimations), and (4) its endorsed narratives (known also as
facts about numbers, such as “2 ! 2 " 4” or “rational numbers have the property of den-
sity”). All these characteristics are publicly accessible and straightforwardly investigable.
The public nature of the numerous aspects of discourse which need to be attended to in
research impose a number of methodological principles, some of which differ from those that
guided the authors of the studies collected in this volume. True, human talk, either vocal or
written, is the basic type of data for both CC and communicational explorer. The distinctive
features of communicational approach, however, are, first, its uncompromising attention to the
verbatim version of the interlocutors’ utterances and second, the fact that the interactions con-
ducted for the sake of data collection are always documented and analyzed in their entirety (as
opposed to their being attended only partially, as is the case when one chooses to analyze just
the interviewees’ parts of oral or written exchanges). Thus, the communicational researcher
begins her report with what was said, rather than with her own story about it. She is also aware
that being a product of interaction rather than of individual doing, students’ answers may be
sensitive even to minute changes in the wording of interviewer’s questions, not to speak about
the interviewer’s facial expressions, her institutional membership, the seemingly unrelated
things she says before and during the interview, the order in which she asks the questions, etc.
All this means, among others, that the communicationalist would be reluctant to satisfy
herself with written questionnaires, which do not give much access to the ways in which all
these factors and, in particular, the formulations of the questions might have framed the
answers. Thus, for example, one needs to consider the possibility that discursive routines
evoked by Christou et al.’s (this volume) written request to “assign numbers to the follow-
ing expressions” would have been quite different if the instruction were formulated in terms
of substituting rather than assigning and if it spelled out explicitly what it was that the num-
bers should be substituted for (in other words, I am speculating that replacing the question
“Are there any numbers … that you think ‘cannot’ be assigned to 4g?” with “Are there any
numbers … that you think ‘cannot’ be substituted for g in the expression 4g?” might have
brought a change in responses). Similarly, my experience taught me that asking “Which of
the two sets is larger?” may elicit responses quite unlike those induced by the seeming
equivalent question “Which of the sets has a greater number of elements?” (This fact is rel-
evant to Tsamir and Tirosh’s (this volume) study; the explicit use of the word number may
be found confusing by the student to whom it never occurred to associate infinite sets with
numbers.) If not followed with an interview, the written answers will always remain subject
to multiple interpretations and non-testable guesses.
Reconceptualizing Conceptual Change 333
Results of Analysis (Interpretation)
To substantiate the claim about the impact of the units of analysis on our interpretation of
observed phenomena, let us consider one of the central claims of Christou et al.’ s (this vol-
ume) study, which was tested and reportedly corroborated also by Vamvakoussi and
Vosniadou’s (this volume) and Merenluoto and Palonen’s (this volume) findings, namely
that “many students even in secondary education tend to think that all numbers share all the
properties of natural numbers.” This sentence can be translated into the communicational
claim that “the students continue to use the word number and all the related terms and sym-
bols the way they did so far.” From our own studies, in which we scrutinized children’s
numerical discourse in its diverse manifestations, we know that the learner’s use of nume-
rical keywords may be quite different from that of expert interlocutors, and that this may be
true even in the case of the discourse on natural numbers. In particular, children tend to use
the word “number” the way experts use the word “numeral,” that is, as referring to strings
of decimal digits rather than to intangible entities for which the numeral is supposed to be
but a representation.4 This interpretation of students’ non-standard answers, while fully con-
sistent with the findings of Christou et al., Vamvakoussi and Vosniadou, and Merenluoto and
Palonen studies,5 is somewhat different from the explanations given by these researchers.
This difference, I wish to claim now, is not without practical consequences.
Teaching
The interpretations provided by Christou et al. (this volume), Merenluoto and Palonen (this
volume), and Vamvakoussi and Vosniadou (this volume) for their findings rest on the
assumption that in all the cases, the interviewees and questionnaire respondents could be
regarded as fully competent in the domain of natural numbers. Indeed, according to the
authors, it was the familiarity with this basic type of number, accompanied by the conviction
that what was true now would also be true in the future, that constituted the main obstacle to
further development. Communicational researcher is less upbeat about the students’ prior
learning. According to the alternative interpretation presented above, this former discourse
may suffer from exactly the same weakness as its various extensions are known to have,
except that in the discourse on natural numbers, this weakness — the lack of objectification6
— does not lead to visibly “misconceived” responses to questions such as those asked in the
present studies. This interpretation of what the students can or cannot do has clear implica-
tions for what the teachers should aim for and how they should act to attain their goal.
4
Unfortunately, there is no space here to explain the operational criteria with which communicational researcher
evaluates similarities and dissimilarities between discursive uses of different words; let me just say that one of
the indicators of the use of number words as signifying abstract objects rather than numerals is that the user views
such different symbolical and verbal forms as 6, 1 + 5, 3·2, 24/3, etc. as fully exchangeable — the feature that is
missing as long as number words are interpreted as referring to specific numerals.
5
For the lack of space, I leave verification of this claim to the reader.
6
To objectify numerical discourse means to be able to use number words as if they signified intangible entities
rather than decimal numerals. (see Footnote 4 ).
334 Anna Sfard
Our ability to make sense of what we see depends on our uses of words. As illustrated
above, the interpretation of the notion of concept that gave rise to the communicational
framework is likely to make a difference in both research and practice. This said, I wish to
stress again that the communicational approach was presented here not as a replacement
for CC but rather as a theoretical lens of choice, to be applied whenever most appropriate.
As I am never tired of reminding to myself, no one theory would ever satisfy all our needs.
Indeed, as suitable and helpful as a theory may be in answering some questions, it will
leave some other queries unanswered. To decide which theory to use in a given context,
the researcher needs to consider, among others, its explanatory power and its potential to
lead to consensus and accumulation of knowledge. The ultimate test of any interpretation,
however, would be in practice. Each of the two approaches discussed in this commentary
yields its own practical advice, and it is the empirically testable effectiveness of these
didactic implications that may constitute the major consideration in our choice of theoret-
ical lens. In the present two cases, such empirical examination is yet to come.
References
Fischbein, E. (1987). Intuition in science and mathematics: An educational approach. Dodrecht, The
Netherlands: Reidel.
Fischbein, E. (2001) Tacit models and infinity. Educational Studies in Mathematics, 48(2,3),
309–329.
Foucault, M. (1972). The archaelogy of knowledge. New York: Pantheon Books.
Piaget, J. (1950). The psychology of intelligence. New York: Harcourt, Brace.
Sfard, A., & Lavie, I. (2005). Why cannot children see as the same what grownups cannot see as dif-
ferent? Early numerical thinking revisited. Cognition and Instruction, 23(2), 237–309.
Vygotsky, L. S. (1987). Thinking and speech. In: R. W. Rieber, & A. C. Carton (Eds), The collected
works of L. S. Vygotsky (Vol. 1, pp. 39–285). New York: Plenum Press.
Wittgenstein, L. (1953). In: G. E. M. Anscombe (Trans.), Philosophical investigations. Oxford, UK:
Blackwell.
Subject Index
affective, 3, 149–151, 160, 165, 202, communication breakdown, 70–71, 78

207–209, 211, 232, 326 communicational framework, 244, 331, 334
algebra, 239, 242, 283–287, 289–291, comprehension, 100, 147, 167–172, 175,
293–295, 310, 323, 326 179–180, 183, 190–192, 305
algebraic, 243, 252, 284–295, 322, 329 community of mathematicians, 239, 251,
expressions, 243, 284, 286, 288–295, 322 258, 324–325
notation, 287 concept, 6–7, 9, 11, 37, 40–42, 48, 50–55,
anti-realism, 22, 54–55 57–61, 63–64, 66, 68–69, 73, 78,
asymmetry, 9, 71–72, 77, 83 84–87, 89–90, 111, 124–128, 136,
approach to learning and studying, 102, 139, 149–150, 168, 171, 202, 204,
151–154, 157–159 211–213, 224, 231, 239, 241–242,
approaches to studying, 101, 123, 135, 140 247–257, 259–261, 265–269,
279–280, 283–285, 299, 302,
background assumptions, 9–11, 86–87, 92, 310–311, 321, 323–326, 330–331, 334
241 conceptions, 2–3, 11, 57, 101, 111,
beliefs, 2–8, 10, 20, 27, 36, 41, 49–50, 116–117, 123–133, 135, 137,
53–55, 59, 72, 90–91, 99–102, 105–120, 139–140, 159, 165–170, 175, 179,
125, 145–155, 157–160, 165–171, 183, 190–193, 199–201, 203–204,
173–193, 197, 208, 213–214, 222–233, 207–209, 213, 222, 224, 242, 247,
240, 243, 295, 299, 301, 303, 305–306 301–302, 311, 322, 330
conceptions of learning, 123, 125, 127,
cantorian set theory, 300–301, 304, 308, 310 129–133, 135, 137, 139, 159
certainty, 64–65, 86, 100, 117–118, 129, conceptions of knowledge, 123, 128–130,
147, 170, 225–226, 252–253, 140, 159
256–257, 259, 290 conceptual, 1–12, 17, 19, 21, 25–27,
classification, 51, 53, 58, 91, 124 31–32, 35, 37, 39, 42–43, 47–55, 57,
coherentism, 107–111, 114 59, 63–78, 83–87, 89–94, 97, 99–102,
cognitive, 1–7, 12, 20–22, 27, 39, 41, 43, 105–120, 123–129, 133, 135–137,
90, 93, 100, 106–107, 109–112, 139–140, 145–155, 157–160,
116–117, 120, 125, 131, 140, 149–150, 165–169, 171, 183, 189–193, 197,
159–160, 165–166, 191–193, 202, 204, 199, 201, 203–205, 207, 209–215,
207, 209–213, 230, 244, 250, 255, 306, 221–233, 237, 239–245, 247–253,
320, 322, 326, 330 255, 257, 259–261, 265–269, 276,
conflict, 2–3, 159–160, 204, 250, 306 278–280, 283, 285–287, 294–295,
development, 1, 4–6, 27, 320 299–301, 303–307, 309, 311,
processes, 2, 7, 109–110, 112, 116–117 319–327, 329–331, 333
science, 93 asymmetry, 83
336 Subject Index
change, 1–12, 17, 19, 21, 25–26, 31–32, discreteness, 11, 242, 266–268, 270, 273,
35, 47–55, 57, 59, 63–69, 71, 73, 75, 275–278, 284
77, 83–87, 89–93, 97, 99–102, domain learning, 228
105–107, 109–120, 123–126, domain specific, 4–5
145–151, 153, 155, 157–160,
165–169, 171, 183, 189–193, 197, education, 1–2, 11, 20–21, 26–27, 64,
199, 201, 203–205, 207, 209, 78, 113, 117, 126, 130–131, 135,
211–215, 221–233, 237, 239–245, 148, 150, 167, 192, 239–240,
247–249, 251–253, 255, 257, 243–244, 248, 261, 267, 283–284,
259–260, 265–267, 269, 276, 299, 302, 311, 319–321, 323–326,
278–280, 283, 285–287, 294–295, 330, 333
299–301, 303, 305–307, 309, 311, enumeration, 66–67, 242
319–325, 327, 329–331, 333 epistemic, 10, 55, 99, 101–102, 105–106,
development, 101, 123–124, 126–128, 108–110, 113–114, 116–120, 145,
133, 137, 139–140, 228–229, 261, 330 183, 222–233
revolution, 93–94 beliefs, 10, 99, 101–102, 105–106, 108,
system, 3, 65–66, 70–76, 90, 94 113–114, 117–118, 145, 222–233
understanding, 101–102, 106, 123, cognition, 99, 113, 116, 118, 183
135–136, 148–150, 152, 154, epistemological, 10, 12, 20, 23, 99–102,
159–160, 193 105–111, 113, 115, 117–120, 123,
constructivist, 1, 10–11, 100–102, 128–129, 145–155, 157–160,
146–149, 151, 153, 158–160, 166, 165–193, 197, 208, 214, 222–227,
197–201, 203–204, 206, 208–209, 229, 231, 233, 240
212–213, 224, 226–228, 240, 269 development, 101, 123, 129, 147
contextual, 11, 47, 100–101, 108, 123, beliefs, 10, 99–102, 105–106, 118, 120,
127, 130, 146–147, 165, 229, 323 145–155, 157–160, 165–171,
contextualism, 90 173–193, 197, 208, 214, 222–227,
counter-intuitive, 7, 126, 147 229
counterintuitive, 63, 71, 241, 304 epistemology, 38, 57, 97, 99–102,
culture of mathematics, 251, 253, 256–259 105–108, 110, 112–113, 116–120,
130, 145–147, 149, 151, 158,
development, 1, 4–6, 10–11, 19–20, 25, 166–167, 221–223, 225–228, 230,
27, 29–31, 48, 53, 60, 64, 66, 91, 232, 258
99–101, 119, 123–129, 131–133, expository text, 167, 170, 193
135–137, 139–140, 146–147, 157, exemplar, 48, 91, 93
159–160, 197–198, 202, 204, experts, 112, 120, 250, 259, 333
207–211, 214–215, 221–222, expertise, 5, 125, 222, 228, 233, 250, 252,
226–233, 241, 247–250, 252, 261, 259
265–267, 269, 283, 300, 302–303,
320–323, 329–331, 333 foundationalism, 105, 107–111, 114–115,
density, 249, 251, 254, 256, 259–260, 266, 119
269–270, 275–278, 284, 286, 323, 332
discourse, 21, 107, 117, 119, 125–126, grammar, 22, 63, 65, 67, 69, 71, 73, 75,
128–129, 131, 137, 139, 204, 206, 77, 86, 92, 241, 320
210–211, 213, 244, 331–333 grammatical conditions, 22, 64–65, 77
Subject Index 337
historicist turn, 20 knowledge objects, 123, 137–140

holism, 84, 90 Kuhn, 1–3, 5–6, 8–9, 11, 17, 19–23,
25–32, 35–43, 47–49, 52, 60,
idealism, 22, 64, 70, 92 63–65, 69–70, 72, 75–76, 78,
incommensurable, 2, 6, 8–9, 26, 40–41, 83–84, 89–94, 100, 146,
43, 52, 279, 320 193, 247
incommensurability, 3, 8–9, 11, 20–22, Kuhnian, 2, 22, 26, 28–30, 36–37, 40,
30–31, 36, 40–43, 47, 52, 70, 72, 77, 43, 65, 74, 78, 91
83–84
local, 8, 22, 41, 83, 84 language, 4, 41, 47, 64, 67, 71, 73, 78,
global, 8, 14, 42, 43, 83 86, 90, 93, 224, 248, 250, 261,
infinity, 242, 275, 279, 300–302, 305, 307, 322–323, 330
310, 321 learning, 1–11, 19–20, 25, 35–36,
initial explanatory frameworks, 243, 39–41, 47, 63, 83, 89, 91–92,
266–268, 270, 279 99–102, 105, 108, 113, 117–118,
intuitions, 9, 41, 53–54, 60, 90, 240, 249, 120, 123, 125–137, 139–140,
300–301, 321, 324 145–147, 149–155, 157–160,
intuitive, 4, 7, 38, 114–115, 126, 165–171, 173, 175, 177, 179, 181,
147, 209, 240, 243, 248, 251, 183, 185, 187, 189, 191–193,
299–301, 307, 321–322, 324–325, 197–206, 208–215, 221–224,
330 226–233, 237, 239–245, 247–248,
instruction, 1–2, 4, 7–8, 19, 25, 35, 47, 63, 251–255, 258–261, 265, 267, 280,
83, 89, 99, 102, 105, 113, 115, 117, 283, 299–303, 309–311, 319–323,
123, 126, 145, 149–150, 160, 165, 326, 329–330, 332–333
197–198, 201–202, 206–210, from text, 167
212–214, 221, 239–244, 247, goals, 10, 101, 149, 151, 230
265–267, 270, 280, 283, 295, literal symbols, 239, 242–243, 283–291,
299–302, 304–307, 311, 319–322, 293–295
326, 329, 332 logical positivists, 1, 37
justification, 66, 89, 105–110, 112, meaning, 8, 22, 25, 37, 40, 42–43,
117, 119–120, 145, 152, 166, 47–51, 54, 64–66, 69–73, 78,
224–226, 326 89–90, 93, 125, 127–132, 136–137,
139, 151–152, 154–155, 159, 172,
knowledge, 1–8, 10–12, 25–27, 36, 209, 241, 244, 265–266, 269,
38–40, 48–50, 53, 57, 63, 89–90, 278–279, 286, 294–295, 307,
99–102, 105–110, 114–120, 123–124, 321, 330
126, 128–131, 133–135, 137–140, meaning change, 51, 70
145–152, 159–160, 165–167, mediators, 209, 332
169–172, 175, 179, 183, 190–193, metacognitive, 131–132, 135, 140, 149,
198–204, 206, 208, 212–214, 159, 190–191, 193, 228, 230, 326
222–233, 239–241, 243, 247–252, 255, metaconceptual awareness, 4–5, 7–8, 10,
258–259, 261, 267, 269–270, 278, 280, 12, 125, 131, 150–152, 154, 159,
283–288, 294, 299–302, 304, 311, 319, 168–169, 171–172, 190, 192, 228,
329–332, 334 244, 303
338 Subject Index
mathematics, 1, 5, 8–11, 36, 38, 40, paradigm, 2, 8–11, 21–22, 26, 28, 30–31,
66–67, 72, 168, 198, 214, 237, 35–38, 40, 43, 47, 63–78, 83, 86–87,
239–245, 247–261, 265, 267–269, 90, 92
280, 283–285, 289, 291, 299–302, paradigm shift, 26, 31, 35
304, 310–311, 319–326, 329–330 personal epistemology, 97, 99–102, 118,
mathematics learning, 11, 237, 239–245, 145–147, 151, 166–167, 223, 227–228
247, 260, 300, 310–311, 322, 326, phenomenal sign, 243, 286–289, 291–294
330 philosophy, 2, 19–22, 27, 35, 37–38, 43,
methodological principles, 332 47–48, 90, 99, 105–107, 120, 225,
motivation, 3, 134, 149, 202, 222, 241, 320
305–306 physics, 1, 5–6, 9, 11, 23, 25–26, 28–32,
motivational, 3, 149–151, 159–160, 165, 51, 64, 69, 72, 75, 78, 84, 93–94,
230 101–102, 134, 145–155, 157–160,
167, 197–198, 201–202, 206,
naive 13–14, 162, 163, 195 209–214, 226–228, 242, 295
negative sign, 243, 288, 293–294, 322 principles, 4, 11, 31, 36–37, 56, 90–91,
normal science, 2, 9, 26, 28, 36–37, 47, 112, 117, 147, 151, 155, 159,
65, 91 198–199, 239–240, 242–243, 256,
number, 4, 7–8, 11, 22, 27, 30, 67, 73, 89, 259, 299–301, 311, 319, 321–322,
101–102, 111, 140, 148, 160, 167, 325, 332
199, 203, 206, 208, 212–214, prior knowledge, 1, 107, 114, 118, 126,
232–233, 239–243, 247–261, 133, 135, 150, 167, 169–171, 175,
265–280, 283–287, 291, 294, 179, 183, 199–201, 203–204, 206,
299–300, 302–310, 319–324, 326, 208, 227, 239–241, 243, 247–248,
331–333 251, 267, 269, 280, 283–284,
concept, 11, 239, 241–242, 247–249, 286–287, 294, 299, 301–302, 304,
251, 253, 255, 257, 259–261, 311, 331
265–269, 280, 283–284, 324, 326 puzzle solving, 21, 36–37, 91
line 249, 251, 254, 256, 257, 259, 265,
269, 272, 274, 275, 323 realism, 22, 47–49, 51–55, 57, 59, 61, 70,
numbers, 8, 67, 242–243, 247–261, 107–108
265–273, 275–280, 283–295, reference, 41, 43, 53–55, 58–61, 75, 85,
299–300, 302–305, 307, 309–310, 90–91, 93, 113, 127, 168, 201, 214,
320, 323–326, 329, 331–333 249, 268, 287, 305, 311, 324
natural, 67, 242–243, 248–250, 265, refutational text, 102, 118, 147, 165,
284, 333 167–171, 175, 179, 183, 190–193,
negative, 266, 268, 280, 287–290, 300, 226–227, 230
323 relativism, 22, 64, 76–77, 92, 128–130,
rational, 8, 242, 243, 249, 261, 284, 300, 137
324, 332 reliabilism, 105, 107–108, 112–113, 117
real, 247, 260, 267, 269, 284, 288, 299, representations, 11, 39, 94, 114,
323 125–126, 130, 169, 191, 193,
novices, 250, 259–260, 267–268, 285 243–244, 256, 259, 261, 267–270,
275–280, 301, 303, 306, 308–310,
Objectification, 333 322–323
Subject Index 339
revolutionary science, 26 study strategies, 10, 101, 135, 151,

revolutions in mathematics, 320, 326 158–159
symbolic representations, 243, 267–268,
science, 1–3, 7–9, 11, 19–23, 25–32, 270, 275–280, 309
35–41, 43, 47–49, 51–52, 58, 60, 65, synthetic models, 7–9, 11, 115, 150,
70, 73, 76, 78, 83, 89–93, 116–118, 242–243, 247, 265, 269–270,
124, 139, 146–148, 150–151, 165, 276–278, 283, 299, 311, 330
171–172, 191–192, 197–200,
203–204, 206, 208–210, 212–214, teaching, 3, 11–12, 20, 22, 25–26, 68, 71,
226–227, 240–243, 247, 258, 78, 101, 120, 123, 126–128, 133–135,
299–302, 304, 306, 311, 319–320, 140, 192–193, 197–199, 201–202,
330 204–206, 208–215, 224, 226, 239–240,
scientific, 1–5, 7–9, 11, 19–23, 25–31, 243, 245, 247, 251–252, 256, 260, 280,
35–38, 40–41, 43, 47–61, 63–66, 70, 283–284, 299–301, 303, 305, 307,
75–77, 89–93, 108, 110, 112, 309–311, 321–323, 326, 330, 333
114–117, 126–127, 146–147, 150, teaching and learning processes, 197, 215
166–169, 172, 183, 191–193, theory-laden, 36, 40, 75
198–200, 206, 211, 226, 247–248, theory-ladness, 36
299, 306, 308–309, 330
community, 3, 21, 36 understanding, 1, 3–4, 9–10, 19, 21,
practice, 36, 56, 92 26, 43, 56–57, 64, 68, 72, 74, 76, 78,
progress, 26, 77, 92 89, 94, 99–102, 105–107, 111, 113,
realism, 22, 47, 49 115, 117–118, 120, 123–139,
revolution, 1–2, 8, 11, 20–21, 25–27, 29, 147–152, 154–155, 158–160, 165,
31, 35, 40, 47 167, 190–193, 209–210,
school context, 198, 248, 250–251, 227–229, 239, 242, 244, 248,
259–261, 268–269, 324 252–256, 261, 266–270, 275–276,
shared practices, 248–249, 251, 260, 324 279, 283–284, 294, 310, 321–322,
similarity, 6–7, 84, 86, 91–93, 138, 251 325
skepticism, 55, 106–108 unit of analysis, 12, 330–331
socio-cultural, 2–4, 7, 10, 12, 150 unobservable entities, 48, 51, 53, 55–56, 59
sociocultural, 101, 111, 150, 232
social epistemology, 107–108, 112, 117 variable, 42, 112, 133, 154, 168, 175,
sociologist turn, 21 179, 183, 190, 192, 233, 284–287,
structure change, 89, 91, 93 294, 323

Reframing The Conceptual Change Approach in Learning and Instruction Advances in Learning and Instruction Advances in Learning and Instruction Ad PDF

Caricato da

Informazioni sul documento

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Reframing The Conceptual Change Approach in Learning and Instruction Advances in Learning and Instruction Advances in Learning and Instruction Ad PDF

Caricato da

Copyright:

Formati disponibili

RE-FRAMING THE CONCEPTUAL

CHANGE APPROACH IN LEARNING AND

Series Editors: K. Littleton, C. P. Constantinou, L. Mason, W.-M. Roth and R. Wegerif

Further details: www.elseviersocialsciences.com

Recently Published Volumes

VERSCHAFFEL, DOCHY, BOEKAERTS AND VOSNIADOU

TYNJÄLÄ, VÄLIMAA AND BOULTON-LEWIS

ELEN AND CLARK

HAKKARAINEN, PALONEN, PAAVOLA AND LEHTINEN

TUOMI-GROHN AND ENGESTROM

DE CORTE, VERSCHAFFEL, ENTWISTLE AND VAN MERRIËNBOER

Related journals – sample copies available online from:

Learning and Instruction

Amsterdam ● Boston ● Heidelberg ● London ● New York ● Oxford

First edition 2007

Copyright © 2007 Elsevier Ltd. All rights reserved

No part of this publication may be reproduced, stored in a retrieval system

British Library Cataloguing in Publication Data

Library of Congress Cataloging-in-Publication Data

For information on all Elsevier publications

Printed and bound in The Netherlands

P. Boscolo, University of Padova, Italy

R. Wegerif, Open University, UK

List of Tables xiii

1. The Conceptual Change Approach and its Re-Framing 1

PART 1: Foundations of the Conceptual Change Approach: Kuhn’s

3. In the Wake of Thomas Kuhn’s Theory of Scientific Revolutions:

4. Kuhn’s Philosophical Successes? 35

5. Conceptual Change and Scientific Realism: Facing Kuhn’s Challenge 47

6. Background ‘Assumptions’ and the Grammar of Conceptual Change:

8. Conceptual Change as Structure Change: Comment on Kuhn’s Legacy 89

PART 2: Personal Epistemology and Conceptual Change 97

10. Epistemological Threads in the Fabric of Conceptual Change Research 105

11. Conceptions of Learning and the Experience of Understanding:

12. Conceptual Change in Physics and Physics-Related Epistemological

13. Effects of Epistemological Beliefs and Learning Text Structure on

PART 3: Extending the Conceptual Change Approach to Mathematics

17. When We Clashed with the Real Numbers: Complexity of Conceptual

19. Students’ Interpretations of Literal Symbols in Algebra 283

22. Reconceptualizing Conceptual Change 329

Subject Index 335

Figure 10.1: Relational model of epistemological stances. 108

Figure 14.9: Achievement gains (Rasch parameters and residuals) for

Table 11.1: Contrasting forms of understanding in revising for final

Table 18.4: Number and percentage of students in categories formed with

The Conceptual Change Approach and its

The Classical Approach to Conceptual Change

Reframing the Conceptual Change Approach in Learning and Instruction

strategy. In addition, attention needs to be paid to the development of motivated and

Re-Framing the Conceptual Change Approach

Domain Specificity and Naïve Theories

As mentioned earlier, the conceptual change approach considers conceptual change as

How does Conceptual Change Happen?

Mechanisms of conceptual change We believe that if we consider that existing knowl-

The problem of incommensurability The problem of incommensurability is the claim

Epistemic beliefs The above considerations regarding the importance of background

Conceptual change in mathematics The re-framed approach to conceptual change that

mathematical knowledge. It also constitutes a valuable framework for analyzing and

The re-framed approach to conceptual change is a constructivist, domain-specific

FOUNDATIONS OF THE CONCEPTUAL

The Philosophical Foundations of the

Reframing the Conceptual Change Approach in Learning and Instruction

In the Wake of Thomas Kuhn’s Theory of