Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
natural sciences but also leads to the In defending his seemingly teleologi-
long overdue internal unification of psy- cal attitude Holt points to the empha-
chology. sis some positivist physicists—we may
We will develop our arguments first think primarily of Mach—place on the
with the use of an example involving relatively descriptive study of "func-
behavior as a "constant function" of a tional relationships"; these may con-
characteristic end state, and then in nect events over space or time regard-
reference to the functionalism of the less of the traditional tracing of causal
perceptual constancies where progress chains in near-action.
along methodological lines is somewhat Tied -variables. Functionally irrele-
further advanced. vant generalizations. Holt's problem of
constant function involves generaliza-
SYSTEMATIC DESIGNS AND THE STUDY tion. One of the most time-honored
OF DISTAL ACHIEVEMENT traditions in experimentally testing gen-
Behavior as constant function. One eralizability demands that one, or per-
of the earlier functional behaviorists, haps a few, conditions be varied in a
Holt, suggested that the movements of planful manner decided upon by the
an individual be defined in terms of experimenter while all others are held
"that object, situation, process . . . of constant. The purpose is to assure iso-
which his behavior is a constant func- lation of the so-called independent vari-
tion. . . . So in behavior, the flock of ables. For their arbitrary orderliness
birds is not with any accuracy, flying and confinement such designs may be
over the green field; it is, more essen- called "systematic."
tially, flying southward" (24, pp. 161- For example, we may move the birds
166). This statement, rather paradig- backward along the line of their flight,
matic of functionalist modes of thought, say, to position I in Fig. 1. The birds
involves selective description. The pre- may persist in their original direction
ferred hypothesis contains reference to and in this sense show generality of be-
a remote end (south); or, in the words havior. But it is easily seen that this
of Heider (21), it sets "distal deter- experiment is irrelevant to Holt's chosen
mination" over "proximal determina- alternative. The two directions, "south-
tion," that is, in our case, over de- ward" (solid arrow) and "over the
scription in terms of momentary posi- green field" (broken arrow) coincide;
tion (green field). allowing for all possible types of re-
sponse, the two variables involved—
TIED DIACRITICAL south vs. non-south, and green-field vs.
VARIABLES DESIGN
non-green-field—are perfectly correlated
and thus inseparable so far as the
available evidence is concerned. This
constitutes artificially induced perfect
confounding, and may be labeled "tied-
variables" design or, in short, tied de-
Green sign.
field Responder replication, that is, repeti-
tion with new individuals or with the
South original individuals at other occasions,
is likewise irrelevant to Holt's alterna-
FIG. 1. Systematic designs in the study of
behavior constancy (applied to an example tive, regardless of the interindividual or
from Holt, 24). intraindividual consistencies that may
REPRESENTATIVE DESIGN AND PROBABILISTIC THEORY 195
"3
o 3 ' t* •
••
!•*••
•
o.
S
o>
o
o
£
Q.
o
0>
0 I 3 4 5
Measured bodily size, 8 (log)
REPRESENTATIVE DESIGN
Fro. 3. Scattergram of an example of representative design used in the study of size constancy
(adapted from Brunswik, 6, 7). Analogously in * dimensions.
198 EGON BRUNSWIK
mary depth cues, such as binocular dis- factorial design, but would merely be
parity, are foolproof in our ecology. analogous, forming in the end an *-di-
For example, binocular disparity is mensional space in which all factors
present in the stereoscope, yet depth is could be considered simultaneously.
absent in the underlying reality; in Quite aside from the avoidance of the
viewing reality through a camera, on pitfalls of systematic design—in which
the other hand, binocular disparity is all factors held constant are lost for the
absent while depth is present in the investigation, and the resultant tied-
chain of causal ancestry. variable clusters only lead to confusion
In our present example the analysis —representative design, while cumber-
was not carried to an explicit treatment some and laborious, is thus potentially
of depth criteria and other context fac- a very economical technique.
tors. With the use of photographs, Projections of the frequencies in Fig. 3
such as those Seidner had available for upon the main axes are shown in Fig. 4. Our
his analysis of depth cues, a great va- sample, restricted to sizes not tilted into the
riety of them could be analyzed in a third dimension and thus by-passing the prob-
single enterprise. In fact, Fig. 3 must lem of shape constancy, is represented by the
solid curves. It is reassuring to find the dis-
be seen as combining but two out of a tribution for B fairly normal. (The third
practically unknown number, x, vari- graph shows the distribution for distance as
ables. Since covariation must be al- related to a dependent distance axis that may
lowed to take its natural course, the be imagined to run across the upper right-
hand part of Fig. 3 under 45°, forming a tri-
different juxtapositions would of course angle with the main axes; the crowding of
not look exactly alike, as they do in both small and intermediate objects along the
J. 2 > 4 5 6 0 1 3 * -i 6 ' i a
1mm 1cm lOoulm lOnlDOm llcm 1mm la 10m 10cm 1m 10m DOm Ikn 10km
BODILV SIZE (log of mm) FROJECTIVK SIZE DISTANCE (log of meters)
(log of mm at 1 m distj
Fio. 4. Frequency distributions of three ecological variables for the representative design
shown in Fig. 3 (solid curves; the dotted curve includes tilted objects and the broken curves
refer to a subsample of vertical objects, both not discussed here). (From Brunswik, 6.)
REPRESENTATIVE DESIGN AND PROBABILISTIC THEORY 201
5 * *
o •
*T
V
/.*
S* •
N
'5 3
•5
ff I \-
• •. V .V*
•o
<o
4*
O
S<b
0 on
E |
in
o ofi
oo
UJ 0 r <§
0
t 1 ' 1 ' 1 ' 1 1 1 i i . i . i
0 1 2 3 4 5 0 1 2 3
Measured Measured
bodily size, B(log) photographic size, P (log)
Fio. S. Scattergrams of perceptual achievement (functional validity) for a subject respond-
ing to the representative design shown in Fig. 3. (Open symbols are used here to designate ob-
jects smaller than 10 cm., see below; the differences in shape are irrelevant in the present con-
text.) (Adapted from Brunswik, 6, 7.)
10-inch limit of near vision leads to skewed degree of perfection of size constancy
distribution and is the chief source of the
curvilinearity that may be observed in Fig. 3.)
may again be sought in the correlation
coefficient. This represents what may
Functional validity (achievement). be called the "functional validity" (7)
We now turn to covariation between or "achievement" (Leistung, 2; for defi-
the distal stimulus and the response nition see 5, p. 255) of the class of re-
variable which will define perception in sponses, b, relative to the distal stimu-
its approximation to "constant func- lus variable, B. For the total sample
tion" as understood by Holt. Figure S, of 93 situations this correlation is close
left, shows the stimulus-response scat-
to .99 (more accurately, .987 when com-
tergram for the logarithms of the per-
ceptual estimates, b (so labeled because puted from ungrouped logarithms with
they were given in the natural, naive- three significant places [see 7, p. 44]),
realistic attitude toward bodily size, in spite of the occasional sizable errors
B) as plotted against the logarithms of referred to above.
B. The entries cling fairly close to the By contrast, the functional validity of b
diagonal, much more closely than the relative to photographic size, P (Fig. 5, right),
major cue variable, P, was found to is only .73; in addition, this is quite close to
cling to B in Fig. 3. Appraising the the purely ecological association of B and P,
maximum errors committed in the par- .70, so that little independent focusing of b
on P is indicated. Perceptual restructuring
ticular estimates, however, we find that toward photographic size (painter's attitude,
some entries deviate as much as one- p) raises the correlation of the response with
half of a power of 10 from the diagonal, P no more than to .85. Other attitudes,
corresponding to about three- to four- deliberately inviting intellectually controlled
fold over- or underestimation of the judgment ("betting") and thus the "stimulus
error," will not be discussed here, nor shall
length in question. we go into the question as to whether or not
A generic summary description of the the estimates b and p are purely perceptual.
202 EGON BRUNSWIK
dividual differences. For example, we this kind, rather than the emotive or
may use partial correlation as a mathe- motivational aspects of value, per se,
matical means of holding constant a prompted the study of what could be
certain variable. Partialing out P from called perceptual value constancy. In
our above correlation between b and B experiments concerning the apparent
(and thus in effect reducing the eco- numerosity of stamps and of coins,
logical validity of P from .70 to 0) still Zuk-Kardos and Fazil (reported in 2,
yields a functional validity as high as pp. 140-1 SO) as well as Ansbacher (re-
.98; whereas factually eliminating the ferred to in 7) found number con-
sizes under 10 cm, in Fig. 3—to the stancy with value variant and value
right of log B — 2—and thus reducing constancy with number variant fairly
the ecological validity of P from .70 to high, although tainted with compromise
.14 in a quasi-systematic move, reduces between the two variables.
the functional validity to .95 (see also Another quasi-representative step is
7, Fig. 9 and the accompanying text). to gear the manner of variation or of
It must also be noted that, in con- covariation between variables to the
tradistinction to systematic design, the general scheme of natural conditions in
process of analysis may be stopped at a planfully controlled way, as when the
any point, falling back on the nonre- association between a certain cue or
ductive aim of functional research, to- means and the object or reward is made
gether with the assurance that the un- probabilistic rather than absolute (4,
resolved part of the associations is 10).
safely within the fold of the ecology to Experiments centered about an ex-
which the investigation has been geared emplary instance. Successive omission
from the beginning. vs. successive accumulation of cues. A
certain effort toward representativeness
SYSTEMATIC EXPERIMENTS WITH is discernible whenever a "lifelike" situ-
REPRESENTATIVE FEATURES ation is taken as the starting point of
We now turn to certain experimental the experiment. In the field of size per-
policies, some of them common, which ception such experiments are likely to
may be considered transitory between abandon, at least in some of their
systematic and representative design. phases, the chin rests, darkrooms,
Representative features may be in- screens with small openings, alleys of
jected in otherwise systematic designs edges without thickness, or other labo-
in a variety of ways. ratory paraphernalia in vogue during
Quasi-representative choice of vari- the late nineteenth century. This lib-
ables and of their variation or covaria- eralization owes much to David Katz
tion. Some measure of representative- (29) and his work in another area of
ness may be achieved by the choice of perceptual constancy—color constancy
variables with particular life relevance, with illumination variant. For size con-
such as "value," as a factor in constant stancy, a study by Holaday (23; see
function. Since ecology embraces cul- also 7, p. 23), and the studies by Hoi-
tural norms held valid by the law en- way and Boring (25), by Gibson (16),
forcement policies of a society, along and to some extent one by Joynson
with those connected with physical law (28) have proceeded by essentially the
or geographical contingency, monetary same functional scheme.
value becomes a challenge to perceptual In each case there is what we may
attainment on a par with other object call a core or "exemplar" situation—
properties. Cognitive considerations of somewhat arbitrarily chosen, to be sure
204 EGON BRUNSWIK
psychologists have seen in such a re- ism; but it does not include the uni-
ductive explanation of behavior the ma- vocality of prediction which is the ma-
jor task of psychological theory. The jor raison d'etre of the nomothetic ap-
nomothetic behaviorism of Hull and of proach (5).
his sympathizers, mentioned at the be- We may therefore take the position
ginning of this paper, reveals its reduc- that in the end the unity of science is
tionist aim most clearly in the use of a better served by allowing the reaffirma-
"physiologizing" terminology. tion or elaboration of this unity to be
To the study of distal function and superseded by a working out of the the-
of its grand strategy the reductive ap- matic diversity of the sciences within
proach adds the study of tactics; to the the minimum common platform. This
study of achievement and of its macro- diversity of themata involves both the
mediation, both of which fall within the aim of the different disciplines and the
province of functional-representative de- designs capable of serving these aims.
sign, it adds the study of microme- Thematic physicalism and the nomo-
diation, which falls in the province thetic-reductionist-systematic syndrome.
of nomothetic-systematic design. The The different explicit and implicit trends
functional approach has its place mainly toward the unification of science which
in the appraisal of the interplay and have dominated the last half century
relative contribution or weight of fac- have been under the spell of a some-
tors in the adjustment to a given ecol- what stereotyped image of physics. The
ogy, while the reductive approach re- thematic element in this cliche may be
veals the technological details of the explicated as the emphasis on the "gen-
machinery that brings about such ad- eral," and notably on the strict, uni-
justment. vocal regularities or laws which pos-
Theory as comparative methodology: sess universal applicability; this is also
observational unity and thematic di- known as the nomothetic approach.
versity of science. The injection of Universality of law presupposes homo-
physiology into the discussion brings us geneity of the universe; hence it mat-
to a branch of psychological theory ters little where and when and over how
which overlaps with the philosophy of large an area a phenomenon is studied.
science and is best labeled comparative Experimental design may thus safely be
methodology of science. One of the left to the convenience and liking of
major concerns of such a science of sci- the experimenter and thus become sys-
ence, or metatheory, is with the basic tematic. And, as Mises (36) has pointed
unity of the sciences. In present-day out in discussing probabilism in phys-
psychological discussion this problem is ics, macrolaws have their origin in dif-
subsumed mainly under the watchword ferential equations, that is, in principles
"operationism." Here it is often for- conceived for minute space-time splin-
gotten that the basic requirement for ters. The triad of nomothetic aim,
scientific exactitude is a relatively mod- microreductionist procedure, and sys-
est one, and in a certain sense a more tematic design which we have come to
commonplace one than anticipated. It recognize as a syndrome in traditional
involves no more than the inter- and psychology thus is revealed as an emu-
intrasubjective univocality of observa- lation of a pattern indigenous to the
tion and communication which is some- specific thema of physics.
times called "methodological physical- As has been pointed out in greater
ism" (8), but should better be specified detail elsewhere (8, sec. 9), the basi-
as observational or procedural physical- cally elementistic character of physical
REPRESENTATIVE DESIGN AND PROBABILISTIC THEORY 209
law is not obviated by, and has little to mapping. The fact that no application
do with, the empirical fact of a prob- of the general physical laws is possible
abilistic microstructure in thermody- without the constants, parameters and
namics or within the atom. Nor must boundary conditions furnished by geo-
such structures be seen as related to, or graphic types of information is fre-
in any way supporting, our arguments quently neglected in theoretical discus-
for the probability approach in psychol- sion. Except at the level of control
ogy; functional psychology is macro- ordinarily accessible only to the physi-
probabilistic in that the identity of the cist observer, these constants are not
individual case is maintained (as in a available (at least not fully so) to the
scattergram) while it is lost in the responding organism. Hence the chains
physical macrolaw which for ordinary from select distal to proximal to cen-
practical purposes is absolute. tral variables in perception are chains
General vs. particular. Physical law, of (probable) partial, rather than of
geographic fact, ecological correlation. total, causation. The universal lawful-
The first to warn against the overesti- ness of the world is of limited comfort
mation of the general over the particu- to the perceiver or behaver not in a po-
lar in science was Windelband (52). sition to apply these laws, and he there-
In a somewhat stilted application to fore must rely largely on whatever
personality and to the humanities, he snitches of particular or semigeneral-
was led to suggest the well-known dis- ized information he may be able to as-
tinction between nomothetic and idio- semble. This is what we meant earlier
graphic disciplines. Unfortunately, the in this paper by the assertion that ordi-
latter term encompasses both the low- narily organisms must behave as if in
brow, strictly enumerative approach to a semierratic ecology.
historic-geographic fact seen in isola- With data from our representative
tion, and the extremely high-brow em- size-constancy survey described above,
phasis on the "unique" lawfulness of the relation between physical law and
the individual or culture; more prop- ecological correlation is illustrated in
erly, the latter case should therefore be Table 1. The first row of ecological
labelled the "idionomothetic" approach. validities has been introduced above in
Within the natural sciences, an ex- discussing Fig. 3, and the remaining
ample of the purely enumerative ap- two pairs are added here from our ma-
proach is given by those branches of terial. Partial correlations derived from
geography that deal with topographical either of the two columns of three co-
TABLE 1
TEXTURAL ECOLOGY AND PHYSICS
(Adapted from Brunswik 7, Fig. 10)
Nomothetic Approach
Ecological Validities
Variables Partial Correlation
Correlated Law of
Full Sizes over Physical
Sample 10 cm. Variable Coefficient Optics
(N - 93) (N - 59) "Held Constant" Obtained
from the tying of variables and from perimposes artificial "laws" upon an
other characteristics of nonrepresenta- ecology which it thereby depletes.
tive designs have rarely been completely In the case of the systematic experi-
avoided. The history of science fur- ment it is these artificial laws in the de-
nishes ample evidence of harmful effects sign which, as we have suggested, are at
of systematic design upon practice or least in part responsible for the often
theoretical outlook. As a layman, one striking lawfulness of the results. At
may think of hygiene in medicine with this point the suspicion arises that the
its dramatic changes between the com- didactic role which systematic experi-
plete neglect and the excesses of aseptic mentation obviously plays in the men-
or antiseptic policies; the boiling of tal economy of,the scientist, by virtue
milk and devitaminization; anemia and of the simplicity and order it both re-
the eating of liver, and so forth. All of quires in the design and furnishes in
these practices appear as playballs of the result, may outweigh the fact-find-
variables arbitrarily selected for study, ing competence of systematically de-
which thus acquire undue prominence signed experiments. Certainly the more
and throw the picture of the interplay drastically simplified forms of art of
of factors out of balance; and the short- which we have spoken, and which are
cuts involved in the ecological overgen- so similar in pattern to /the systematic
eralization of results may be even more experiment, are clearly didactic rather
serious. The biochemist or nutritionist than informative in any realistic sense;
who shuns eggs but is a chain smoker, we do not go to the movies to find out
and the cancer specialist who does the about life or to form a scientifically air-
opposite, are too familiar examples to tight theory about personality, in spite
require elaboration. of the fact that movies may sometimes
Systematic design in cliche literature. be helpful in temporarily smoothing the
We may also regard the "world" of perplexities of life.
popular novels and movies as an arti- The main function both of art and of
ficial, cultural subecology. Its outstand- systematic experimentation, then, is to
ing feature is the presence of cliches; shake and mold us by exaggeration and
these include both personality stereo- extreme correlation or absence of cor-
types and plot formulas. Cliches are relation. But exaggeration is distortion,
similar to experiments using systematic and this distortion must in science even-
design. Factors that show some degree tually be resolved by allowing the more
of independent variability in real life palatable systematic design to mature
are artificially tied. Old-fashioned opera into, and to be superseded by, the more
plots or soap melodramas that "drip truthful representative design.
with generosity" share with cowboy
movies what David Hume would have CONCLUSION: UNITY OF SCIENCE AND
called "inseparable" associations of no- UNITY OF PSYCHOLOGY
ble character, overpowering strength, Our considerations in comparative
fairness, courage, youth, final success, methodology have brought into focus
and so forth, or of their opposites. The the thematic diversification that is pos-
cliche' is a "worn" case or incident, by sible within the over-all unity of ex-
no means impossible or nonexistent but act, observationally physicalistic sci-
made prominent out of all proportion to ence. They may facilitate the cheerful
its frequency, and to the detriment of relinquishment of the overheated nomo-
all other types of incident. Like the thetic bias under which the development
systematic design of experiments, it su- of psychology has long suffered and
216 EGON BRUNSWIK