A Study of Thinking

A Study of Thinking by Jerome S. Bruner; Jacqueline J. Goodnow; George A. Austin; Roger

W. Brown
Review by: Robert Oppenheimer
The Sewanee Review, Vol. 66, No. 3, The University of the South 1858-1958: The Centennial
Symposia (Summer, 1958), pp. 481-490
Published by: The Johns Hopkins University Press
Stable URL: http://www.jstor.org/stable/27538750 .
Accessed: 10/06/2014 21:29

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].

The Johns Hopkins University Press is collaborating with JSTOR to digitize, preserve and extend access to The
Sewanee Review.

http://www.jstor.org

This content downloaded from 62.122.73.213 on Tue, 10 Jun 2014 21:29:20 PM

All use subject to JSTOR Terms and Conditions
 ARTS AND LETTERS 481

A STUDY OF THINKING*

By ROBERT OPPENHEIMER

In the spring of 1949, the Massachusetts Institute of Technology

held a convocation devoted to the celebration of recent successes in the
sciences, and to reflections on their portent. Mr. Churchill, then Prime
Minister of the United Kingdom, commented in the course of his ad
dress, "The Dean of the Humanities spoke with awe, cof an approach
ing scientific ability to control men's thoughts with precision.' I shall
be very content to be dead when that happens." This book which is
before us need lend no haste to Churchill's thoughts of death, nor to
those of the many others who, without words or with words of less
austere finality, share his anxieties.
Even the lay reader will recognize in this book some fresh and solid
steps toward an understanding of characteristic traits of man's rational
behavior. He will also see that the psychological sciences have a very
long way indeed to go. For A Study of Thinking has in many ways
the flavor of the opening of a new science. This is not because its au
thors so regard themselves. They are learned in the literature of their
science; they make discriminating and frequent use of the findings of
those who have worked in the study of perception, of concept forma
tion, of linguistics and of learning. They are happy to recognize that
the logical ideas to which they are led are as well explicated by Aris
totle as ever since; it is clear that the models of von Neumann and
the theory of games have contributed both to their explicit terminology,
and to their ways of thinking and of formulating problems.
But the book has a unity of view and a fervor of conviction which
makes it point to the future. If it raises more questions than it answers,
the are new and because there are some answers
questions newly sharp
in the book. Although there is little in the findings here reported that
shocks or transcends common sense, these findings are cast with the
and objectivity which is indeed the mark of a science finding
precision
its bearings.
We are concerned throughout with the discovery and creation of
order in man's cognitive life. We are dealing with a work which both
*A Study of Thinking. By Jerome S. Bruner, Jacqueline J. Goodnow, the late
George A. Austin; with an on Language by Roger W. Brown. Publication
Appendix
of the Harvard Cognition Project. New York: John Wiley and Sons, Inc. London:
Chapman and Hall, Limited.

This content downloaded from 62.122.73.213 on Tue, 10 Jun 2014 21:29:20 PM

All use subject to JSTOR Terms and Conditions
 482 ARTS AND LETTERS

studies a part of this great theme and, in itself, exemplifies it. The
theme is vast. Man has a great capacity for discrimination. His po
tential sense of otherness is almost unlimited. Rational life begins
with the selective
practice of ignoring differences, failing in truth to per
ceive them; rational life begins with the failure to use discriminatory
power in anything like its full potentiality. It lies in the selection, ar
rangement, and appropriate adequation to the objects of perception
and thought, of limited traits, of a small residue of potential wealth.
Dr. Brown
in his Appendix assaults the analogous problems of lan
guage, language made possible only because so many differences of
sound and tone, aspiration, and articulation are ignored in the forging
of a common tongue. Lest this creation of order from chaos be seen
wholly as attrition, one the powers of discrimina
anticipatory finding:
tion which are subordinated in learning and in language, are not lost;
they are in a sense set to one side, or on the back burner. They do not
come easily and into use; but once there is evidence that
immediately
they are relevant to the subject of learning or thought, they can be
back. Thus we can learn whose
brought languages phonetic and sonic
distinctions differ in profound ways from those of our own; thus it is
that we can not only see the world as ordered, but can learn when we
have missed the point, and can attain new levels of insight, complexity,
richness and structure.

A physicist, reflecting on these matters, is likely to be of reminded

his experiences with the quantum theory. In perception,
in learning
concepts, only a ^mall part of our potential is engaged.
perceptivity
The small part, if we are to be successful, must have a proper adequa
tion to our theme. The rest cannot be involved in the operation if
there is to be clarity or order at all. Some of the rest will be involved
in other problems and other contexts. These limited constellations of
our human powers are used in a complementary none of them ex
way,

hausting what we could do, but each in a measure excluding the con
current use of all the others. In physics we have learned that in ex
on an atomic one and the
periments scale, experimental arrangement
words and ideas appropriate for describing the findings of that experi
ment, are complementary to others which might indeed have been used,
which will again be used, but which cannot be used concurrently.
What then is this book all about? Certainly not about the whole of
perception and of thinking; certainly not even about the whole of the

This content downloaded from 62.122.73.213 on Tue, 10 Jun 2014 21:29:20 PM

All use subject to JSTOR Terms and Conditions
 ARTS AND LETTERS 483

processes of abstraction and categorization. But the authors tell us

in their preface,

This book is an effort

to deal with one of the simplest and most
ubiquitous phenomena of cognition: categorizing or conceptualiz
On closer it is not so The
ing. inspection, simple. spirit of the
inquiry is descriptive. We have not sought "explanation" in terms
of learning theory, information
theory, or personality theory. We
have sought to describe and in a small measure to explain what
happens when an intelligent human being seeks to sort the en
vironment into significant classes of events so that he may end
by
treating discriminably different things as equivalents. In dealing
with the problem, we have found ourselves far afield.
travelling

More specifically, the book is an extended analysis, and a report of

some done earlier by these and other
experiments, scientists, but the
majority done for testing the analysis and answering the questions
which it raises. It deals with situations neecessarily almost ludicrously
abstract and unlifelike, with how people learn to
categorize, with how
they learn to recognize instances that they see or hear as belonging
or not to a class, as illustrating or not
belonging illustrating a concept.
It asks by what procedures people learn to do these
things. Are they
procedures which show a certain consistency of logical method? Do the
various steps in learning form a coherent pattern, or is each step a
more or less random probe? Can one
perceive the logical structure un
derlying the effort of learning? Can one describe it as a strategy of
learning? And to the extent to which the answers of these questions
are affirmative, can one understand the choice of strategy in terms of
its difficulty, its economy, its certitude, its abstractness, the burden it
puts on memory and inference? How is the choice of strategy affected
by the circumstances of the learning process, the pressure of
time, the
availability of confirmation, the penalties attaching to error, the purity
or allusiveness of the material being ordered?
In short, this is an inquiry, limited to the most controll
antiseptic,
able and gamelike situations, into one great part of human reason:
the learning of categories and concepts on the basis of an experience
either wholly, or in some cases to
only partially, logically adequate
teach them. But neither the authors nor the reader will mistake the
vast scope of the domain to which this research is relevant. "The re
search we have reported has mainly been drawn from the field of
'concept formation' so-called, but we would that our conclu
propose

This content downloaded from 62.122.73.213 on Tue, 10 Jun 2014 21:29:20 PM

All use subject to JSTOR Terms and Conditions
484 ARTS AND LETTERS

sions are applicable to any phenomenon where an organism is faced

with the task of identifying and placing events into classes on the basis
of using certain criterial cues and ignoring others." Or again, very
near the end,

We have chosen to call this volume A Study of Thinking. A word

in explanation of this title, brings the enterprise to a close. Concept
attainment is, to be sure, an aspect of what is conventionally called
thinking, and in this sense the title justifies itself. But we have
also urged a broader view: that virtually all cognitive activity in
volves and is dependent on the process of categorizing. More cri
tical still, the act of categorizing derives from man's capacity to
infer from !sign to signif?cate, and in so far as we have shed any
light on categorizing as such it is our hope that we have also made
clearer the nature of inference as a psychological phenomenon.

It is in passages like these that the authors point with firm hope to
the future. With even greater frequency, they acknowledge their debt
to their colleagues, present and past, and even to the "Zeitgeist," that
has made it natural for psychologists to turn their attention to man
afs a rational being, and not only to the problems of his appetites, his
folly, and his will.
How have the authors set about their enterprise? The main part of
the book is divided into two principal sections: the first is primarily
an analysis of categorizing activity, and its relations to general infer
ence; and the discussion of the factors involved in learning how to sort
the environment into functionally significant categories and equivalence
classes. Here we find discussed the form in which decisions on how
to learn occur in practice, how sequences of such decisions come to de
fine a learning strategy, what general considerations may favor one or
another course of learning, and how one may alter the weight of these
considerations by changing the "problem." It is right that these des
criptive and analytic chapters come first, before the discussion of the
experimental material; the experiments could hardly be intelligible
without this. Yet the authors themselves state that the clarity which
illuminates these earlier chapters, and gives order and sense to the
sequence of experiments, and meaning to the questions that they an
swer, emerged after long periods of observation and did not underlie
the initial designs. ". . . the ideal strategies that have served us so
steadily in this chapter are essentially refined versions of what we
have observed our subjects doing. They were not invented by us in

This content downloaded from 62.122.73.213 on Tue, 10 Jun 2014 21:29:20 PM

All use subject to JSTOR Terms and Conditions
 ARTS AND LETTERS 485
an a priori manner. Our description of ideal strategies is a descrip
"
tion of what, it seemed to us, our subjects were trying to 'bring off.'
Perhaps this also has to do with the hopeful quality of the book. If
in these simple experiments, stylized as we have said and piteously
artificial, enough could emerge to warrant an ordered review of so much
of logic and of common sense, the experiments themselves take on an
interest far greater than any single one of them reported in this, their
summation.

Throughout the whole of the analytic discussion, and throughout

most of the presentation of the experiments, the authors see, in the
learning problems which they have studied, a very close analogy to the
situations confronting the scientist when he plans his research pro
gram and tries to decide between alternative concepts and theories,
which are at the moment still compatible with the evidence. Some parts
of science are clearly not touched on in this?the moments of discov
ery, the times of creation. Here, rather, is the scientist doing his con
scientious best, disentangling puzzles fairly well defined and finite. In
deed, it is the part of science which is most like a game that is cldsest
to the situations dealt with in this study. The rules are well defined;
the information is honest; there are no tricks: a world, to alter Ein
stein's "Raffiniert ist der Herr Gott, aber boshaft ist er nicht,"?in
which God is not only not malicious but not even very subtle. But
no is where most of us as and
this, doubt, begin children, this, for
sure, is where most of us still stay most of our time as men.
Chapters 4 to 7 describe the experimental designs, the experiments
themselves, and the questions they raised; and chapter 8 returns to
the summary of what has been learned, and to some of the principal
questions raised to a new sharpness by the findings and their analysis,
questions only partially answered or as yet wholly unanswered.
What do the experiments do? They are designed to make manifest
the sequence of choices the subjects use in trying to learn. The verbal
accounts the subjects give are occasionally referred to; they are not
part of the experiment.

In studying concept attainment, then, it has been our aim to

externalize for observation as many of the decisions as could pos
sibly be brought into the open in the hope that regularities in these
decisions might provide the basis for making inferences about the
processes involved in or a . . . To
learning attaining concept. put
the matter perhaps too simply, the analysis of performance strat
egy consists in comparing the actual performance of a subject with

This content downloaded from 62.122.73.213 on Tue, 10 Jun 2014 21:29:20 PM

All use subject to JSTOR Terms and Conditions
486 ARTS AND LETTERS

a set of rational or ideal and determining a best fit.

strategies
... In the study of thinking, inference, and
conceptualizing,
other such diversely labelled activities, the great technical problem
is precisely this one. If behavior is to be viewed as strategy, the
task of analysis can only be accomplished by devising experiments
that can get a lot of sequentially linked behavior out of the or
ganism where it can be observed.

The subjects of the experiments were students, mostly from Harvard,

a few from Radcliffe. The number of subjects in one given experi
ment runs in the tens, seldom in the hundreds. The students are as
signed a task of learning, usually from visual material, cards on which
objects are printed or abstract, some sketchily the
drawings, mostly
matic. As far as the experimentor can arrange it, the world of the
student is limited to these cards that he may see or select or be shown.
The students are told that
they are to learn about a class which con
tains some but not all of the cards. They are told whether the defining
concepts are conjunctive (round and blue), or disjunctive (round or
blue or both), and whether the relationship between the instance and
the category is certain or only probable. Then the students are watched
going about their
learning. Sometimes there are limited opportunities
for selecting instances, or for volunteering a solution; sometimes not.
Sometimes the cards are presented with maximum order; sometimes
not. Sometimes what is on the cards refers to a human world; some
times not. What goes on is duly noted. On occasion, things which
cannot be recorded as part of the experiment, such as an expression of
at or an evidence of mastery, are noted down as a pro
pleasure insight,
blem for the future.
The four chapters on experiments, sometimes reviewing earlier work,
but for the most part undertaken for the purposes of this study, are
called, "Selection Strategies in Concept Attainment," "Reception Strat
egies in Concept Attainment," "On Disjunctive Concepts and Their
Attainment," "On Categorizing with Probabilistic Cues."
We shall turn now to a sketchy and incomplete account of some of
the findings. But before we do that, in order to stress again the ab
stract nature of the experiments, we quote the final sentences of this
study. "We have idealized the experimental situations employed in our
investigations beyond what is normal in daily life, but that is the price
one pays for experimenting at all. It is our hope that by reaching a
fuller understanding of these more idealized forms of thinking in vitro,

This content downloaded from 62.122.73.213 on Tue, 10 Jun 2014 21:29:20 PM

All use subject to JSTOR Terms and Conditions
 ARTS AND LETTERS 487
the complexities of thinking as we observe it around us will be more
readily understood."
Huizinga, who sees human culture and man's rational life as an out
growth of Homo Ludens, would, no doubt, concur.
A Study of Thinking is replete with findings based on experiments
and their analysis. They are almost all statistical findings, in the sense
that the traits of behavior discovered can not be predicated with uni
versality and certainty of every subject. One of the findings, for in
stance, is that when students are trying to learn a disjunctive concept,
they fare far better when their first instances are negative. This does
not mean that every sequence started with a instance gives
negative
the quickest possible nor does it mean that no
learning, subject whose
first instance is positive learns quickly. But the statistical traits are
striking, and strikingly different from a record of chance behavior; and
in none of the important conclusions is there any question of their sta
tistical significance. The samples are small, but they appear to be
more than large enough.
They are, however, quite specialized, and the
question of whether Hindus, Senegalese, or Senators would behave like
Harvard students is left over for the future.
The samples which are here briefly sketched seem typical to this re
viewer in their interest, in their scope, and in the extent to which they
seem to or refine common sense views. I have at
confirm, refute,
to select findings which could be
tempted presented with a minimum of
specialized language, and this has caused me to omit many which are
sharp and, to me, most illuminating. Even where no very specialized
language is needed, we must remember that these refer to well-defined
experimental situations, that they are in this sense objective, and in
this sense limited.
A. It is possible, in the problem solving behavior of a subject, to
identify a strategic pattern, and to get a measure of strategic con
. . it is to describe
sistency. ". possible and evaluate strategies in a
relatively systematic way, both in terms of their objectives and in terms
of the steps taken to achieve these" . . . "Each of the is
strategies
amenable to relatively rigorous description, and it is fairly simple to
employ a quite precise measure for describing the shift from one strat
egy to the other." Thus the authors indicate that they can usually
identify a strategy from observation of the subject's choices and se
quence of choices. It is all-important for the future of their work that

This content downloaded from 62.122.73.213 on Tue, 10 Jun 2014 21:29:20 PM

All use subject to JSTOR Terms and Conditions
488 ARTS AND LETTERS

this identification be established as objective, in the sense that the cri

teria for identification can be learned by others.
The authors write further, ". . . it is possible to demonstrate the ef
fect of relevant conditions upon measurable aspects of categorizing stra

tegies. By so doing, we have been able to 'get into' the process of con
cept attainment. . . ." in
Subjects general adopt strategies appropriate
to the problem before them, the amount of information they are being
given, and the instructions of the game. And then again, "In general
we are struck by the notable flexibility and intelligence of our subjects
in adapting their strategies to the information, capacity, and risk re
quirements we have imposed on them." If he expects to be allowed
to continue, the student will take steps, the value of which lies in the
future use of the information they yield. Some of the best strategies
on purely logical grounds are unused because they require too much
inference and too much thinking to be done readily in the head.
Through all this, a strategy is a sequence of decisions made by the sub
ject in an attempt to learn the concept. Through all this, the question
of the relevance of education and culture comes often to mind. How
universal are these traits?
B. A second set of findings has to do with a comparison between in
stances based on pictures, pictures of people that evoke stories and
familiar scenes on the one hand, with pictures of shapes and figures of
a neutral and abstract quality on the other. The thematic examples
reduce the logical appropriateness of the strategies employed, and gen
to lead to the of a series of hypotheses evoked
erally tend testing by
the themes. ". . . in attempting to differentiate exemplars from non
as one so frequently must in science, medicine,
exemplars of a category,
and indeed in daily life, the person will, in the absence of other infor
mation, tend to fall back on cues that in the past have seemed useful,
whether these cues have been useful in an analagous situation or not."
This is surely not the whole reason why the questions which most move
and touch men are among the hardest to answer.

C. A third set of findings has to do with decisive evidence for the

aversion and the ineptitude that are involved in the negative: in hand
ling the negative instance, in learning from the indirect test, and in
about disjunctive concepts. It is not that negative
learning anything
instances are not essential. They are in fact the principal instrument
of correcting error and learning. But they are not liked and there is
a marked "tendency on the part of subjects to utilize only positive in

This content downloaded from 62.122.73.213 on Tue, 10 Jun 2014 21:29:20 PM

All use subject to JSTOR Terms and Conditions
 ARTS AND LETTERS 489
stances as a basis for Subjects also do not like
forming hypotheses."
to use indirect tests of a They do not like to have to trans
hypothesis.
form negative information into positive information or a positive con
clusion. subjects do not like and show little skill in attaining
Finally,
disjunctive concepts. The authors think that this may be a character
istic of our culture. "It is a characteristic of much scientific thinking
to assume at the outset that whatever behaves in a common way does
so for a common cause." This finding is clearly relevant to the pro
blem of causality in history.
D. Strategies and the purposes of strategy are markedly affected,
among many other things, by the extent to which the student can learn
whether he is right or wrong. If he is told of his errors, he works
hard to eliminate them and to solve the problem. If he has little op
portunity to know whether he is right or wrong, he abandons this hope
and adopts a prudential policy, keeping error within limited bounds.
Such is our government.

E. In a beautiful little experiment on categorization of spoken sounds,

the Harvard student regarded as distinctive sounds characterized by
different letters in English. The mono-lingual Navajos, not Harvard
students, categorized by the length of the vowel. Once the material
was so arranged as to indicate the possible relevance of vowel
length,
English-speaking students categorized by it too. This is an example
of a theme running through the whole book: powers of discrimination,
unused but latent, are resurrected when evoked by the
problem.
None of the findings sketched above, and indeed very few in the
book as a whole, upsets what we would have expected, or outrages
common sense; but in every case
they suggest further studies; and in
most cases, the questions to which they lead find no firm answer in
common sense. This reviewer, at least, does not know the answers.
A Study of Thinking ranges through territory which has been for
millennia a battleground of epistemology and metaphysics. It is thus
not astonishing that from time to time the authors may be found tem
porarily manning one of the old trenches. Thus at one point they say,
"To be sure, the defining criteria in terms of which equivalence classes
are formed exist in nature as discriminate." This does
potentially
not keep them from writing, further on, "The categories in terms of
which we group the events of the world around us are constructions
?r inventions. . . . do not 'exist' in the environment." Nor is
They
Dr. Brown able to avoid a bout with Locke and Berkeley. But the

This content downloaded from 62.122.73.213 on Tue, 10 Jun 2014 21:29:20 PM

All use subject to JSTOR Terms and Conditions
490 ARTS AND LETTERS

authors' more steady view, it appears to me, they record as follows:

"We have found it more meaningful to regard a concept as a network
of sign-significate inferences by which one goes beyond a set of ofe
served criterial properties exhibited by an object or event to the class
identity of the object or event in question, and thence to additional in
ferences about other unobserved properties of the object or event."
That man must act in order to know, that he must thereby reject
other actions of which he is capable, and lose other knowledge of what
is knowable in the world, will not solve the old philosophical questions;
but it will alter, deepen, and illuminate them.

POOR MONKEY! *
NOW, GOD HELP THEE,

By CHARLES TOMLINSON

Mr. Coveney's study of the child in literature (its title comes from
the words of Lady Macduff to her son) is typical of a good deal of
the critical writing that appears in England at the present. At its best
level, his book is highly informative and stimulating. At its average
level it seems uncertain of the nature of the audience it has to deal
with, and consequently disperses much of its force in retailing back
ground information of the kind that the educated reader already pos
sesses. At its worst level Poor Monkey is somewhat derivative in sub
stance and tired in tone. Now this is a pity. For the book possesses
a real core and its materials are potentially and, at times, actually a
valuable addition to our notions about literature.
The author traces the passage of the literary image of the child from
Blake to Lawrence. He sees in the Romantic use of that image a
creative symbol, "a focal point of contact," as he says, "between the
growing human consciousness and the 'experience' of an alien world."
In short, the child focuses a disquiet and, at the same time, the artist's
hopes for human salvation: it is a touchstone for life. The corruption
of that symbol provides Mr. Coveney with matter for some of his most
interesting and original chapters; the symbol of growth gives place to
its antithesis, to the Victorian image of the "dying child" and to the
*Poor Monkey: The Child in Literature. By Peter Coveney. London: Rockcliff. 30s.

This content downloaded from 62.122.73.213 on Tue, 10 Jun 2014 21:29:20 PM

All use subject to JSTOR Terms and Conditions