Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Contents
List of Figures
ix
Preface
Michael Lissack and Abraham Graber
xiii
Acknowledgments
xvii
Context
1 Introduction: Thoughts on Explanation
Michael Lissack and Abraham Graber
A Place in History
Alicia Juarrero
17
25
Case Study
4
59
75
93
Dialogue
7
109
115
Contents
121
10
Getting a Grip
Nancy J. Nersessian
133
11
143
12
151
13
Economic Explanations
Paul Thagard
161
14
171
15
197
203
Conclusion
Michael Lissack and Abraham Graber
215
229
233
241
249
Reprise
Michael Lissack
257
References
263
285
Notes on Contributors
289
Index
295
MODES OF EXPLANATION
2014026012
CHAPTER 1
his is a book about explanation. Its origins lie in the all too frequent
observation that our way of thinking often does not match the world.
Such mismatches give rise to ambiguity and uncertainty. The ambiguity, in turn, acts as both a constraint on possible actions (including the
action of reliable prediction) and the desire to explain what is going on.
Explanation is the name for the process we use to answer the questions raised
by observed ambiguities. Explanation is also the name for the product of such
processes. This process/product divergence is merely a hint of the many conf licting approaches to be found in the contemporary understanding of explanation. This book is the first in decades to attempt to bring these conf licting
approaches together and to offer a compelling narrative to explore how those
conf licts can converge.
Such convergence is important because explanation is important. Often we
work with an idiosyncratic conception of explanationa conception that may
not match those of our neighbors. In this dissonance lies both potential gain
and potential trauma. The lack of an explanation often leads to either creative
inquiry or troubling confrontations between holders of differing beliefs. Such
occurrences may be found even when some believe that an explanation has
been forthcomingan explanation that others find explains nothing.
Explanations are central to our way of navigating the world. Some explanations appear in the everyday life of the average person. Thus, the best explanation of the fact that there is dog food all over the kitchen f loor is that, while
we were away at work, Fido got into the food. Other explanations are more
rarified. For example, one might explain the blueness of the sky in terms of
the comparatively long wavelength of blue light and the comparative predilection of longer wavelengths to disperse when passing through the atmosphere.
There are important similarities and differences between these two sketches of
explanation.
Contemporary philosophy is characterized by a fascination with explanation. The philosophical literature on explanation is rapidly expanding; the
philosophical literature that attempts to use explanations is vast. This fascination with explanation appears to correspond with the contemporary trend
toward naturalized philosophy. More and more, philosophers are coming to
take their cues from the sciences. Thus, philosophers are increasingly expending energy on studying the methods and results of the sciences. Explanation
appears to be central to the practice of actual scientists; a brief glance at
scientific practice suggests that scientists are in the business of offering
explanations.
This focus on scientific practice, however, overlooks an important set of
practitioners who also rely heavily on explanation. Explanation is important
for managers, consultants, entrepreneurs, investors, and so on. The parallels
between the work of these practitioners and the work of scientists are notable.
Just as scientists construct explanations to make sense of observed phenomena, practitioners create explanations to make sense of the world around them.
Furthermore, just as scientists use accepted explanations to make the world
respond as they want it to, practitioners rely on explanation to navigate the
complicated social, financial, and political world that they inhabit. In each
case, explanations allow humans to manipulate the world successfully. Put
another way, explanations offer affordances (Gibson, 1977). Some of these may
be affordances for action; others are affordances for prediction.
There are, however, also important differences between the ways in which
scientists and practitioners construct and use explanations. The scientist aims
to use established explanations as a starting point for the production of further, true explanations. In practice, the scientist makes use of explanations
as the basis on which to make predictions. Successful predictions, in turn,
help to generate the theories that then become the basis for further explanations. The practitioners aims are more pragmatic. The practitioner relies
on explanations insofar as they are useful; that is, truth is incidental to the
practitioners aims. Explanations have value if they lead to affordances for
action. Explanations have little value if they do not create an affordance but
rather merely offer more description. For the scientific realist, an explanation
is good if it is accurate. Pragmatic success is, at best, a secondary desideratum.
Priorities are reversed for the practitioner: for the practitioner, an explanation
is good if reliance on the explanation leads to pragmatic success. Truth is, at
best, a secondary goal.
The distinction is perhaps best illustrated by considering two disciplines,
each of which is interested in offering explanations: physics and economics.
Physics offers reductive explanations in terms of the properties of the constituents and sub-constituents of matter. For the past century, Western thinking has
been guided by the physics paradigm: the world is organized around discrete
objects that aggregate and have simple relationships. Everything is explainable
through rules, laws, and algorithms. The observer is not a part of the observation but is external to the closed systems under consideration.
Economics carries the mark of the last century of Western thought and so
is modeled on the physics paradigm. Physics has been strikingly successful;
economics, less so. There are at least two fundamental differences between
the object of study of physics and that of economics. While physics studies the
interactions of mindless particles, economics studies the interactions of autonomous and semiautonomous agents. Furthermore, while in the study of physics
the physicist stands outside of the closed system being studied, the same cannot
be said of the economist.
Despite the successes of the frame of thinking that characterized physics,
it has a serious deficiency: How can it be that the actions and behaviors of
reflexive, anticipatory creatures are best described by rules for nonthinking,
non-reflexive, non-anticipatory objects? How can it be that context is deemed
not to matter? And what about complexity or those relationships that cannot
be described by the simple? The physics-based frame has no answer and instead
discards these issues with the magic words ceteris paribus (all other things
being equal). Ceteris paribus clauses need not be problematic for the physicist,
for physics studies closed systems. However, we do not live in a closed system;
thus, the need arises for some other frame of thought to enable our tools for
understanding to be adequate for the world around us.
The philosophical literature suggests that explanation and understanding, while intertwined, are also different. Ricoeurs (1973, 1974) hermeneutical method, for example, unfolds through the dialectic of understanding,
explanation, and comprehension. Understanding seems to be better thought
of as the acceptance of a structure into which the target understanding can
be comfortably placed. Another way of saying this is that understanding
involves locating the target into a context in which it seems to be coherent. While contexts are often quite large, the frames we use when seeking
to explain need not be. If the mode of our explanation is to place the target
into a pre-given structure, then both context and frame will be as large or
small as the structure itself. If, by contrast, the mode of our explanation is
to detail a mechanism for how something happens or the conditions that
allow for action to occur, then the context will be large but the frame rather
small. This contrast between frame and context ref lects the notion that each
explanation we encounter contributes to the larger environment that in the
aggregate makes up our cognitive understanding. This contrast also sheds
some light on the role that recursive inquiry among description, explanation,
and understanding can have in constituting and revising our cognitive environs (cf. Runciman, 1983).
Forms of explanation are themselves context dependent. Social systems differ from physical systems in that the use of theories changes the behavior of
social systems. As participants in these systems act, they do so on the basis
of reflexive consideration of context, goals, and affordances drawing on their
own mental models (which are themselves the product of prior contexts and
current attention) in anticipation of possible outcomes. These recursive ref lexive considerations (or as Piaget (1929) would have called it, learning through
actions) have no parallel among the physical sciences. The additional considerations give rise to questions of objectivity, discovery, and the basis for scientific
explanation.
The basis for social sciences and design (pragmatic assumptions) is different from the hard sciences. There is a need to deal with ideas and communication in social systems. Thus, the philosophy of science needs expansion to
include paths to the potential logics of the social sciences. Example questions
might include asking What is the basic unit (individual, group, set, dynamic,
environment, etc.)? Sciences of the sentient will require different languages
and different frameworks of thinking than are commonly used in the hard
sciences of non-sentient beings. Meta-level thinking is an opportunity that
can create the need for new strategies of simplification so as to meet requisite
variety.
Objectivity and a goal of reliable predictivity are the hallmarks of what we
shall label Science 1. These are the hard sciences as traditionally taught and as
used as references by philosophers of science. Physics is the exemplar of Science
1. In the Science 1 world, we label and categorize via deduction, probabilistic
inference, and induction. Science 1 excludes context dependence; thus, when it
is forced to deal with the possibility instead asserts ceteris paribus.
Discovery and attunement to context are the hallmarks of what we shall
refer to as Science 2. In the Science 2 world, we instead seek to identify relationships, affordances, and potential actions. We ask questions rather than
seek to label or categorize. Science 2 explicitly makes room for the context
dependencies that Science 1 has excluded. These can be characterized as emergence, volition, ref lexive anticipation, heterogeneity, and design, among others. The philosophical sources necessary to understand the hermeneutics of
social experience can be found in the field known as systems sciences, with
a focus on the underlying models, feedback loops, ref lection, and anticipation that goes by the label of systems thinking. In the social science modeling
embraced by systems science, apparent inconsistencies raised by the inclusion
of the observer are replaced by a need to pay close attention to processes and
to multiple adjacent possibles. Once participants are admitted as part of the
process being modeled and their decision-making and design abilities are taken
into account, then the multiple possibilities to which they give rise must also
be taken into account and not seen as contradictory. The broad applicability
of context dependence and observer questions throughout the anticipatory sciences demands the exploration of both logical foundations and narrative application. The possibility for implementation or action lies in the reconciliation
of experience and models in the anticipatory science.
The inability of Science 1 models to capture the essence of Science 2 events
adequately has been well documented. For example, consider social science
domains where reflexivity and reflexive anticipation are characteristic traits of
actors. Actors can become reflexive by learning and by modifying their cognitive repertoire. More advanced forms of ref lexive anticipation at the actors
level occur when actor A possesses an image of actor Bs image of A, actor B
an image of actor As image of B, and so on (the explicit basis of interaction in
Figure I.1
world as we label and categorize it. Along the complex and attuned side of the
surface lies the world as we act in it.
These two sides of the surface have strikingly different characteristics despite
being part of a continuous surface. The simple and ordered side on the right
(Science 1) corresponds roughly to our traditional way of thinking. It excludes
context dependence. It is the world of reliable predictions, truth claims, and
invariants. The complex and attuned side on the left (Science 2) corresponds to
a more relationship way of thinking. It explicitly includes context dependence.
This is the world of affordances, anticipations, and actions. It is devoid of truth
claims in favor of abductive hypotheses.
The very notion of what counts as an explanation seems to differ between
these two worlds. Adherents of both worldviews in general agree that a description of a mechanism in response to a how? question constitutes an explanation. The disagreements arise over the kinds of answers offered in response to
a why? question, those that tend to arise when an expectation is not met.
While the Science 1 worldview inquires why as a means of revealing truth
and will keep asking until this criterion is met (an optimization strategy), the
Science 2 worldview inquires why as a foundation for further action (or nonaction) and will stop asking when a satisfactory narrative has been offered (a
satisficing strategy). The discussion that follows will attempt to outline the
basis for these orthogonal divergences.
Traditionally, Science 1 is concerned with regularities. Thus, observations
of individual events or occurrences are important only to the extent that the
occurrence of such an individual event is the basis for the falsification of a
claim about regularities. Within the Science 1 context, the answers to questions of why concern the placement of regularities (observed or conjectured)
within an overall schema of regularities. The relevant questions seem to be
those of order and of fit. Both further descriptions of regularities within an
ordered regime (functional explanation) and measurements of adherence to a
pure (noncontingency messed-with) regularity are both offered and accepted
as explanatory. The how? question implied by the why? questions is How
does this fit within the established order? where the answer is a mechanism for
how fit happens.
Some of the unarticulated assumptions in the Science 1 worldview are the
pre-givenness of an established order, the idea that there should be fidelity to
that order, that the correct granularity for inquiry is at the level of regularities,
and that regularities can be referred to adequately by labels and models. Given
these assumptions, it is reasonable to eliminate contingency with a further
claim of ceteris paribus, to treat fit as measurable, to rely on noun forms, and
to posit truth as a justificatory variable. While each of these reasonable
approximations can be discarded in the pursuit of better explanation, our
human cognitive limits and our reliance on the least action principle enable
us to simplify why explanations in the Science 1 world as category membership questions, and allows a pragmatic scientific realism to guide the articulations of the abbreviated worldview that results.
In the Science 2 world, the focus is on individual actions and occurrences,
whereas the regularities of Science 1 are part of the context in which these individual events occur. In Science 2, the why? questions tend to demand answers
in the form of narrativehere are the constraints/affordances that given this
particular context allowed or prevented a particular action. Once again, the
least action principle combined with human cognitive limits means that
while a particular context includes an infinitude of variables, the observer/
actor is limited in what is attended to and processed. The regularities that are
the granular focus of Science 1 frequently are treated as assumed in the attention/cognition processing of Science 2. The granular focus of Science 2 is on
individual actions and events and the regularities are part of the context.
Some of the unarticulated assumptions in the Science 2 worldview are the
contingency and context dependence of any observed or assumed order, the
idea that fidelity to any particular order only has relevance as part of an
observation/expectation/further action feedback loop, that the correct granularity for inquiry is at the level of individuality, and that regularities can be
referred to only contingently by labels and models. Given these assumptions,
it is never reasonable to eliminate contingency with a further claim of ceteris
paribus (for in that claim one might eliminate the explanatory variables themselves), to treat measurement of deviance from expectations as a further contingent variable in the feedback loop, to rely on verb forms, or to posit actions
as a justificatory variable. While each of these reasonable approximations
can be discarded in the pursuit of better explanation, our human cognitive
limits and our reliance on the least action principle allow us to simplify
why explanations in the Science 2 world as narratives about affordances and
constraints, and allow a pragmatic constructivism to guide the articulations of
the abbreviated worldview that results.
The differences between Science 1 and Science 2 echo as we seek to answer:
What do we mean by explanations and how are we comfortable with them?
The two perspectives ask different questions that might affect what we believe
or do not believe about explanations. Are we looking at the right things? Are
Figure I.2
11
While Science 1 explanations allow for and are structured around predictions, Science 2 explanations suffer from the contingencies accompanying
explicit rejection of ceteris paribus. As such, these explanations by definition
allow for a series of possible errors that are seldom found or asserted in the
Science 1 world. These errors include the possibility of the wrong model being
used, the wrong contingencies happening or failing to happen, overlooked context, inadequate metaphor, inappropriate synecdoche, misdirected awareness or
attention, intervening volition or coercion, and incommensurable worldviews.
Despite these differences, we must remember that Science 1 and Science 2
are on the same surface and part of the same continuum (that Mobius strip in
Figure I.1). As such, our mission is trying to make sense of this giant muddle,
define what we mean, suggest where it might work and where it might not, and
then try to explore what it is that we are talking about.
Worldviews
Some of our contributors have suggested that the muddle of explanation and its
meaning that we describe above can be clarified when approached from the
perspective of scientific realism; still others suggested that the answer can be
found in the perspective known as pragmatic constructivism. The philosophy
of science literature often portrays these perspectives in opposition; much like
the worlds of Science 1 and Science 2. It can be very tempting to attempt an
overlay and then to suggest that Science 1 can be mapped to scientific realism
and Science 2 to pragmatic constructivism. However, as the collection of chapters in this book will reveal, such a mapping is far too simple and overlooks the
very nuances that make the question of explanation of interest.
Exploration requires perspective and the philosophy of science offers two
perspectives that seem to be helpful. Scientific realism is often modeled as
taking Newtonian physics to be the paradigm instance of science: other sciences are understood via assimilation to the Newtonian model; explanations
are understood to be reductionist and law driven. While the scientific realism
practiced by scientists and philosophers is much more nuanced, what it shares
with the common-sense version is an underlying belief in the independent
existence of reality and of the fundamental importance of truth. The takeaway
of importance here is that scientific realism makes truth claims, judges those
claims for coherence against a pre-given world, and affords as real entities
whose existence cannot be observed and can only be inferred.
The pragmatic constructivism approach begins by asking what actions are
being contemplated and how judgments regarding those actions can be arrived
at. The key to these observations lies in the recognition of the ontological
difference between natural entities and those that are the product of human
constructionwhile the natural entities can be referred to as pre-given
and thus described (functional explanation), human constructions are always
changing and requisite explanations demand mechanisms and explication of
relationships. This form of constructivism is less concerned with the idea that
man constructs reality and more with the notion that what matters is the
13
Approximately half of the authors in this volume hold one belief or the other
with respect to scientific realism and some form of pragmatic constructivism.
The interchange between these two worldviews formed the heart of the interesting dialogue during our event: Modes of Explanation. Both perspectives
have a concern for explanation by means of category membership. Still other
kinds of explanation raise concerns for one perspective and not the other. For
example, functional explanations are not explanatory from the perspective of
pragmatic constructivism, because a functional explanation fails to create any
kind of first-order affordance for action (descriptions may provide background
information, and thus create a second-order affordance, but fail to create an
enablement or a constraint on action in and of themselves). By contrast, the
two kinds of context-dependent explanations that play critical roles in pragmatic constructivismexplanations that point to aspects of the context that
enabled such-and-such and explanations that point to aspects of the context
such that, were these contexts absent, this-and-that would have come about
are similarly not considered to be explanatory from the perspective of the scientific realist, for each kind of explanation points to contingent features of the
world as opposed to bottoming out in robust, exceptionless laws.
Rich (2011), in his farewell column in the New York Times, noted that the
pressures of writing for a readership can push you to have stronger opinions
than you actually have, or contrived opinions about subjects you may not care
deeply about, or to run roughshod over nuance to reach an unambiguous conclusion. We believe that unambiguous conclusions about the nature of explanations are a mistake and thus have undertaken to find a way to preserve the
very ambiguity that gives nuance its due.
To accomplish this, a concept that we believe helps to reconcile the Science 1
and Science 2 perspectives is the notion of concurrent but orthogonal. Science
1 and Science 2 are indeed different, but they are not oppositional. They are
also not super-positional, where one would claim a status of truth only in the
light of a revealed contingency. Concurrent but orthogonal suggests a simultaneity that is perpendicular, much like the planes in Figure I.3.
Figure 1.3
Perpendicular planes.
Of course, these planes are not existing in their own space, but have a contextthus the shape we suggest looks more like the plane-crossed ellipsoid in
Figure I.4.
Initially we conjecture that the intersecting planes can be thought of as
the two Sciences (1 and 2) and the two philosophies of science (scientific
realism and pragmatic constructivism). This conception helped us shape the
conference, the workshops, and this book. Yet the exercise of organizing,
gathering, speaking, listening, transcribing, questioning, editing, and writing has led to a revision in this conception. We now are suggesting that the
two planes are those of ontologyrepresented by the Mobius strip of the two
Sciences and epistemologyrepresented by a model of question generation that we will further discuss in chapter 5 and beyond. This questiongeneration model represents a guide to the pragmatic hermeneutic process
(see Figures I.5 and I.6 ).
Figure I.4
Plane-crossed ellipsoid.
Figure I.5
Question-generation model.
Figure I.6
15
Question-generation model.
Index
Index
297
Index
299
Index
Raiffa, 1612
Rakover, 31, 35, 271, 273
Ravetz, 171, 268
Rawls, 221, 277
realism, 4, 9, 11, 1315, 279, 32, 37, 3945,
479, 514, 73, 75, 77, 7991, 93, 958,
1001, 1045, 1712, 175, 215, 2206,
2301, 235, 2378, 2491, 258, 261,
2647, 2691, 274, 27680, 285, 287
recursion, 5, 7, 12, 31, 151, 1545, 217
Regt de, 26, 141, 267, 288
regularities, 89, 22, 312, 36, 39, 43,
512, 110, 112, 21617, 224, 282
representation, 12, 14, 26, 28, 35, 37, 39,
413, 456, 97, 99102, 107, 116,
118, 135, 1379, 153, 1634, 167, 185,
2001, 2046, 21820, 222, 2345,
237, 2601, 277, 281, 283, 286, 2903
Rescher, 30, 277, 288
Reutlinger, 31, 277
Rich, 13, 49, 195, 260, 277
Ricoeur, 5, 277
Robertson, 93, 98, 277
Rorty, 1516, 21, 28, 49, 104, 220, 277, 281
Rosen, 39, 95, 99100, 172, 174, 177,
1845, 260, 266, 2778
Rosenberg, 40, 278
Rota, 967, 278
Runciman, 5, 278
Russell, 50, 286
Russo, 31, 269, 278, 286
Salmon, 22, 26, 30, 33, 43, 47, 51, 201,
2034, 243, 272, 278
Salthe, 11519, 222, 263, 274, 278, 290
Samarapungavan, 46, 265
Satish, 198, 278
Sayer, 32, 42, 53, 278
Schalk, 201, 266
Schank, 32, 46, 52, 54, 2789
Scheines, 128, 27980, 283
Schiele, 199, 279
Schruijer, 201, 266
Schueler, 38, 40, 279
Schutz, 218, 279
Scriven, 30, 32, 206, 279
Searle, 38, 43, 53, 153, 157, 279
sentience, 6, 45, 48, 188
Shakespeare, 166
Shermer, 29, 31, 279
301
Index