The intent of this paper is the pres- interest and curiosity. There are ex-
entation of an associative interpreta- tended analyses of novels and novel-
tion of the process of creative thinking. ists, poems and poets, mathematics and
The explanation is not directed to any mathematicians, both biographical and
specific field of application such as art autobiographical. Perusal of the intro-
or science but attempts to delineate spections of manifestly creative in-
processes that underlie all creative dividuals uncovers a surprising vein of
thought. similarity in the processes they de-
The discussion will take the follow- scribe (Ghiselin, 1952). Thus we
ing form, (a) First, we will define find Albert Einstein's self-searching to
creative thinking in associative terms suggest that "The psychical entities
and indicate three ways in which crea- which seem to serve as elements in
tive solutions may be achieved—seren- thought are certain signs and more or
dipity, similarity, and mediation, (b) less clear images which can be com-
This definition will allow us to deduce bined . . . This combinatory play
those individual difference variables seems to be the essential feature in
which will facilitate creative perform- productive thought." Samuel Taylor
ance, (c) Consideration of the defi- Coleridge is described as having de-
nition of the creative process has sug- veloped his ideas in the following man-
gested an operational statement of the ner : "Facts which sank at intervals out
definition in the form of a test. The of conscious recollection drew together
test will be briefly described along with beneath the surface through the almost
some preliminary research results. chemical affinities of common ele-
(d) The paper will conclude with a ments." In the field of art, we find
discussion of predictions regarding the Andre Breton referring to a collage
influence of certain experimentally by Ernst as being distinguished by a
manipulable variables upon the crea- "marvelous capacity to grasp two mu-
tive process. tually distant realities without going
Creative individuals and the proc- beyond the field of our experience and
esses by which they manifest their to draw a spark from the juxtaposi-
creativity have excited a good deal of tion." Most explicit, however, is the
1 oft-quoted statement by the mathema-
The essence of this paper was written tician, Poincare, who talks about an
while the writer was a Visiting Research
Psychologist at the Institute of Personality evening when "ideas rose in crowds; I
Assessment and Research, University of felt them collide until pairs interlocked
California, Berkeley. The author wishes to so to speak, making a stable combina-
acknowledge his indebtedness to Benton J. tion. By next morning I had estab-
Underwood for his encouragement and the
stimulation of his paper, "Orientation to lished the existence of a class of Fuch-
research on thinking" (1952). Work on this sian functions." From these experi-
material has been supported by the Coopera- ences, Poincare felt that he could state
tive Research Program of the Office of
Education (Project No. 1073) and the Na- that "to create consists of making new
tional Science Foundation (Grant G-3855). combinations of associative elements
220
THE ASSOCIATIVE BASIS OF THE CREATIVE PROCESS 221
facts from the fishbowl, looking for ing creative solutions may be deduced.
new and useful combinations. His Any ability or tendency which serves
procedure represents the operational to bring otherwise mutually remote
embodiment of this method of achiev- ideas into contiguity will facilitate a
ing creative solutions. creative solution; any ability or ten-
Similarity. The requisite associative dency which serves to keep remote
elements may be evoked in contiguity ideas from contiguous evocation will
as a result of the similarity of the inhibit the creative solution.
associative elements or the similarity Listed below are several illustrative
of the stimuli eliciting these associative predictions concerning individual dif-
elements. This mode of creative solu- ferences that one may make from
tion may be encountered in creative this theoretical orientation.
writing which exploits homonymity,
rhyme, and similarities in the structure Need for Associative Elements
and rhythm of words or similarities in It should be clear that an individual
the objects which they designate. The without the requisite elements in his
contiguous ideational occurrence of response repertoire will not be able
such items as alliterative and rhyming to combine them so as to arrive at a
associates may be dependent on a fac- creative solution. An architect who
tor such as primary stimulus general- does not know of the existence of a
ization. It seems possible that this new material can hardly be expected
means of bringing about contiguity of to use it creatively.
associational elements may be of con-
siderable importance in those domains Associative Hierarchy
of creative effort which are less di- The organization of an individual's
rectly dependent on the manipulation associations will influence the proba-
of symbols. Here we might include bility and speed of attainment of a
certain approaches to painting, sculp- creative solution. There is a whole
ture, musical composition, and poetry. family of predictions that one may
Mediation. The requisite associa- draw from this concept of the associa-
tive elements may be evoked in con- tive hierarchy. As an initial example,
tiguity through the mediation of com- let us take the question of the manner
mon elements. This means of bringing in which the associative strength
the associative elements into contigu- around ideas is distributed. If we
ity with each other is of great impor- present an individual with the word
tance in those areas of endeavor where "table," what sort of associative re-
the use of symbols (verbal, mathemati- sponses does he make? The individ-
cal, chemical, etc. . . .) is mandatory. ual who tends to be restricted to the
For example, in psychology, the idea stereotyped responses, such as "chair,"
of relating reactive inhibition and cor- may be characterized as having an as-
tical satiation may have been mediated sociative hierarchy with a steep slope
by the common associates "tiredness" (see Figure 1). That is, when you get
or "fatigue" (Kohler & Fishback, past the first one or two conventional
19SO). responses to the stimulus, the individ-
ual's associative strengths to other
INDIVIDUAL DIFFERENCES words or ideas (lower in the hier-
From the definition given above, the archy) drops rapidly. We can also
factors that will make for individual conceive of a second sort of individual
differences in the probability of achiev- whose associative hierarchy is charac-
THE ASSOCIATIVE BASIS OF THE CREATIVE PROCESS 223
first thing that strikes us is that while on the number of door openings.), does
some requirements have been stated, not interfere with the normal use of the
there are even more that are strongly refrigerator, and does not require spe-
implied and essential, many that are cial training. Note that the principle
desirable, and a number that we would behind this highly creative solution
only become aware of after some (not allowing massive build-ups of
method of satisfying them had been frost) was infrequently suggested in
suggested. the classroom group. However, this
Let us examine some possible so- solution is not wholly successful at
lutions : meeting some criteria. The frequent
1. Simply refraining from opening heating and cooling may injure frozen
the refrigerator door would solve the food stored near the heating element.
problem as stated since this would pre- Secondly, since the heating process
vent moisture from entering and con- must be brief and mild, it is inevitable
densing as frost. This solution meets that not all frost is removed. While
many of the implied requirements. It this solution does effectively curtail the
is cheap, convenient, effective, does number of defrostings, it does not elim-
not require special training, etc. . . . inate them completely. It is clear that
However, it is not an optimal solution a method which would encompass all of
since it violates one essential, implied the advantages of the "counter" method,
requirement—the usefulness of the but which would, in addition, eliminate
refrigerator must not be impaired. defrosting altogether would be even
(This is the cutting-off-your-nose-to- more creative. What is suggested by
spite-your-face solution.) this discussion is that the creativeness
2. A primitive solution is the of a product is some function of the
hammer-and-screwdriver method. This number of requirements that the prod-
is tried and true and meets many of the uct meets. The most ready application
essential requirements. It falls down of this definition will be in laboratory
in that it is inconvenient, messy, un- research in which tasks, solutions, and
economical (when caked with frost, the requirements may be arbitrarily con-
refrigerator unit is very inefficient), en- structed and varied.
dangers the mechanism, and is hardly
automatic. A TEST OF CREATIVITY
3. In a refrigerator we once owned
another solution was used. The open- The definition of the creative process
ing and closing of the refrigerator door has suggested a way of testing for indi-
operated a counter. At a certain count vidual differences in creativity. The
the refrigerator unit was automatically test items are intended to require the
heated and the melted water evaporated testee to perform creatively. That is,
outside the refrigerator. The superi- he is asked to form associative elements
ority of this solution is immediately ap- into new combinations by providing
parent. The source of this superior- mediating connective links. Since the
ity lies in the number of requirements test situation is contrived, the combina-
which it meets. It is economical, auto- tion must meet specified criteria that
matic, convenient, peculiarly appropri- are experimenter imposed.
ate (the operation of the heating The definition dictates the structure
element is contingent upon the number of the test. We must provide stimulus
of door openings. The amount of frost items from two mutually distant reali-
accumulated is also in part dependent ties and ask the subject to "draw a
THE ASSOCIATIVE BASIS OF THE CREATIVE PROCESS 227
spark from their juxtaposition." To at least present) form, the test items
state it more usefully, we must provide consist of sets of three words drawn
stimulus elements from mutually re- from mutally remote associative clus-
mote associative clusters and have the ter. One example might be:
subject find a criteria-meeting medi- Example 1: rat blue cottage
ating link which combines them. A
first problem concerns the type of ma- The subject is required to find a fourth
terial of which the stimulus item should word which could serve as a specific
be composed. If the test is to be ap- kind of associative connective link be-
propriate for all fields of creative en- tween these disparate words. The an-
deavor, the material must either be swer to Example 1 is "cheese."
nonsensical so as to avoid bias favoring "Cheese" is a word which is present in
any specific means of creative expres- the word pairs "rat-cheese," "blue-
sion, or it must be so common in so- cheese," and "cottage-cheese." The
ciety that familiarity could be assumed subject is presented with several ex-
to be high across fields of interest. The amples so that he has an adequate op-
problems involved in constructing the portunity to achieve the specific set
nonsense materials so as to avoid favor- necessary for the task.
ing any interest groups soon proved to Example 2: railroad girl class
be apparently insurmountable. This Example 3: surprise line birthday
left us searching for materials with Example 4: wheel electric high
which most individuals in the culture Example 5: out dog cat 2
could claim acquaintance; this, in turn, (None of these examples is a test item
brought us to verbal materials. from any form of the actual test.) The
While it may be true that certain two college level forms of the test (one
occupational groups have extensive ex- coauthored by Sharon Halpern and the
perience in dealing with words, there other by Martha T. Mednick) have 30
are some verbal associative habits that items each; the subject is allowed 40
could reasonably be assumed to be fa- minutes; his score is the number right.
miliar to almost all individuals that The test, called the Remote Associ-
have been brought up in this (USA) ates Test (RAT), has some interesting
culture. Among such habits are the correlations with other measures.
associative bonds between words like Comparisons with Criteria. A study
"ham and eggs," "bed-bug," "pool-hall," was conducted at the College of Archi-
"hound-dog," "whole-wheat," "chorus- tecture, University of California, Berke-
girl," "kill-joy," and "red-hot." These ley, by the writer and Sharon Halpern.
became the materials for the test. Ratings of creativity by faculty mem-
Having decided on the materials, the bers of the College who taught the De-
test almost constructed itself in accord- sign courses were correlated with RAT
ance with the definition. Several words scores. These ratings form an unusu-
from mutually distant associative clus- ally excellent criterion of creative per-
ters must be presented to the subject; formance since the raters had been
his task must be to provide mediating advising and evaluating the students in
links between them. Further, (a factor the creation of new designs and models
of extreme importance), the mediating of structures. They had been working
link must be strictly associative rather with these students for at least a year
than being of a sort that follows elabo- 2
Answers to sample RAT items: 2.
rate rules of logic, concept formation, working; 3. party; 4. chair or wire; 5.
or problem solving. In their final (or house.
228 SARNOFF A. MEDNICK
and in many cases two or more. The Correlation with Grades. One of the
ratings and RAT scores correlated sig- present forms of the RAT correlated
nificantly (r = .70, df -19, p< .01) negatively with the first-two-year grade
In this study an early form of the RAT point averages of a group of under-
was used. graduates at a large Eastern technology
The RAT was administered to a college, (r = -.27, N = 74, p < .05).
group of first year psychology graduate This same correlation was obtained
students at the University of Michigan with the summer grades of a smaller
whose native language was American group, N = 34, of summer students at
English (JV = 35). Faculty research a large Eastern liberal arts college (not
supervisors (who had been directing statistically significant in this case).
the independent research efforts of the In a study by Miller (1960) it was
students), rated the eight highest and found that high RAT scorers tended to
eight lowest RAT scorers either "high" get higher grades from teachers rated
or "low" in research creativity (no as flexible than from teachers rated as
middle category allowed). Research dogmatic. Low RAT scorers received
creativity was defined as being demon- higher grades from teachers rated dog-
strated if the student developed new matic than from teachers rated as
research methods and/or pulled to- flexible.
gether disparate theory or research Correlation with Social Attitudes
areas in useful and original ways. Of and Occupational Interests. It is clear
the 16 research supervisors, one felt that creative individuals must have
that he had not had enough contact access to improbable associative re-
with his student to make the judgment. sponses. Kowalski (1960) hypothe-
His student was a low RAT scorer. sized that this is a general tendency
Of the eight high RAT scorers, six which also manifests itself in their atti-
were rated high on research creativity tudes and interests. She tested and
and two were rated low; of the seven interviewed 15 high RAT scoring and
low RAT scorers, only one was rated 15 low RAT scoring undergraduate
high, the other six being rated low. women. The two groups had radically
By Fisher's exact test the probability differing views on sexual morality and
of these events occurring by chance is women's rights. The views of the high
less than .05. Miller Analogies Test creatives were more atypical and
(MAT) scores were available for these "liberal" (£7 = 37, / > < .001). On
students. Of the seven high MAT the Strong Vocational Interest Blank,
scorers, three were rated high on re- Metis' Form (SVIB), the high creative
search creativity; of the eight low group showed "significantly higher in-
MAT scorers, four were rated high in terest on the artist (p < .05), psycholo-
research creativity. gist (/> < .005), physician (p < .025),
Reliability. The Spearman-Brown mathematician (p < .025), and author-
reliability of the RAT was .92 in one journalist (p < .05) keys. The low
sample (289 women, almost all the creative group showed higher interest
students at an Eastern women's college, on the farmer (p < .05), math-physical
tested as part of a project under the di- science high school teacher (p < .05),
rection of Theodore Newcomb) and office man (p < .05), and pharmacist
.91 in another (215 men tested at the (p < .01) keys" (p. 19). (These are
University of Michigan as part of a the probability values of obtained chi
project under the direction of Warren squares.) The only one of these keys
T, Norman), related to ACE scores was that of
THE ASSOCIATIVE BASIS OF THE CREATIVE PROCESS 229
duced the probability of the remote as- their most probable associate. Asso-
sociation "switch-weight." ciative probabilities were obtained from
Another feasible experimental ap- the Russell and Jenkins (1954) and
proach would make use of the RAT Deese (1960) norms. If the improba-
item as a creative task and test the in- ble response was satisfying a need,
fluence upon it of certain variables. the probability of noun-choice should
For example, the words of which an increase over the 160 trials. It did
item is composed may be presented at significantly in the high RAT experi-
varying rates to test the massed trials mental group; the low RAT experi-
hypothesis. In addition, various pre- mental group showed a decrease. The
training conditions may be evaluated in high and low RAT control groups
terms of their effectiveness in increasing showed no reliable change.
or decreasing RAT performance. Some of the positions which have
Another possible experimental ap- been taken in this paper are assump-
proach would entail separating out high tions and not deductions. As more
and low RAT scorers and observing data are gathered some of these as-
the differential effect of certain vari- sumptions will assume the status of
ables upon their behavior. In an ex- facts, some will be revised. For ex-
periment just completed Houston and ample, the opening paragraph suggests
Mednick (in press) postulated that an that the paper is not meant to apply
important motive impelling the be- only to one field of creative endeavor
havior of the creative individual was a but attempts to delineate processes
need for improbable associative stimu- that underlie all creative thought. This
lation. It was reasoned that if such may require modification. The expla-
stimulation were supplied, it would nation may fit the process of scientific
tend to satisfy this need and be rein- discovery and not be appropriate to
forcing. Further, if such stimulation discussions of painting or music. For
regularly followed a given response the the present (paradoxically enough), the
high creative individual should tend to more encompassing assumptions seem
learn that response. Thirty high and more parsimonious. It may eventually
30 low RAT scorers were asked to turn out (as is hinted at in the body of
read aloud only one of two typed the paper) that the differences between
words on a 3 x 5 card. Excepting the fields are more determined by dif-
buffer items and including 40 pairs ferences in suitability of the three
aimed at gauging the free operant means of achieving contiguity, i.e.,
level of noun-choice, there were 160 serendipity, similarity, and mediation.
pairs of words, each pair consisting of a
noun and a nonnoun (verb, adjective, SUMMARY
adverb, etc.). If a subject in the experi-
mental group (IS high RAT subjects, An associative theory of creative
15 low RAT subjects) responded with thinking has been outlined. Differences
the noun member of a pair, the experi- between high creatives and low crea-
menter responded with an improbable tives have been predicted along speci-
association; if the subject chose the fied dimensions. Predictions have been
nonnoun, the experimenter responded made regarding the effect on the crea-
with the most probable association. In tive process of some experimentally
the control group (15 high RAT, 15 manipulable variables.
low RAT) both the nouns and the The associative definition of the crea-
nonnouns were invariably followed by tive process has taken the operational
232 SARNOFF A. MEDNICK