Thinking styles and modes of thinking: implications for
education and research
The Journal of Psychology: interdisciplinary and applied,
2002, v. 136 n. 3, p. 245-261
The Journal of Psychology: interdisciplinary and applied.
Copyright © Heldref Publications.
Thinking Styles and Modes of Thinking:
Implications for Education and Research
Department of Education
The University of Hong Kong
ABSTRACT. The author investigated the relationship of thinking styles to modes of think-
ing. Participants were 371 freshmen (aged 18 and 19) from the University of Hong Kong.
Participants responded to the Thinking Styles Inventory (R. J. Sternberg & R. K. Wagner,
1992) and the Style of Learning and Thinking (Youth Form; E. P. Torrance, B. McCarthy,
& M. T. Kolesinski, 1988). A major finding was that creativity generating and complex
thinking styles were significantly positively correlated with the holistic mode of thinking
but significantly negatively correlated with the analytic mode of thinking. Thinking styles
that denote the tendency to norm favoring and simplistic information processing were sig-
nificantly positively correlated with the analytic mode of thinking and significantly nega-
tively correlated with the holistic mode of thinking. In a preliminary conclusion, it appears
that the thinking style construct overlaps the mode of thinking construct. Implications of
this finding for teachers and researchers are delineated.
Key words: modes of thinking, thinking styles
IN EDUCATIONAL SETTINGS it is common that one student gets straight As
and another student at the same ability level frequently fails tests. There are var-
ious ways to explain this phenomenon because there are many ways to explain
individual differences in academic performance. For example, two students with
the same abilities may use their abilities differently—that is, they use different
Styles, as an individual-difference variable in human performance, have
long occupied the minds of many scholars. Between the late 1950s and mid-
1970s, there was a proliferation of literature in the area of theories and models of
styles that has become stagnant partially because of the overwhelming output in
the field and partially because of a lack of internal dialogue among researchers
The Journal of Psychology, 2002, 136(3), 245–261
This research was supported by the Wu Jieh-yee Research Fund as administered by The
University of Hong Kong.
Address correspondence to Li-fang Zhang, Department of Education, The Universi-
ty of Hong Kong, Pokfulam Road, Hong Kong; firstname.lastname@example.org (e-mail).
(Jones, 1997). When Riding and Cheema (1991) reviewed the literature on styles,
they identified over 30 labels for the style construct. Consequently, we are left
with a research field that embraces a confusing variety of seemingly different yet
In the past decade or so there has been renewed interest in theories and mod-
els of styles manifested in two topics—conceptually integrating existing style
labels and empirically testing the style labels. Literature on the conceptual inte-
gration of styles is best represented by Curry’s (1983) three-layer “onion” model
of style measures, by Riding and Cheema’s (1991) model of two style dimen-
sions and one family of learning strategies, and by Grigorenko and Sternberg’s
(1995) three traditions of the study of styles. These works have been reviewed in
detail in my recent article (Zhang, 2000b). My present article is based on Grig-
orenko and Sternberg’s conceptualization of styles in the literature. Thus, only
Grigorenko and Sternberg’s work is recapitulated.
Grigorenko and Sternberg (1995) contended that existing models and theo-
ries related to style labels can be classified into three traditions of studying styles:
cognition centered, personality centered, and activity centered. Styles in the cog-
nition-centered tradition most closely resemble abilities. These styles have often
been measured by tests of maximal performance with right and wrong answers.
Within this tradition, Witkin’s (1964) field dependence–independence and
Kagan’s (1966) reflection–impulsivity models have generated the most interest.
Styles in the personality-centered tradition most closely resemble personal-
ity traits, and styles in this tradition are measured with typical performance tests
(no right or wrong answers) rather than maximal performance tests. Models of
styles in this tradition are best represented by Gregorc’s (1982) four main types
of styles and Myers and McCaulley’s (1988) work based on Jung’s (1923) theo-
ry of types. The activity-centered tradition emphasizes the notion of styles as
mediators of various forms of activities that tend to arise from some aspects of
cognition and personality. Literature in this tradition is represented by similar
theories of deep- and surface-learning approaches proposed separately by Mar-
ton (1976), Biggs (1979), and Entwistle (1981).
Empirical studies that attempt to clarify the nature of the relationships
among the different style labels are sparse. All the studies found in the literature
(through a PsycLit search) were reviewed in my (Zhang, 2000b) recent work.
This review suggested that empirical studies about the relationships among the
different style labels have produced diverse results. The results of some of these
studies showed more similarities than differences among various style labels,
whereas others identified more differences than similarities. For example, after
studying 38 university students, Ford (1995) concluded that students’ holist and
serialist competence, as measured by Pask and Scott’s (1972) original testing
materials designed to suit holist and serialist learning strategies, could be pre-
dicted by Riding’s (1991) Cognitive Styles Analysis that is designed to measure
The Journal of Psychology
Harasym, Leong, Juschka, Lucier, and Lorscheider (1996) also found a
strong association between two style labels. Harasym et al. investigated the rela-
tionship between the Myers-Briggs Type Indicator (MBTI; Myers & McCaulley,
1988) and the Gregorc Style Delineator (Gregorc, 1982). They found that each
learning style assessed by the Gregorc Style Delineator corresponded to certain
traits assessed by the MBTI. For example, individuals who scored higher on the
Concrete Sequential learning style scale tended to have traits of sensing and
judging on the MBTI, whereas individuals who scored higher on the Concrete
Random learning style scale tended to have the traits of intuition and perceiving
on the MBTI.
Other researchers, however, found more differences than similarities among
different style labels. For example, Sadler-Smith (1997) carried out a study
among 245 university undergraduates in business studies. The participants
responded to the Cognitive Styles Analysis (CSA; Riding, 1991), the Learning
Preferences Inventory (LPI; Riechmann & Grasha, 1974), the Learning Styles
Questionnaire (LSQ; Honey & Mumford, 1992), and the Revised Approaches to
Studying Inventory (RASI; Entwistle & Tait, 1994). After examining the corre-
lation coefficients among the scales of the different instruments, Sadler-Smith
concluded that there is some overlap between the dimensions measured by the
LSQ and the RASI. However, no statistically significant relationships were iden-
tified between cognitive styles (as measured by the CSA) and any of the other
style constructs investigated.
In a more recent study, Sadler-Smith (1999) examined the relationships
between cognitive styles as measured by the Cognitive Style Index (Allinson &
Hayes, 1996) and learning approaches as measured by the Approaches to Study-
ing Inventory (Gibbs, Habeshaw, & Habeshaw, 1988). Although the results of
this study indicated that analysts tended to adopt a deeper approach to learning
than did intuitives and that intuitives exhibited a stronger preference for collabo-
rative approaches than did analysts, Sadler-Smith concluded that the evidence
found in the relationships between cognitive styles and learning approaches was
not strong and that the two style labels are, at least superficially, independent.
Renewed interest in the style literature also manifested itself through an
additional type of work, that is, the postulation of new theories of styles that
encompass styles from all three traditions to the study of styles. Sternberg’s
(1988, 1997) theory of mental self-government is such a theory. The theory of
mental self-government addresses thinking styles. A thinking style is defined as
our preferred way of using the abilities that we have. Sternberg believes that just
as there are many ways of governing our society, there are many ways to govern
or manage our daily activities. These different ways of managing our activities or
of using our abilities are called thinking styles. People’s thinking styles vary
depending on the stylistic demands of a given situation. Also, thinking styles are
at least partially socialized (Sternberg, 1994, 1997).
The theory of mental self-government proposes 13 thinking styles that fall
along five dimensions of mental self-government. The first dimension is func-
tion, including the legislative, executive, and judicial thinking styles. The second
dimension is related to form, including the hierarchical, oligarchic, monarchic,
and anarchic thinking styles. The third dimension concerns level, including the
global and local thinking styles. The fourth dimension is scope, including the
internal and external thinking styles. The fifth dimension is leaning, including
the liberal and conservative thinking styles. A brief description of each of the 13
thinking styles is provided in the Appendix.
In my opinion, 7 of these thinking styles can be categorized broadly into two
types. The first type (including the legislative, judicial, global, and liberal styles)
is creativity generating and requires complex information processing. People
who use this type of thinking styles tend to be norm challenging and risk taking.
The second type (including the executive, local, and conservative styles) requires
simplistic information processing. People who use this type of thinking styles
tend to be norm favoring and authority oriented.
There are two reasons for this new classification of the 7 styles. The first is
related to the nature of the styles as manifested in their demands for the degree
of complexity in information processing (complex vs. simplistic) and in the
potential products that the use of each type of thinking style is likely to lead to
(creative and norm challenging vs. noncreative and norm favoring). The second
reason is related to the empirical findings based on Sternberg’s theory of mental
self-government. For example, I (Zhang, 2000b; Zhang & Sternberg, 2000)
found that the first type of thinking styles were significantly positively related to
the deep approach to learning but significantly negatively related to the surface
approach to learning. Complementarily, the second type of thinking styles were
significantly positively related to the surface approach to learning but signifi-
cantly negatively related to the deep approach to learning.
In Zhang and Sternberg’s studies, the concept of a deep or surface approach
to learning was the one defined in Biggs’s (1987, 1992) theory of student learn-
ing. An individual who uses a deep approach hopes to gain a real understanding
of what is learned, whereas an individual who uses a surface approach aims to
reproduce what is taught to meet the minimum requirements.
To gain a real understanding of what is learned, one needs to be creative and
to use a nontraditional approach to learning that involves a great deal of complex
information processing, as would an individual who uses the first type of think-
ing styles. To accurately reproduce what is taught, one needs to follow estab-
lished rules in performing tasks, which requires mostly simple information pro-
cessing, as would an individual with the second type of thinking styles. It was on
the basis of this interpretation of the findings on the relationships between think-
ing styles and approaches to learning, along with the nature of styles as discussed
earlier, that I classified the 7 thinking styles into the two types.
Several measures have been constructed to test the theory of mental self-
government. These measures have been used to study a variety of populations
The Journal of Psychology