Creating a Vision of Creativity: The First 25 Years

Psychology of Aesthetics, Creativity, and the Arts
Copyright 2006 by the American Psychological Association
2006, Vol. S, No. 1, 2–12
1931-3896/06/$12.00
DOI: 10.1037/1931-3896.S.1.2
Creating a Vision of Creativity: The First 25 Years
Robert J. Sternberg
Yale University
This article describes three stages of my attempts to understand, measure, and develop creative thinking.
The first stage explored creative intelligence. The second investigated a theory of creativity, the
investment theory. The third proposed a theory of creative leadership. Together, these three stages
comprise the development of my thought on creativity—its nature, measurement, and development.
Keywords: creativity, abilities, motivation, personality, intelligence
For roughly 25 years, I have been trying to understand creativity
of the components of analytical intelligence, I tried to understand
and its various aspects. Because this article is based on an Amer-
as well other kinds of intelligence, in particular, creative and
ican Psychological Association (APA) Division 10 Arnheim
practical thinking. I proposed what I referred to as a “triarchic”
Award address for career achievement studying creativity, this
theory of intelligence (Sternberg, 1984, 1985). One of the three
seems like a good time to review what I think I have learned in 25
kinds of intelligence that I concerned myself with, the topic of
years.
discussion here, is creative intelligence. I then expanded my think-
My goal has been to create some kind of vision of creativity:
ing to a theory of “successful” intelligence, which emphasized the
What is it, how can it be measured, how can it be developed? Over
importance not only of one’s pattern of analytical, creative, and
the course of the years, my attempts to understand creativity have
practical intelligence, but also of capitalizing on strengths and
gone roughly through three stages. The stages have not been
compensating for or correcting weaknesses (Sternberg, 1997a,
wholly sequential. Sometimes I would proceed to a next stage,
1999a).
only later to go back to an earlier one. My goal always has been to
broaden and deepen my, and, I hope, others’ understanding of
creativity.
The Move to Studying Creative Intelligence
What has changed over the years are not the “answers.” I have
not found anything earlier that I later retracted or ceased to believe.
Intelligence tests contain a range of problems, some of them
Rather, what have changed are the questions. As time has gone on,
more novel than others. In some of our componential work, we
the questions that have seemed important to ask have changed as
have shown that when one goes beyond the range of unconven-
my research interests have developed.
tionality of the conventional tests of intelligence, one starts to tap
sources of individual differences measured little or not at all by the
tests (Sternberg, 1985). According to the theory of successful
Stage 1: Creative Intelligence
intelligence, (creative) intelligence is particularly well measured
Early Studies of Intelligence
by problems assessing how well an individual can cope with
relative novelty. Thus, it is important to include in a battery of tests
My earliest work was on intelligence (e.g., Sternberg, 1977,
problems that are relatively novel in nature.
1979). I was trying to understand the mental representations and
In work with convergent problems, we presented 80 individuals
processes involved in solving problems on intelligence tests. But
with novel kinds of reasoning problems that had a single best
after some number of years, I came to the conclusion that this
answer (Sternberg, 1982; Tetewsky & Sternberg, 1986). For ex-
approach was narrow, because the kinds of items used on these
ample, they might be told that some objects are green and others
tests covered only a narrow range. My original “componential
blue; but still other objects might be grue, meaning green until the
subtheory” of intelligence (Sternberg, 1980) came to seem too
year 2000 and blue thereafter, or bleen, meaning blue until the year
narrow. Instead of trying to understand intelligence only in terms
2000 and green thereafter. Or they might be told of four kinds of
people on the planet Kyron, blens, who are born young and die
young; kwefs, who are born old and die old; balts, who are born
Robert J. Sternberg, Yale University, New Haven, Connecticut.
young and die old; and prosses, who are born old and die young.
Preparation of this article was supported by a government grant under
Their task was to predict future states from past states, given
the Javits Act Program (Grant No. R206R000001) as administered by the
incomplete information. In another set of studies, 60 people were
U.S. Institute of Educational Sciences, U.S. Department of Education.
given more conventional kinds of inductive reasoning problems,
Such support does not imply acceptance or endorsement of the ideas
such as analogies, series completions, and classifications, but were
presented in this article.
told to solve them. The problems had premises preceding them that
Correspondence concerning this article should be addressed to Robert J.
were either conventional (dancers wear shoes) or novel (dancers
Sternberg who is now at the Office of the Dean of Arts and Sciences, 3rd
floor, Ballou Hall, Tufts University, Medford, Massachusetts 02155. E-
eat shoes). The participants had to solve the problems as though
mail: robert.sternberg@tufts.edu
the counterfactuals were true (Sternberg & Gastel, 1989a, 1989b).
2

---

CREATING A VISION OF CREATIVITY
3
In these studies, we found that correlations with conventional
Creative-Quantitative:
Novel number operations. Students
kinds of tests depended on how novel or nonentrenched the con-
are presented with rules for novel
ventional tests were. The more novel the items, the higher were the
number operations, for example,
correlations of our tests with scores on successively more novel
“flix,” which involves numerical
conventional tests. Thus, the components isolated for relatively
manipulations that differ as a func-
novel items would tend to correlate more highly with more unusual
tion of whether the first of two op-
tests of fluid abilities (e.g., that of Cattell & Cattell, 1973) than
erands is greater than, equal to, or
with tests of crystallized abilities. We also found that when re-
less than the second. Participants
sponse times on the relatively novel problems were componen-
have to use the novel number oper-
tially analyzed, some components better measured the creative
ations to solve presented math
aspect of intelligence than did others. For example, in the “grue-
problems.
bleen” task mentioned above, the information-processing compo-
nent requiring people to switch from conventional green-blue
Creative-Figural:
In each item, participants are first
thinking to grue-bleen thinking and then back to green-blue think-
presented with a figural series that
ing again was a particularly good measure of the ability to cope
involves one or more transforma-
with novelty.
tions; they then have to apply the
rule of the series to a new figure with
a different appearance, and complete
The Rainbow Project
the new series.
More recently, we have sought to measure creative intelligence
in the context of a test designed to supplement the SAT. This test
Ability scores were computed by combining the responses to
emanated from the so-called Rainbow Project (Henry, Sternberg,
the subscales, using item response theory to create three final
& Grigorenko, 2004; Sternberg & The Rainbow Project Collabo-
scales representing analytical, creative, and practical skills
rators, 2005; Sternberg, The Rainbow Project Collaborators, &
(STAT
, STAT
, and STAT
). The Cronbach
Analytical
Creative
Practical
University of Michigan Business School Project Collaborators,
alpha estimates of reliability were satisfactory but not high (.67,
2004), an attempt to explore how to supplement the SAT in college
.72, and .56 for the analytical, creative, and practical subtests,
admissions with creative and practical tests in addition to the more
respectively), in part because within each subtest, the content
analytical SAT. This work thus represents a recent return to Stage
domains (verbal, quantitative, and figural) are very different from
1—an attempt to measure creative intelligence in the context of
each other.
our most recent understandings.
Creative performance tests had a different character. A series of
tasks was created that required open-ended, creative responses. For
The Participants
each of the tasks, participants were given a choice of topic or
stimuli. Although these different topics or stimuli varied in terms
A battery of tests was administered in the spring and summer of
of their difficulty for inventing creative stories and captions, these
2001 to eight 4-year colleges and five community colleges. The
differences are accounted for in the derivation of Item Response
participants were 793 students predominantly in their first year of
Therapy (IRT) ability estimates using the Many-Facets Rasch
college.
model (Linacre, 1989) of the FACETS computer program (Lina-
cre, 1998).
The Tests
Cartoons.
Participants were given five cartoons purchased
from the archives of the New Yorker, but with the captions re-
The Sternberg Triarchic Abilities Test (STAT; Sternberg, 1993)
moved. The participants’ task was to choose three cartoons and to
was a means of capturing analytical, practical, and creative skills
provide a caption for each cartoon. Two trained judges rated all the
using multiple-choice questions (Sternberg & Clinkenbeard, 1995;
cartoons for cleverness, humor, originality, and task appropriate-
Sternberg, Ferrari, Clinkenbeard, & Grigorenko, 1996). Level H of
ness on 5-point scales. A combined creativity score was formed by
the test (Sternberg, 1993) was designed to measure cognitive skills
deriving ability estimates based on a composite of all ratings
among secondary school and college students, and was used in
except for task appropriateness, which theoretically is not a mea-
this study. Each multiple-choice item in the STAT had four
sure of creativity per se. The IRT reliability for the composite was
different response options, from which the correct response
very good.
could be selected. There were three analytical, three creative,
Written stories.
Participants were asked to write two stories,
and three practical subtests. The creative multiple-choice STAT
spending about 15 minutes on each, choosing from the following
subtests were:
titles: “A Fifth Chance,” “2983,” “Beyond the Edge,” “The Octo-
Creative-Verbal:
Novel analogies. Students are pre-
pus’s Sneakers,” “It’s Moving Backwards,” and “Not Enough
sented with verbal analogies pre-
Time” (Lubart & Sternberg, 1995; Sternberg & Lubart, 1995). A
ceded by counterfactual premises
team of six judges was trained to rate the stories. Each judge rated
(e.g., money falls off trees). They
the stories for originality, complexity, emotional evocativeness,
have to solve the analogies as though
and descriptiveness on 5-point scales. Because the reliability based
the counterfactual premises were
on the total score for each story was satisfactory, for purposes of
true.
efficiency, 64.7% of the stories were rated by one of the six judges.

---

4
STERNBERG
The IRT reliability indices for the composite person ability esti-
included the items analyzed above in the analytic, creative, and
mates for the written stories were very good.
practical abilities. We used the Mplus computer program to ana-
Oral stories.
Participants were presented with five sheets of
lyze the hierarchical multiple regressions, and for specification
paper, each containing a set of 11 to 13 images linked by a
purposes for the structural equation models we constrained the six
common theme (e.g., keys, money, travel, animals playing musical
lowest correlations in the correlation matrix to zero (not doing so
instruments, and humans playing musical instruments). After
would have left the model with negative degrees of freedom). The
choosing one of the pages, the participant was given 15 minutes to
creativity measures in these hierarchical regressions are separated
formulate a short story and dictate it into a cassette recorder. The
from their latent variable because, as noted earlier, these items did
dictation period for each story was not to be more than 5 minutes
not include enough common variance to justify using a latent
long. The process was then repeated with another sheet of images
variable in further analysis.
so that each participant dictated a total of two oral stories. Six
The hierarchical regressions show that measures representing
judges were trained to rate the stories. As with the written stories,
each of the triarchic measures of intelligence significantly pre-
each judge rated the stories for originality, complexity, emotional
dicted success in college as represented by college grade-point
evocativeness, and descriptiveness on 5-point scales. Because in-
average. In the first stage we began with the practical intelligence
terrater reliability based on the total score for each story was satis-
measures. Both measures of practical intelligence significantly
factory, for purposes of efficiency, 48.4% of the stories were rated by
contributed to predicting college GPA, and together predicted
one of the six judges. The IRT reliability indices for the composite
7.4% of the variance in GPA. In the second stage of the regression
person ability estimates for the oral stories were very good.
we added the creativity measures. Both practical intelligence mea-
School performance was measured using cumulative grade point
sures remain statistically significant, and the oral stories and cre-
average as obtained from school transcripts.
ative STAT contribute significantly to the model. Together, the
All materials were administered either in paper-and-pencil for-
added creativity items double the amount of variance contributed
mat (N
325) or on the computer via the World Wide Web (N
by the practical measures (R2
15.3%). Finally, in the last step we
468). Participants were either tested individually or in small
added the analytic STAT measure, which also contributes signif-
groups. During the oral stories section, participants who were
icantly to the regression model and adds 1% to the GPA variance
tested in the group situation either wore headphones or were
predicted by all measures (16.3%). By this last stage, at least one
directed into a separate room so as not to disturb the other partic-
measure from each of the aspects of intelligence was represented
ipants during the oral dictation of the stories.
as significant contributors to the model: the practical performance
There were two discrete sessions, conducted one directly after
latent variable (including all three tacit-knowledge inventories),
the other, for each participant. The order of test administration was
the oral stories, and the creative and analytic STAT measures.
the same for all participants. No strict time limits were set for
completing the tests, although the instructors were given rough
Group-Differences Data
guidelines of about 70 minutes per session. The time taken to
complete the battery of tests ranged from 2 to 4 hours.
Although one important goal of the present study was to predict
As a result of the lengthy nature of the complete battery of
success in college, another important goal involved developing
assessments, participants were administered parts of the battery
measures that might reduce gender, racial, and ethnic group dif-
using an intentional incomplete overlapping design, as described in
ferences in the mean levels for the predictor variables. There are a
McArdle and Hamagami (1992; also McArdle, 1994). The partic-
number of ways one can test for group differences in these mea-
ipants were randomly assigned to the test sections they were to
sures, each involving a test of the size of the effect of race. We
complete. All missing data in the sample were managed using the
chose two different measures of effect size: omega square ( 2),
full-information maximum likelihood (FIML) technique (see
and Cohen’s d.
McArdle, 1994; Wothke, 2000).
To test our success in minimizing group differences with the
new measures, we first considered the omega square coefficients.
The Validity Data
This procedure involves conducting a series of one-way analyses
of variance (ANOVA) considering differences in mean perfor-
An exploratory factor analysis with Varimax rotation was con-
mance levels among different groups. The omega-squared coeffi-
ducted to explore the factor structure underlying the measures in
cient indicates the amount of variance in the variables that is
this study. Three factors were extracted with eigenvalues greater
accounted for by the group differences.
than 1, and these accounted for 59.2% of the variation between the
We first tested for differences between the six ethnic and racial
measures.
groups reported, including White, Asian, Pacific Islander, Latino,
The results suggest that, consistent with the analyses reported
Black, and Native American. Although most of the measures
above, evidence for a unidimensional latent creativity factor is
indicated a statistically significant F statistic, consider the omega-
mixed. It would seem that, in this sample, the common method-
squared statistic, which is not vulnerable to the size of the sample.
ological factor might have overwhelmed the unique creative, prac-
All of the individual measures had
2
.03, suggesting that, at
tical, and analytic contribution offered by the different STAT
most, 3% of the variance in each individual measure could be
subtests. However, as will be demonstrated below, in terms of
accounted for by racial and ethnic difference. For the practical
incremental validity, the three STAT variables each contributed
latent variable, the number becomes somewhat increased,
2
uniquely to the prediction of college grade point average (GPA).
.05.
In order to test the incremental validity provided by our mea-
When considering the differences between men and women,
sures, a series of hierarchical regressions was conducted that
most of the tests resulted in an omega-squared statistic of zero

---

CREATING A VISION OF CREATIVITY
5
(including some with statistically significant F statistics). How-
countries who were identified by their schools as gifted by any
ever, for one of the practical performance measures, the Common
standard whatsoever. Children were selected for a summer pro-
Sense measure, 5% of the variance could be explained by the
gram in (college-level) psychology if they fell into one of five
participant’s gender. As will be shown in the Cohen’s d statistics,
ability groupings: high analytical, high creative, high practical,
this is because women outperformed men on this measure.
high balanced (high in all three abilities), or low balanced (low in
The test of effect sizes using the Cohen’s d statistic allows one
all three abilities). Students who came to Yale were then divided
to consider more specifically a standardized representation of
into four instructional groups. Students in all four instructional
specific group differences. For the test of ethnic group differences,
groups used the same introductory-psychology textbook (a prelim-
each entry represented how far away from the mean for Whites
inary version of Sternberg [1995a]) and listened to the same
each group performs in terms of standard deviations. For the test
psychology lectures. What differed among them was the type of
of gender differences, the entries represented how far away from
afternoon discussion section to which they were assigned. They
the mean for men that the women perform in terms of standard
were assigned to an instructional condition that emphasized either
deviations.
memory, analytical, creative, or practical instruction. For example,
These results illuminate findings that go beyond the omega-
in the memory condition, they might be asked to describe the main
squared statistic. First, the Black and Latino students seem to be
tenets of a major theory of depression. In the analytical condition,
negatively affected by the group differences. However, when com-
they might be asked to compare and contrast two theories of
pared to reported group differences on other predictors of college
depression. In the creative condition, they might be asked to
success, like the SAT and ACT, these scores show a marked
formulate their own theory of depression. In the practical condi-
improvement. For example, a study conducted in 1998 showed that
tion, they might be asked how they could use what they had
Blacks and Latinos consistently score from between half to a full
learned about depression to help a friend who was depressed.
standard deviation away on standardized tests:
.83,
.92, and
Students in all four instructional conditions were evaluated in
.98 for Blacks on the SAT-Verbal, SAT-Math, and ACT-
terms of their performance on homework, a midterm exam, a final
Composite, respectively; and
.63,
.61, and
.60 for Latinos on
exam, and an independent project. Each type of work was evalu-
the SAT-Verbal, SAT-Math, and ACT-Composite, respectively
ated for memory, analytical, creative, and practical quality. Thus,
(Camara & Schmidt, 1999).
all students were evaluated in exactly the same way.
Second, to the extent that gender differences exist in our mea-
Our results suggested the utility of the theory of successful
sures, they are most visible among the practical performance
intelligence. This utility showed itself in several ways.
measures where women tend to outperform the men. For the
First, we observed when the students arrived at Yale that the
Common Sense measure, this results in scores as much as half a
students in the high creative and high practical groups were much
standard deviation above men. Nevertheless, on measures like the
more diverse in terms of racial, ethnic, socioeconomic, and edu-
STAT-Analytic, women still perform below men, but this is only
cational backgrounds than were the students in the high-analytical
by about one fifth a standard deviation.
group, suggesting that correlations of measured intelligence with
Overall, although the group differences are not perfectly re-
status variables such as these may be reduced by using a broader
duced, these findings suggest that measures may be designed to
conception of intelligence. Thus, the kinds of students identified as
reduce ethnic and racial group differences on standardized tests,
strong differed in terms of populations from which they were
particularly for historically disadvantaged groups like Black and
drawn in comparison with students identified as strong solely by
Latino students. These findings have important implications for
analytical measures. More importantly, just by expanding the
reducing adverse impact wherever these tests might be used, such
range of abilities measured, we discovered intellectual strengths
as college admissions.
that might not have been apparent through a conventional test.
Second, we found that all three ability tests—analytical, cre-
Instructional Studies
ative, and practical—significantly predicted course performance.
When multiple-regression analysis was used, at least two of these
Another way of exploring creativity is to look at how one can
ability measures contributed significantly to the prediction of each
develop it and at the same time enhance school achievement. In a
of the measures of achievement. Perhaps as a reflection of the
first set of studies, we explored the question of whether conven-
difficulty of deemphasizing the analytical way of teaching, one of
tional education in school systematically discriminates against
the significant predictors was always the analytical score. (How-
children with creative and practical strengths (Sternberg &
ever, in a replication of our study with low-income African-
Clinkenbeard, 1995; Sternberg, Ferrari, Clinkenbeard, & Grigor-
American students from New York, Deborah Coates of the City
enko, 1996; Sternberg, Grigorenko, Ferrari, & Clinkenbeard,
University of New York found a different pattern of results. Her
1999). Motivating this work was the belief that the systems in most
data indicated that the practical tests were better predictors of
schools strongly tend to favor children with strengths in memory
course performance than were the analytical measures, suggesting
and analytical abilities. However, schools can be unbalanced in
that what ability test predicts what criterion depends on population
other directions as well. One school Elena Grigorenko and I visited
as well as mode of teaching.)
in Russia in 2000 placed a heavy emphasis upon the development
Third and most importantly, there was an aptitude-treatment
of creative abilities—much more so than on the development of
interaction whereby students who were placed in instructional
analytical and practical abilities.
conditions that better matched their pattern of abilities outper-
We used the Sternberg Triarchic Abilities Test, as described
formed students who were mismatched. In other words, when
above, in some of our instructional work. The test was adminis-
students are taught in a way that fits how they think, they do better
tered to 326 children around the United States and in some other
in school. Children with creative and practical abilities, who are

---

6
STERNBERG
almost never taught or assessed in a way that matches their pattern
sell high” in the realm of ideas (see also Rubenson & Runco, 1992,
of abilities, may be at a disadvantage in course after course, year
for use of concepts from economic theory). Buying low means
after year.
pursuing ideas that are unknown or out of favor but that have
A follow-up study (Sternberg, Torff, & Grigorenko, 1998a,
growth potential. Often, when these ideas are first presented, they
1998b) examined learning of social studies and science by third-
encounter resistance. The creative individual persists in the face of
graders and eighth-graders. The 225 third-graders were students in
this resistance, and eventually sells high, moving on to the next
a very low-income neighborhood in Raleigh, North Carolina. The
new, or unpopular idea.
142 eighth-graders were students who were largely middle to
Research within the investment framework has yielded support
upper-middle class studying in Baltimore, Maryland, and Fresno,
for this model (Lubart & Sternberg, 1995). This research, which
California. In this study, students were assigned to one of three
later served in part as the basis for the Rainbow Project work on
instructional conditions. In the first condition, they were taught the
creativity discussed earlier, has used tasks such as (a) writing
course that basically they would have learned had there been no
short-stories using unusual titles (e.g., “The Octopus’ Sneakers”),
intervention. The emphasis in the course was on memory. In a
(b) drawing pictures with unusual themes (e.g., the earth from an
second condition, students were taught in a way that emphasized
insect’s point of view), (c) devising creative advertisements for
critical (analytical) thinking. In the third condition, they were
boring products (e.g., cufflinks), and (d) solving unusual scientific
taught in a way that emphasized analytical, creative, and practical
problems (e.g., how we could tell if someone had been on the
thinking. All students’ performance was assessed for memory
moon within the past month?). This research showed creative
learning (through multiple-choice assessments) as well as for
performance to be moderately domain-specific and to be predicted
analytical, creative, and practical learning (through performance
by a combination of certain resources, as described below.
assessments).
According to the investment theory, creativity requires a con-
As expected, students in the successful-intelligence (analytical,
fluence of six distinct but interrelated resources: intellectual abil-
creative, practical) condition outperformed the other students in
ities, knowledge, styles of thinking, personality, motivation, and
terms of the performance assessments. One could argue that this
environment. Although levels of these resources are sources of
result merely reflected the way they were taught. Nevertheless, the
individual differences, often the decision to use the resources is the
result suggested that teaching for these kinds of thinking suc-
more important source of individual differences.
ceeded. More important, however, was the result that children in
the successful-intelligence condition outperformed the other chil-
Facets of Creativity
dren even on the multiple-choice memory tests. In other words, to
the extent that one’s goal is just to maximize children’s memory
Intellectual Abilities
for information, teaching for successful intelligence is still supe-
rior. It enables children to capitalize on their strengths and to
Three intellectual skills are particularly important (Sternberg,
correct or to compensate for their weaknesses, and it allows
1985): (a) the synthetic ability to see problems in new ways and to
children to encode material in a variety of interesting ways.
escape the bounds of conventional thinking; (b) the analytic ability
Grigorenko and her colleagues extended these results to reading
to recognize which of one’s ideas are worth pursuing and which
curricula at the middle-school and the high-school level (Grigor-
are not; and (c) the practical-contextual ability to know how to
enko, Jarvin, & Sternberg, 2002). In a study of 871 middle school
persuade others of—to sell other people on—the value of one’s
students and 432 high school students, researchers taught reading
ideas. The confluence of these three abilities is also important.
either triarchically or through the regular curriculum. At the mid-
Analytic ability used in the absence of the other two abilities
dle school level, reading was taught explicitly. At the high school
results in powerful critical, but not creative thinking. Synthetic
level, reading was infused into instruction in mathematics, physi-
ability in the absence of the other two abilities results in new ideas
cal sciences, social sciences, English, history, foreign languages,
that are not subjected to the scrutiny required to make them work.
and the arts. In all settings, students who were taught triarchially
And practical-contextual ability in the absence of the other two
substantially outperformed students who were taught in standard
may result in the transmittal of ideas not because the ideas are
ways.
good, but rather because the ideas have been well and powerfully
Thus, the results of three sets of studies suggest that teaching for
presented. To be creative, one must first decide to generate new
creativity (as well as other kinds of thinking) can improve school
ideas, analyze these ideas, and sell the ideas to others. The studies
achievement. Creative students do not profit as much as some
described above address the issue of intelligence and creativity.
other students from standard kinds of teaching. Moreover, the
results suggest that the theory can make a difference not only in
Knowledge
laboratory tests, but in school classrooms and even the everyday
life of adults as well.
Concerning knowledge, on the one hand, one needs to know
enough about a field to move it forward. One can’t move beyond
Stage 2: The Investment Theory of Creativity
where a field is if one doesn’t know where it is. On the other hand,
knowledge about a field can result in a closed and entrenched
Terms of the Theory
perspective, resulting in a person’s not moving beyond the way in
which he or she has seen problems in the past (Frensch & Stern-
With Todd Lubart, I proposed an “investment” theory of cre-
berg, 1989). Thus, one needs to decide to use one’s past knowl-
ativity (Sternberg & Lubart, 1991, 1995). According to this theory,
edge, but also decide not to let the knowledge become a hindrance
creative people are ones who are willing and able to “buy low and
rather than a help.

---

CREATING A VISION OF CREATIVITY
7
Thinking Styles
ests and defies the crowd. The crowd does not maliciously or
willfully reject creative notions. Rather, it does not realize, and
With regard to thinking styles, a legislative style is particularly
often does not want to realize, that the proposed idea represents a
important for creativity (Sternberg, 1988, 1997b; Zhang & Stern-
valid and advanced way of thinking. Society generally perceives
berg, 2005, 2006), that is, a preference for thinking and a decision
opposition to the status quo as annoying, offensive, and as reason
to think in new ways. This preference needs to be distinguished
enough to ignore innovative ideas.
from the ability to think creatively: Someone may like to think
Evidence abounds that creative ideas are often rejected (Stern-
along new lines, but not think well, or vice versa. It also helps, to
berg & Lubart, 1995). Initial reviews of major works of literature
become a major creative thinker, if one is able to think globally as
and art are often negative. Toni Morrison’s Tar Baby received
well as locally, distinguishing the forest from the trees and thereby
negative reviews when it was first published, as did Sylvia Plath’s
recognizing which questions are important and which ones are not.
The Bell Jar. The first exhibition in Munich of the work of
Norwegian painter Edvard Munch opened and closed the same day
Personality
because of the strong negative response from the critics. Some of
the greatest scientific papers have been rejected not just by one, but
Numerous research investigations (summarized in Lubart, 1994,
by several journals before being published. For example, John
and Sternberg & Lubart, 1991, 1995) have supported the impor-
Garcia, a distinguished biopsychologist, was immediately de-
tance of certain personality attributes for creative functioning.
nounced when he first proposed that a form of learning called
These attributes include, but are not limited to, willingness to
classical conditioning could be produced in a single trial of learn-
overcome obstacles, willingness to take sensible risks, willingness
ing (Garcia & Koelling, 1966).
to tolerate ambiguity, and self-efficacy. In particular, buying low
From the investment view, then, the creative person buys low by
and selling high typically means defying the crowd, so that one has
presenting a unique idea and then attempting to convince other
to be willing to stand up to conventions if one wants to think and
people of its value. After convincing others that the idea is valu-
act in creative ways. Note that none of these attributes are fixed.
able, which increases the perceived value of the investment, the
One can decide to overcome obstacles, take sensible risks, and so
creative person sells high by leaving the idea to others and moving
forth.
on to another idea. People typically want others to love their ideas,
but immediate universal applause for an idea usually indicates that
Motivation
it is not particularly creative.
Intrinsic, task-focused motivation is also essential to creativity.
Creativity is as much a decision about and an attitude toward life
The research of Amabile (1983) and others has shown the impor-
as it is a matter of ability. Creativity is often obvious in young
tance of such motivation for creative work, and has suggested that
children, but it may be harder to find in older children and adults
people rarely do truly creative work in an area unless they really
because their creative potential has been suppressed by a society
love what they are doing and focus on the work rather than the
that encourages intellectual conformity. Yet, anyone can decide for
potential rewards. Motivation is not something inherent in a per-
creativity.
son: One decides to be motivated by one thing or another.
In work with divergent reasoning problems having no one best
answer, we asked 63 people to create various kinds of products
(Lubart & Sternberg, 1995; Sternberg & Lubart, 1991, 1995, 1996)
Environment
where an infinite variety of responses were possible. Individuals
Finally, one needs an environment that is supportive and re-
were asked to create products in the realms of writing, art, adver-
warding of creative ideas. One could have all of the internal
tising, and science. In writing, they would be asked to write very
resources needed in order to think creatively, but without some
short stories for which we would give them a choice of titles, such
environmental support (such as a forum for proposing those ideas),
as “Beyond the Edge” or “The Octopus’s Sneakers.” In art, the
the creativity that a person has within him or her might never be
participants were asked to produce art compositions with titles
displayed.
such as “The Beginning of Time” or “Earth from an Insect’s Point
of View.” In advertising, they were asked to produce advertise-
Confluence
ments for products such as a brand of bow tie or a brand of
doorknob. In science, they were asked to solve problems such as
Concerning the confluence of components, creativity is hypoth-
one asking them how people might detect extraterrestrial aliens
esized to involve more than a simple sum of a person’s level on
among us who are seeking to escape detection. Participants created
each component. First, there may be thresholds for some compo-
two products in each domain.
nents (e.g., knowledge) below which creativity is not possible
We found, first, that creativity comprises the components pro-
regardless of the levels on other components. Second, partial
posed by our investment model of creativity: intelligence, knowl-
compensation may occur in which strength on one component
edge, thinking styles, personality, and motivation. Second, they
(e.g., motivation) counteracts a weakness on another component
found that creativity is relatively although not wholly domain-
(e.g., environment). Third, interactions may also occur between
specific. Correlations of ratings of the creative quality of the
components, such as intelligence and motivation, in which high
products across domains were lower than correlations of ratings
levels on both components could multiplicatively enhance
and generally were at about the .4 level. Thus, there was some
creativity.
degree of relation across domains, at the same time that there was
Creative ideas are both novel and valuable, but they are often
plenty of room for someone to be strong in one or more domains
rejected because the creative innovator stands up to vested inter-
but not in others. Third, we found a range of correlations of

---

8
STERNBERG
measures of creative performance with conventional tests of abil-
• Advance forward incrementation. This type of leadership is an
ities. As was the case for the correlations obtained with convergent
attempt to move an organization forward in the direction it is
problems, correlations were higher to the extent that problems on
already going, but by moving beyond where others are ready for it
the conventional tests were nonentrenched. For example, correla-
to go. The leader moves followers in an accelerated way beyond
tions were higher with fluid than with crystallized ability tests, and
the expected rate of forward progression. Advance forward incre-
correlations were higher, the more novel the fluid test was. These
mentations usually are not successful at the time they are at-
results that tests of creative intelligence have some overlap with
tempted, because followers in fields and organizations are not
conventional tests (e.g., in requiring verbal skills or the ability to
ready to go where the leader wants to lead. Or significant portions
analyze one’s own ideas—Sternberg & Lubart, 1995) but also tap
of them may not wish to go to that point, in which case they form
skills beyond those measured even by relatively novel kinds of
an organized and sometimes successful source of resistance.
items on the conventional tests of intelligence.
• Redirection. This type of leadership is an attempt to redirect
The work on creativity revealed a number of sources of indi-
an organization, field, or product line from where it is headed
vidual and developmental differences.
toward a different direction. Redirective leaders need to match to
environmental circumstances to succeed (Sternberg & Vroom,
1.
To what extent was the thinking of the individual novel
2002). If they do not have the luck to have matching environmental
or nonentrenched?
circumstances, their best intentions may go awry.
• Reconstruction/redirection. This type of creative leadership is
2.
What was the quality of the individual’s thinking?
an attempt to move a field or an organization or a product line back
to where it once was (a reconstruction of the past) so that it may
3.
To what extent did the thinking of the individual meet the
move onward from that point, but in a direction different from the
demands of the task?
one it took from that point onward.
• Reinitiation. This type of leadership is an attempt to move a
Stage 3: Creative Leadership
field, organization, or product line to a different as yet unreached
starting point and then to move from that point. The leaders takes
Types of Creative Leadership
followers from a new starting point in a direction that is different
In my most recent work, I have become interested in the
from that the field, organization, or product line previously has
question of what constitutes creative leadership (Sternberg, 1999b;
pursued.
Sternberg, Kaufman, & Pretz, 2001, 2002, 2003, 2004; Sternberg
• Synthesis. In this type of creative leadership, the creator
& Vroom, 2002). I believe that creative leadership can be of
integrates two ideas that previously were seen as unrelated or even
different types (Sternberg, 1999b; Sternberg, Kaufman, & Pretz,
as opposed. What formerly were viewed as distinct ideas now are
2003). Some of these types accept current ways of doing things,
viewed as related and capable of being unified. Integration is a key
others do not; and still another attempts to integrate different
means by which progress is attained in the sciences. It represents
current practices. Which types are more acceptable depends upon
neither an acceptance nor a rejection of existing paradigms, but
the interaction of the leader with the situation. The types of
rather, a merger of them.
creative leadership are
• Replication. This type of leadership is an attempt to show that
Stories of Creative Leadership
a field or organization is in the right place at the right time. The
leader therefore attempts to maintain it in that place. The leader
Contents of Creativity: Stories
keeps the organization where it is rather than moving it. The view
of the leader is that the organization is where it needs to be. The
Leaders generate stories that appeal in various degrees to their
leader’s role is to keep it there.
followers. These stories, like stories of love (Sternberg, 1998),
• Redefinition. This type of leadership is an attempt to show
attract followers in various degrees. In unsuccessful leaders, they
that a field or organization is in the right place, but not for the
leave followers indifferent, or even repulsed, as in stories of hate
reason(s) that others, including previous leaders, think it is. The
(see Sternberg, 2003). Whether a story works or not, therefore, is
current status of the organization thus is seen from a different point
a contingency dependent upon the leader, the followers, and the
of view. Redefiners often end up taking credit for ideas of others
situation (Ayman, 2004).
because they find a better reason to implement the others’ ideas, or
Stories provide much of the content of creativity in leadership.
say they do.
They are the way in which a leader distinguishes him or herself
• Forward incrementation. This type of leadership is an attempt
and the contribution he or she plans to make. The story a leader
to lead a field or an organization forward in the direction it already
tells is not necessarily constructive, as leaders such as Hitler and
is going. The leader specializes to forward motion. Most leader-
Stalin have shown. Hitler’s story as a savior of the Aryan race
ship is probably forward incrementation. In such leadership, one
resulted in millions of death and much suffering. People must
takes on the helm with the idea of advancing the leadership
understand their leaders’ stories if they are to understand whether
program of whomever one has succeeded. The promise is of
the leader will do good or ill.
progress through continuity. Creativity through forward incremen-
tation is probably the kind that is most easily recognized and
Characteristics of Stories
appreciated as creativity. Because it extends existing notions, it is
seen as creative. Because it does not threaten the assumptions of
Gardner’s view of leadership stories.
Gardner (1995) has sug-
such notions, it is not rejected as useless or even harmful.
gested that successful leaders have a story to tell and a message to

---

CREATING A VISION OF CREATIVITY
9
convey. The story tends to be more effective to the extent that it
it distressing that many people cared so little about the changes in
appeals to what Gardner (1991) refers to as the “unschooled
the stories. They cared more about having a story with resonance
mind,” that is, a mind that, in terms of modern cognitive theory, is
than one that was necessarily “true” in any meaningful sense.
more experiential than rational in its thinking (Sloman, 1996).
Stories also have plots, themes, and characters. For example, a
Stories need to address both individuals’ own identities and those
common story now for political leaders is the warrior chieftain
of the group or groups to which they belong. A story is more likely
who will fight terrorists. The plot is the battle against terrorists.
to succeed if it is central to what the leader actually does in his or
Themes give stories meaning. They help people understand why
her action, if the story can be unfolded over a long period of time,
the story is important and what script it will follow. One theme is
and if it can be stated in a time of relative calm. In times of crisis,
that the leader must constantly prepare his followers to combat the
according to Gardner, stories need to be simplified.
terrorists; another is that followers must give up some of their
Stories may be inclusionary or exclusionary. Inclusionary lead-
liberty to enable the leader to fight the terrorists in an effective
ers try to ensure that all of the followers for whom they are
way. Vladimir Putin, for example, announced in September 2004,
responsible somehow are made to feel inside the fold. Exclusion-
a major reorganization of the Russian government to enable effec-
ary leaders do not include everyone and in extreme cases, such as
tive mobilization against terrorists. The reorganization concen-
Hitler or Stalin, turn on segments of the population whom they are
trated more power in his hands. The characters in the battle are the
entrusted to leader.
terrorists, the victims, the warriors who oppose the terrorists, and
The story must reach an audience. Gardner (1995) points out
the audience that watches what is happening,
that no matter what the story, if there is no audience for it, it is
Perceptions of leaders are filtered through stories. The reality
dead. So a leader needs a story to which his or her audience will
may be quite different from the stories. For example, Stalin was
respond. The leader needs to take into account the experiential
responsible for the deaths of many millions of Soviet and other
mode of thinking of the audience and the kinds of changes in
citizens. Yet when he died, there was a great deal of sadness
points of view to which the audience is likely to be responsive. The
among many citizens of the USSR. For many years, Stalin was
leader must also have an organizational structure to implement the
idolized, despite his responsibility for so many deaths. Even today,
structure. Further, he or she needs in some way embody the story
many people still idolize Hitler. People see the leaders only
he or she has to tell. If the leader fails to do so, then that leader’s
through their stories, not through any objective reality. The stories
leadership may come to be seen as bankrupt. For example, cover-
may be based in part on objective reality, but the part may be fairly
ups by Richard Nixon and Bill Clinton seriously undermined their
small.
leadership because they came to be seen as leaders who held others
It is important to realize that stories are social constructions.
to one standard and themselves to another. Many people recently
Different people and different groups may interpret the same
have lost faith in certain church leaders who held their flocks to a
events in different ways. Leadership is the attempt to capture the
standard of morality that they themselves flagrantly violated by
minds of the people to accept one’s version of events. In presi-
abusing children or covering up such abuse. One cannot lead
dential campaigns, such as that of 2004 between Bush and Kerry,
effectively if one asks people to do as one says, not as one does.
much of the campaign is devoted to the fight for the storyline that
Gardner (2004) has further suggested that a good story over-
people will accept. For example, was the war in Iraq a war against
comes resistances. Leaders must expect groups of followers to
international terrorists or against a bad regime unconnected with
resist some of the leaders’ ideas. It is the leaders’ responsibility to
the terrorists? The candidates took opposite positions, each trying
devise ways to overcome these resistances. Good stories also
to persuade listeners to believe their story. Of course, there is a
involve representational redescription. Ideas can be expressed in
truth underlying the battle: The regime either was or was not
many ways. The more different ways in which a leader’s ideas can
connected to international terrorists. For better or worse, truth
be expressed, and the more compelling these ways are, the more
plays at best a minor role in persuading people one way or another.
likely the leader is to persuade followers to come along. A good
Strong emotions, such as fear, rage, joy, and sorrow, probably play
story also embodies resonance. At a given time and in a given
much more powerful roles.
place, certain ideas will resonate with followers, others will not.
Stories are hierarchically arranged so that people have multiple
Establishing resonance can go a long way toward persuading
stories they can accept at a given time. The challenge of the leader
people to listen. Finally, the story will be more effective to the
is to create a story that is higher in people’s hierarchies rather than
extent it incorporates real-world events. Followers need to see how
lower. Moreover, the leader in a competition may try to undermine
the leader’s ideas relate to the lives the followers live from day to
the story or stories of his or her competitors, trying to show that the
day.
story he or she proposes is the one that followers should accept.
Sternberg’s characterization of stable story elements.
Stories
Again, truth may play a relatively small role in what stories people
have certain stable elements (Sternberg, 1995b; Sternberg, Hojjat,
accept. Rather, their emotional needs are likely to be key. Effective
& Barnes, 2001). First, they have beginnings, middles, and ends.
leaders know this, and pitch their stories to resonate with people’s
In this way, they are like scripts (Schank & Abelson, 1977).
emotions.
Sometimes, leaders start with a story that works well in the
Stories can become self-fulfilling prophecies. For example, the
beginning and discover that the end does not work. They either
governments of both the United States and Russia have a history of
change stories, or they lose the support of their followers. Indeed,
acting aggressively toward nations or interest groups that displease
stories are constantly being rewritten in order to suit the needs of
them. In Chechnya, the Russian government has acted in very
the leader-follower unit. For example, the story of the war in Iraq
harsh ways to suppress rebellions. The harshness of the actions
changed multiple times in 2004 in order to accommodate emerging
creates resistance, which in turn creates more harshness, and so
facts and the perceived needs of followers. Some individuals found
forth. The same dynamic has played out in the Israeli-Palestinian

---

10
STERNBERG
conflict. When people have stories, they act in ways to make them
In terms of the model of types of creativity described above, the
come true, and often they do.
kinds of leaders vary widely. Replicators and conservers pretty
Stories always have two principal roles. One is for the leader,
much leave existing paradigms as they are. Doctors change things
the other for followers. Some of the stories are more symmetrical,
that are wrong. Turn-around specialists make major changes in the
others less so. For example, a democratic leader expects a great
organization they lead. They are redirectors or reinitiators.
deal of participation from followers in setting and determining
policies. An autocratic leader expects little or no participation.
Success or Failure of Stories
Leaders and followers clearly differ in the level of symmetry with
which they are comfortable. For example, Russia has a history of
Leaders succeed to the extent that they (a) have a story that fits
less symmetry, and when more symmetry was introduced, the
their followers’ needs, (b) communicate that story in a compelling
system as implemented under Boris Yeltsin was not particularly
way, (c) implement the story in a way that suggests it is succeeding
successful. Today, Vladimir Putin is moving back toward a more
(given that there may be a difference between the perception and
asymmetrical system.
the reality), and (d) persuade followers, in the end, that the story
accomplished what it was supposed to have accomplished. Leaders
Classification of Stories
fail to the extent that they (a) have a story that fails to fit their
followers’ needs, (b) fail in communicating their story, (c) fail in
Christopher Rate, a graduate student at Yale, and I are working
implementing the story, (d) fail in persuading followers that they
to create and test a taxonomy of stories. Our main hypothesis is
have accomplished what they promised, (e) fail to have any co-
that leaders will succeed differentially well, depending in large
herent story at all, (f) seem to move from story to story without
part on the extent of match between the stories of the leaders and
convincing followers that there is a need to change stories, or (g)
the followers. Some tentative examples of stories we are exploring
allow a story of successful leadership to be replaced with a story
are
of personal failings. For example, the leader may come to be
• The carpenter—The leader who can build a new organization
viewed as in power not to lead, but to maintain power at all costs,
or society
to enrich him or herself personally, to increase his or her power to
• The CEO—The leader who can “get things done”
the maximum extent possible, or to harm groups not obeying him
• The communicator—The leader who can communicate with
or her. In these cases, stories of leadership come to be replaced
diverse followers
with stories of personal failings. Joseph Stalin, Adolph Hitler, and
• The conquerer—The leader who is going to conquer all
many other leaders eventually lost the mantle of leadership as a
enemies
result of such personal failings.
• The conserver—The leader who will make sure things stay the
In the 2004 U.S. presidential election, for example, John Kerry’s
wonderful way they are
campaign floundered over the summer. Why? First, it was not
• The cook—The leader who has the recipe to improve the life
clear what his story is. Or if he had a clearly defined story, he
of his or her followers
failed adequately to convey it. Second, he did not persuade people
• The deep thinker—The leader who will make sense out of
that he would be able to implement a story, if he had one. Third,
what is going on
he allowed Bush to define the story to which people paid attention,
• The defender—The leader who will save all followers from
namely, one of a leader who would be a defender and even as a
harm
warrior in the face of threats of terrorism. Fourth, he allowed the
• The deity—The leader who presents him or herself as savior
Bush campaign to portray him as a “flip-flopper,” that is, someone
• The diplomat—The leader who can get everyone to work
who kept changing stories in the hope of finding one that worked.
together
Leaders need to be creative in inventing their stories, analyti-
• The doctor—The leader who can cure what is wrong with the
cally intelligent in addressing the strengths and weaknesses of their
organization
stories, practically intelligent in implementing the stories and
• The ethicist—The leader who pledges to clean up the place
persuading followers to listen to them, and wise in generating and
• The lifesaver—The leader who will rescue followers from
instantiating stories that are for the common good. They may fail
otherwise certain death
if they lack creativity, intelligence, or wisdom, and especially if
• The organizer—The leaders who can create order out of chaos
they foolishly succumb to the fallacies described earlier (such as
• The plumber—The leader who can fix all the leaks
egocentrism), which can divert them from a successful leadership
• The politician—The leader who understands how “the sys-
story to a story of failed leadership.
tem” works
Stories fit into a contingency-based notion of leadership. There
• The replicator—The leader who is going to be like some past
is no one story that works for all organizations in all times or all
individual
places. Rather, success of a story fits into the situation at a given
• The scout—The leader who can lead followers to new and
time and place. When Tolstoy speculated, in Anna Karenina, that
uncharted territory
if it had not been Napoleon, it would have been someone else
• The ship captain—The captain of a ship navigating through
fitting that particular situation, he was partially right. The situation
turbulent times
demanded a certain kind of story. But it was not certain that
• The turn-around specialist—The leader who can turn around
anyone would come along who could tell that story in a compelling
a failing organization
way and convince people to listen to him or her.
• The warrior chieftain—The leader who will lead followers to
Creativity is in itself insufficient for successful leadership. In
fight, defensively or offensively, enemies, seen or unseen
addition to being able to come up with good ideas, a good leader

---

CREATING A VISION OF CREATIVITY
11
needs to know how to analyze and evaluate his or her ideas.
and cross-sectional data using latent growth structural models. Experi-
Successful intelligence is instrumental in this evaluation process.
mental Aging Research, 18, 145–166.
Rubenson, D. L., & Runco, M. A. (1992). The psychoeconomic approach
to creativity. New Ideas in Psychology, 10, 131–147.
Conclusions
Schank, R., & Abelson, R. P. (1977). Scripts, plans, goals, and under-
standing: An inquiry into human knowledge structures. Hillsdale, NJ:
For 25 years or so, I have been trying to understand the nature
Erlbaum.
of creativity, as well as how to measure and develop it. In this
Sloman, S. A. (1996). The empirical case for two systems of reasoning.
essay, I have tried to characterize work I have done in three stages
Psychologica