Cognition and Emotion
ISSN: 0269-9931 (Print) 1464-0600 (Online) Journal homepage: http://www.tandfonline.com/loi/pcem20
How does emotion influence different creative
performances? The mediating role of cognitive
flexibility
Wei-Lun Lin, Ping-Hsun Tsai, Hung-Yu Lin & Hsueh-Chih Chen
To cite this article: Wei-Lun Lin, Ping-Hsun Tsai, Hung-Yu Lin & Hsueh-Chih Chen (2014) How
does emotion influence different creative performances? The mediating role of cognitive
flexibility, Cognition and Emotion, 28:5, 834-844, DOI: 10.1080/02699931.2013.854195
To link to this article: http://dx.doi.org/10.1080/02699931.2013.854195
Published online: 18 Nov 2013.
Submit your article to this journal
Article views: 750
View related articles
View Crossmark data
Citing articles: 6 View citing articles
Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=pcem20
Download by: [National Taiwan Normal University]
Date: 14 December 2015, At: 20:31
COGNITION AND EMOTION, 2014
Vol. 28, No. 5, 834–844, http://dx.doi.org/10.1080/02699931.2013.854195
BRIEF REPORT
Downloaded by [National Taiwan Normal University] at 20:31 14 December 2015
How does emotion influence different creative
performances? The mediating role of cognitive flexibility
Wei-Lun Lin1, Ping-Hsun Tsai1, Hung-Yu Lin1, and Hsueh-Chih Chen2
1
Department of Psychology, Fo Guang University, Jiaosi Shiang, Yilan County, Taiwan
Department of Educational Psychology and Counselling, National Taiwan Normal University,
Taipei City, Taiwan
2
Cognitive flexibility is proposed to be one of the factors underlying how positive emotions can improve
creativity. However, previous works have seldom set up or empirically measured an independent index to
demonstrate its mediating effect, nor have they investigated its mediating role on different types of
creative performances, which involve distinct processes. In this study, 120 participants were randomly
assigned to positive, neutral or negative affect conditions. Their levels of cognitive flexibility were then
measured by a switch task. Finally, their creative performances were calibrated by either an open-ended
divergent thinking test or a closed-ended insight problem-solving task. The results showed that positive
emotional states could reduce switch costs and enhance both types of creative performances. However,
cognitive flexibility exhibited a full mediating effect only on the relationship between positive emotion
and insight problem solving, but not between positive emotion and divergent thinking. Divergent
thinking was instead more associated with arousal level. These results suggest that emotions might
influence different creative performances through distinct mechanisms.
Keywords: Emotional state; Cognitive flexibility; Divergent thinking; Insight problem solving.
The relationship between emotion and cognition
has been ardently investigated in the past few
decades. Among cognitive functions, creativity is
the highest human mental activity (Guilford,
1967). Mumford (2003) called the influence of
emotion one of the most important topics in the
Correspondence should be addressed to: Wei-Lun Lin, Department of Psychology, Fo Guang University, No. 160, Linwei Rd.,
Jiaosi Shiang, Yilan County 26247, Taiwan. E-mail: [email protected]
We thank Professor Rothermund and other anonymous reviewers for their extremely helpful comments on the draft. We also
thank Chao-Yuan Tseng for helping with the data analysis. This research was supported by grants to the first author from the
National Science Council of Taiwan [grant number 99-2410-H-431-015] and to the fourth author from the “International
Research-Intensive Center of Excellence Program” of National Taiwan Normal University [grant number NSC102-2911-I-003-301]
and the Ministry of Education, Taiwan, under the Aiming for the Top University Plan at National Taiwan Normal University.
© 2013 Taylor & Francis
834
Downloaded by [National Taiwan Normal University] at 20:31 14 December 2015
EMOTION, COGNITIVE FLEXIBILITY AND CREATIVITIES
field of creativity. This study was conducted to
inspect the mechanism of how affect could influence creativity. More specifically, this study investigated the mediating role of cognitive flexibility.
Previous studies have generally shown that
positive affect can inspire more creativity than
neutral or negative affect (for a review, see Hirt,
Devers, & McCrea, 2008). For example, research
has shown that induced positive affect can
enhance remote-association performance (Isen,
Daubman, & Nowicki, 1987), facilitate the correct
solving of insight problems (Isen et al., 1987) and
promote performance in divergent thinking (Hirt,
Levine, McDonald, Melton, & Martin, 1997;
Hirt et al., 2008).
In response, researchers have proposed several
possible mechanisms to explain this phenomenon
(for a review, see Hirt et al., 2008). Some believe
that positive affect enhances cognitive flexibility
and, hence, promotes creativity (De Dreu, Baas, &
Nijstad, 2008; Fredrickson, 2001, 2003; Hirt et al.,
2008; Isen, 2008). Cognitive flexibility refers to
the elasticity of strategies or ability to switch
among stimuli, operations and mental sets when
responding to demands (Isen, 2008; Miyake,
Friedman, Emerson, Witzki, & Howerter, 2000;
Vartanian, 2009). Empirical evidence shows that
positive affect can help individuals employ flexible
strategies in the formation and testing of hypotheses (De Dreu et al., 2008; Hirt et al., 2008;
Vartanian, 2009), switch between aspects of
information, decrease anchoring in a diagnosis
(Estrada, Isen, & Young, 1997) and in a decision
(Djamasbi, 2007), improve cognitive organisation
as reflected in flexible word association (Isen,
Johnson, Mertz, & Robinson, 1985) and categorisation (Isen & Daubman, 1984), enable flexible
deployment of attention in a global–local task
(Baumann & Kuhl, 2005), improve executive
functions (Yang, Yang, & Isen, 2013), broaden
momentary thought–action repertoires, and ensure
the flexible exploration of new ideas (broadenand-build theory: see Fredrickson, 2001, 2003).
However, after reviewing previous studies,
two issues must still be clarified. First, although
many researchers claim that positive affect
enhances cognitive flexibility and, hence, facilitates
creativity, previous studies generally conflate the
measures of cognitive flexibility and creativity. For
example, some have adopted flexibility indices in
divergent thinking tests (Hirt et al., 2008) or
category inclusion in classification tasks (De Dreu
et al., 2008) as measures of cognitive flexibility
while also making these indicators dependent
variables for creativity. The better way to explore
the mechanisms of how emotions could influence
creativity is to set up measures of cognitive
flexibility independently and then conduct mediation analyses. To our knowledge, few studies have
adopted this approach. The only study we
reviewed found no significant mediating effect of
cognitive flexibility (Hsu, 2009). However, Hsu
(2009) used only open-ended creativity measure as
a creativity index, which might have created an
additional problem; this will be discussed next.
Second, past studies have seldom distinguished
different types of creative performances; however,
accumulating evidence indicates that different
creative tasks exhibit distinct properties (Wakefield, 1989) and involve different processes (Isen,
2008; Lin, Hsu, Chen, & Wang, 2012; Lin &
Lien, 2013). It is therefore important to differentiate how positive emotion affects different creative performances. Researchers have usually used
either open-ended (such as divergent thinking
tests) or closed-ended creativity tasks (such as
insight problem-solving tasks), but not both, to
measure creative potential. In a divergent thinking
question (e.g., “What are the uses of a brick?”),
respondents have to answer in as many diverse
ideas as possible (so it is open-ended), to make
novel output more likely (Guilford, 1956). On the
other hand, an insight problem-solving task has a
specific solution goal (in other words, it is closedended). The respondents have to escape from the
obstacles of familiar frames or existing rules under
task constraints (Weisberg, 1995) to find the
correct answer. In this case, successful solutions
require not only expansive thinking but also
constrained thinking in a reciprocal process to
construct a novel and appropriate final solution
(Isen, 2008).
Researchers have recently developed the “dual
process account of creativity” from dual process
COGNITION AND EMOTION, 2014, 28 (5)
835
Downloaded by [National Taiwan Normal University] at 20:31 14 December 2015
LIN ET AL.
theories (e.g., Stanovich & West, 2000) to differentiate open- and closed-ended creative tasks (Lin
et al., 2012; Lin & Lien, 2013). It is proposed that
open-ended thinking mainly relies on intuitive,
associative System 1 processing to obtain numerous and unique responses, while closed-ended
thinking involves a reciprocal process between
System 1 and the analytical, evaluative System 2
processing to find a creative correct answer.
Empirical evidence supports this differentiation
(Lin et al., 2012; Lin & Lien, 2013). For example,
studies have shown that the individual performances of the two tasks have no correlation (e.g.,
Lin, Lien, & Jen, 2005). Furthermore, the two
measures have exhibited different relationships
among certain cognitive factors and personal
variables, such as working memory (Lin & Lien,
2013), personality traits and gender (Lin et al.,
2012). Our recent studies have also demonstrated
that different types of induced affects have their
respective impacts on the two measures of creativity. Divergent thinking performances can be
enhanced by both positive and negative affects,
while insight problem-solving performances can
only be improved by positive affect (Tsai, Lin, &
Lin, 2013). Accordingly, the question of whether
emotions influence divergent thinking and insight
problem solving through different mechanisms is
worth further investigation.
With respect to the issues addressed above, the
present study aimed to use an independent
measurement of cognitive flexibility in the field
of attention research (the switch task, Dreisbach
& Goschke, 2004) to test whether there are
different mediating effects following the different
influences of emotions in an open-ended divergent thinking test and a closed-ended insight
problem-solving task, respectively. Past studies
have demonstrated that only positive affect, not
negative affect, can enhance cognitive flexibility
(e.g., Dreisbach & Goschke, 2004), but divergent
thinking performance was found to be facilitated
by both positive and negative affects (e.g., Tsai
et al., 2013). Furthermore, in contrast to openended divergent thinking that relies only on
System 1 processing (or expansive thinking),
closed-ended insight problem solving requires a
836
COGNITION AND EMOTION, 2014, 28 (5)
reciprocal process between Systems 1 and 2
processing modes (or expansive and constrained
thinking) according to Isen (2008) and the dual
process account of creativity (Lin et al., 2012; Lin
& Lien, 2013). To switch flexibly between
different operations or processing modes may be
more crucial to success in a closed-ended creativity task.
In the following experiment, we induced participants’ positive, neutral or negative emotional
states. There were two reasons to include three
affect conditions in the design. First, it allowed us
to replicate past findings on the influence of
different emotions on cognitive flexibility (e.g.,
Dreisbach & Goschke, 2004) and on different
creative performances (e.g., Tsai et al., 2013).
Second, by contrasting between positive versus
neutral and negative versus neutral conditions, we
could further examine our hypotheses: only positive affect, not affect per se, could improve closedended insight problem solving, but not openended divergent thinking, through the mediation
of cognitive flexibility.
METHODS
Participants, design and general procedures
A total of 120 college students from Fo Guang
University participated in this study (64% women,
mean age 20.6 years). Participants were randomly
assigned to 3 (affect induction: positive/neutral/
negative) × 2 (creativity type: insight problem/
divergent thinking) conditions for a total of six
conditions with 20 participants in each condition.
All the participants received individual tests. To
disguise the purposes of this study, we told
participants that they were going to proceed with
three independent tasks. The first was to rate a
film as teaching materials for next semester. The
participants then watched a film to induce a
certain emotional state. They then filled out a
checklist about their emotional state after the film.
Next, they carried out a switch task to show their
cognitive flexibility, and, finally, completed either
the insight problem task or the divergent thinking
test. After having completed the three tasks, every
EMOTION, COGNITIVE FLEXIBILITY AND CREATIVITIES
participant was debriefed and received appropriate
compensation.
Materials and procedures
Downloaded by [National Taiwan Normal University] at 20:31 14 December 2015
Affect-inducing films and assessments
Three films adopted from a past study (Tsai et al.,
2013) induced positive, neutral and negative
affects; each lasted for about four minutes.1 After
the film, participants rated their current emotional
states from very unpleasant ( 100 points) to very
pleasant (100) and their arousal state from calm
(0) to aroused (100). One of the main purposes of
measuring arousal level in this study was to control
and purify the effect of emotional valence (e.g.,
Russell & Carroll, 1999).
Switch task
Miyake et al. (2000) have conducted meta-analysis
on various tasks designed to inspect human
executive functions. Switching (or shifting)
between multiple tasks, operations or mental sets
is one of the fundamental executive functions, and
a switch task is considered as one of the representative tasks with which to inspect this function.
In switch tasks, participants must respond according to a certain rule, which will change during the
task. The general findings of this task show that,
when the rule changes, participants’ reaction times
lengthen. The extra time spent switching rules
(switch cost) indicates the perseveration of the
prior rule or, conversely, cognitive flexibility (e.g.,
Smillie, Cooper, Tharp, & Pelling, 2009). The
present study modified a switch task from Dreisbach and Goschke (2004) to measure cognitive
flexibility because it was found that positive affect
could effectively reduce the switch cost in this task.
Participants used the e-prime computer programme to perform the task. In each trial, the
screen first showed the fixation point for 250 ms,
followed by a blank screen for 250 ms, and then
presented two numbers (ranging from 2 to 9)
simultaneously in Times New Roman, with a
length of 2.4°, and a width of 1.4°, but in different
colours. Participants had to judge whether the
certain coloured number that appeared was odd or
even. After they responded, the next trial proceeded after a wait of 1000 ms. The position of
relevant and irrelevant digits changed randomly
from trial to trial.
For the first 40 trials, participants had to judge
the green number and ignore the purple one. After
this, instructions changed—participants needed to
judge the grey number and ignore the green one.
There were 20 trials for the new rule. Before the
formal trials, a 12-trial practice (eight trials for the
first rule and four trials for the second rule)
acquainted participants with the procedures.
We also adopted the calculation of cognitive
flexibility from Dreisbach and Goschke (2004);
that is, we calculated two 5-trial average reaction
times for correct responses (excluding the error
trials) after the switch (from the 42nd to the 46th
trial) and before the switch (from the 36th to the
40th trial). The difference between these two
(after/before) was the switch cost. The lower the
switch cost, the higher the cognitive flexibility.
This indicates the ease of preventing interruption
of the old rule and adapting a new one (Dreisbach
& Goschke, 2004).
Insight problem task
The insight problem task, adopted from Lin et al.
(2012), measured participants’ closed-ended creative potentials. This task consists of 10 pure
insight problems (Weisberg, 1995), five verbal and
five figural and is reasonably reliable (Cronbach’s
α = .68). Participants proceeded with this task
within a 20-minute limit and had to indicate for
each problem whether they already knew the
answer. The index of closed-ended creativity was
the revised number of correct items = (correct
items of unknown problems)/(total items of unknown problems) × 10 (total number of items).
1
The contents of each affect induction film are positive (a mischievous monkey plays tricks on a tiger), neutral (the
fabrication process of Michelin tires) and negative (conflict on a football field).
COGNITION AND EMOTION, 2014, 28 (5)
837
Downloaded by [National Taiwan Normal University] at 20:31 14 December 2015
LIN ET AL.
Divergent thinking test
We adopted the Chinese version of the Creative
Thinking Test (Wu, 1998) to measure participants’
open-ended creative potentials. This test was
designed from subtests of the Torrance Test of
Creative Thinking with culturally familiar materials; it has developed a large sample norm in Taiwan,
as well as stable reliability and validity results (Wu,
1998). It consists of a verbal and a figural subtest.
The participants had to write down or complete as
many unusual uses or drawings as possible in the
verbal and figural subtests, respectively. They were
allowed 10 minutes for each subtest. Two independent raters then scored the results for fluency,
flexibility, originality and (for the figural subtest)
elaboration. The inter-rater reliabilities of each
score were above .94. We also computed the
standardised total scores (T scores) for verbal and
figural subtests.
RESULTS
Manipulation check of affect induction
The results indicated the effectiveness of affect
manipulation: F(2, 117) = 63.36, p < .001, η2 = .52.
Further least significant difference (LSD) comparisons (p’s < .001) showed that participants’ selfassessments in the positive affect condition
(M = 55.13, SD = 25.53) were significantly more
positive than those in the neutral affect condition
(M = 15.42, SD = 24.54), and the assessments in
the neutral affect condition were significantly more
positive than those in the negative affect condition
(M = 14.18, SD = 32.18).
2
Arousal also showed significant differences:
F(2, 117) = 13.35, p < .001, η2 = .19. The par‐
ticipants’ arousal levels in the positive affect condition (M = 38.50, SD = 23.43) and negative affect
condition (M = 42.38, SD = 25.55) were both
significantly higher than those in the neutral affect
condition (M = 18.85, SD = 15.10), ps < .001. There
was no difference between positive and negative
affect conditions, p = .43.
Emotional states and switch task
Participants’ error rate on the sampled trials
(36th–40th, 42nd–46th) in switch task was 2.9%.
After excluding error trials, we also compared
participants’ mean response times (RTs) between
correct trials immediately following an error with
mean RTs of their other correct trials (if they did
make an error). The results showed that, before
the switch, RTs were not different, t(10) = 1.67,
p = .13, d = .22. However, the mean RTs for
correct trials right after an error were significantly
longer than the other correct trials after the switch,
t(14) = 2.15, p = .049, d = .25. We therefore
omitted RTs immediately following an error after
the switch for further analyses.
We conducted a two-factor mixed design
analysis of variance (ANOVA) with affect induction (positive/neutral/negative) and rule switch
(before switch/after switch) as independent variables and five-trial average RTs as the dependent
variable.2 The results showed no significant main
effect of affect induction: F(2, 117) = .35, p = .71,
η2 = .006. However, rule switching did have a
significant main effect: F(1, 117) = 46.38, p <
.001, η2 = .28. This indicated a stable switch cost
According to Dreisbach and Goschke (2004), results of the switch task demonstrate a compatibility effect in which RTs
in incompatible (incongruent) trials, where targets and distracters map to different response keys, are longer than in
compatible trials. We conducted a three (affect induction: positive/neutral/negative) × 2 (rule switch: before switch/after
switch) × 2 (response compatibility: compatible/incompatible) three-way ANOVA to inspect compatibility effects. Neither
the main effect of compatibility nor interactions between compatibility and other variables were significant. We therefore
combined RTs of compatible and incompatible trials for later analyses. Possible reasons for the lack of compatibility effect in
our study are that our participants had fewer blocks of trials than in the previous study in order to preserve their emotional
states for further creativity tests. Another possibility is that our participants followed instructions well and focused on the
targets, which would also explain their lower error rate. Our study collecting two blocks of trials from another independent
sample also showed no compatibility effect and a lower error rate (1.82%, Li & Lin, 2013).
838
COGNITION AND EMOTION, 2014, 28 (5)
Downloaded by [National Taiwan Normal University] at 20:31 14 December 2015
EMOTION, COGNITIVE FLEXIBILITY AND CREATIVITIES
(after switch RT: M = 790.29, SD = 138.31;
before switch RT: M = 720.21, SD = 126.40).
There was also a significant interaction effect
between affect induction and the rule switch:
F(2,117) = 3.53, p = .03, η2 = .06. Simple main
effect analyses showed that (F(1, 40) = 25.13;
22.31, p’s < .001 and η2 = .40; .37), for both the
neutral and the negative affect conditions, RTs
after the switch (neutral: M = 788.64, SD =
155.44; negative: M = 799.84, SD = 135.20)
were significantly slower than those before the
switch, (neutral: M = 700.11, SD = 126.37;
negative: M = 709.57, SD = 108.81). However,
for the positive affect condition, RT difference was
only marginally significant before (M = 750.94,
SD = 139.44) and after (M = 782.39, SD =
125.56) the rule switch, F(1, 40) = 3.78, p = .07,
η2 = .08.
For the following mediation analyses, the
switch costs (RTs after switch minus RTs before
switch, the cognitive flexibility index) of the three
conditions (see Table 1) were also computed and
compared. LSD analyses revealed switch costs for
both the neutral affect condition (M = 88.52, SD
= 111.05) and the negative affect condition (M =
90.27, SD = 118.95) were significantly higher (p =
.025; p = .021) than those for the positive affect
condition (M = 31.45, SD = 107.87). The above
results demonstrated that participants in the positive affect condition could switch rules more
quickly, which replicated a previous study
(Dreisbach & Goschke, 2004) which demonstrated that a positive emotional state can foster
better cognitive flexibility.
Emotional states and creativity performance
We compared the participants’ performance on the
insight problem task and divergent thinking test
separately in three induced affect conditions3 (see
also Table 1). There was a significant difference
between the conditions in insight problem performance, F(2, 57) = 5.45, p = .007, η2 = .16. LSD
comparisons (p = .002; .029) showed that the
numbers of correct items were significantly higher
in the positive affect condition (M = 4.10, SD =
1.71) than those in both neutral (M = 2.45, SD =
1.32) and negative affect conditions (M = 2.95, SD
= 1.79).
As for divergent thinking performance (see also
Table 1), there were also significant differences
among conditions for the total scores in the verbal
subtest, F(2, 57) = 6.30, p = .003, η2 = .18, and for
the total scores in the figural subtest, F(2, 57) =
6.02, p = .004, η2 = .17. LSD comparisons (p =
.001; .03) showed that the total score on the verbal
subtest in the positive affect condition (M =
175.05, SD = 40.75) was significantly higher
than those in neutral (M = 139.60, SD = 20.05)
and negative affect conditions (M = 152.8, SD =
31.52). Although total scores in the negative
affect condition seemed higher than those in the
Table 1. Means and SDs of participants’ switch costs and creative performance in three induced-affect conditions
Cognitive flexibility
Switch cost (ms)
Creative performance
Insight problem solving (correct items)
Verbal divergent thinking (standardised total scores)
Figural divergent thinking (standardised total scores)
Positive affect
Neutral affect
Negative affect
31.45a (107.9)
88.52b (111.1)
90.27b (119.0)
4.10a (1.7)
175.05a (40.8)
163.20a (24.3)
2.45b (1.3)
139.60b (20.1)
141.00b (22.7)
2.95b (1.8)
152.80b (31.5)
157.05a (14.4)
Note: Different superscript alphabets mean that there are significant differences between conditions (p < .05).
3
The two-way ANOVA analysis with affect induction and creativity type as independent variables showed no interaction
effects.
COGNITION AND EMOTION, 2014, 28 (5)
839
Downloaded by [National Taiwan Normal University] at 20:31 14 December 2015
LIN ET AL.
neutral condition, the difference did not reach a
significant level (p = .20). For the figural
subtest performance, LSD comparison (p = .001;
.018) showed that total scores in the positive
(M = 163.20, SD = 24.25) and negative affect
conditions (M = 157.05, SD = 14.42) significantly
exceeded those in the neutral affect condition
(M = 141.00, SD = 22.66).
The above results mostly support previous
findings that insight problem-solving performance
benefits only from a positive emotional state, while
both positive and negative affects can improve
divergent thinking performance (Tsai et al., 2013).
Emotional states influence creativity
performance through cognitive flexibility
The results of the analyses above, respectively,
showed that positive affect could reduce switch
costs (and, hence, increase cognitive flexibility in
the switch task) as well as enhance performance in
two types of creativity measures. With three
indices obtained on emotional states (self-rated
affect), cognitive flexibility (switch cost) and creative performances (correct items in insight problem solving/T scores for verbal and figural subtests
in divergent thinking), we further investigated the
mediating role of cognitive flexibility in the affect–
creativity relationships.
Although manipulation checks on self-assessed
ratings and creative performances differed significantly between the different affect conditions,
participants’ self-assessed rating of emotional
states did not strongly correlate with their creative
performances.4 To examine our hypotheses that
only positive affect, not affect per se, could
improve insight problem solving but not divergent
thinking through the mediation of cognitive
flexibility, we computed two contrasts (positive
vs. neutral and negative vs. neutral) on the two
creative performances. In the first contrast, we
coded “the positive condition = 1” as an experimental condition and coded “the neutral condition
= 0” as a control condition. In the second contrast,
4
we coded “the negative condition = 1” and “the
neutral condition = 0”.
To perform mediation analyses on the two
contrasts, partial correlations [control for arousal
to exclude the possible confounding from the
arousal component of emotional states, Russell &
Carroll (1999)] were first calculated. When considering the contrast of positive and neutral affect
conditions on insight problem performances (n =
40), the results showed that there were significant
positive correlations between affect condition and
insight problem-solving performance (r = .49, p =
.008), between affect condition and cognitive
flexibility (r = .45, p = .02), and between cognitive
flexibility and insight problem-solving performance (r = .44, p = .004). However, when considering the contrast of negative and neutral affect
conditions on insight problem performances (n =
40), the results did not show any significant
correlations. The correlations between affect condition and insight problem-solving performance,
affect condition and cognitive flexibility, and
cognitive flexibility and insight problem-solving
performance were .18 (p = .32), .13 (p = .70) and
.01 (p = .97), respectively.
Similar analyses were computed for divergent
thinking performances. For the contrast of positive
versus neutral affect conditions (n = 40), the
results showed that the affect condition significantly correlated with both verbal and figural
divergent thinking performances (r = .49, p =
.004; r = .50, p = .004) and that affect condition
and cognitive flexibility were marginally correlated
(r = .33, p = .06), but cognitive flexibility was not
correlated with either verbal (r = .25, p = .13) or
figural (r = .15, p = .38) divergent thinking
performances. For the contrast of negative versus
neutral affect conditions (n = 40), the results
revealed that affect condition was not correlated
with verbal divergent thinking (r = .07, p = .68)
and cognitive flexibility was not correlated with
figural divergent thinking (r = .18, p = .28).
According to Baron and Kenny’s (1986) criterion, only the first contrast (positive and neutral
The results might be due to a relatively large variance in separated affect conditions.
840
COGNITION AND EMOTION, 2014, 28 (5)
EMOTION, COGNITIVE FLEXIBILITY AND CREATIVITIES
Cognitive Flexibility
β=0.33*(0.44**)
β=0.45*
Downloaded by [National Taiwan Normal University] at 20:31 14 December 2015
Affect Condition
(Positive vs. Neutral)
Insight Problem
β=0.34(0.49**)
Figure 1. The mediating effect of cognitive flexibility on the
relationship between positive affect and performance on an insight
problem (the analysis included data of positive and neutral affect
conditions). The values in parentheses indicate the strength of the
path prior to the inclusion of the mediating variable, *p < .05;
**p < .01.
affect conditions) on insight problem-solving
performances could further perform the mediation
analysis of cognitive flexibility, as all three pairwise correlations were significant. The following
mediation analysis was conducted using affect
conditions (positive vs. neutral) as independent
variables, cognitive flexibility as a mediator and
performance in the insight problem task as the
dependent variable. Arousal level was included as a
control variable.
The regression analyses of Model 1 (affect
condition /arousal → cognitive flexibility) and
Model 2 (affect condition /arousal → insight
problem) showed that affect condition could
positively and significantly predict cognitive flexibility and insight problem performance (β = .45,
p = .02, R2 = .14; β = .49, p = .008, R2 = .24). When
we included both affect condition and cognitive
flexibility in Model 3 (affect condition /cognitive
flexibility/arousal → insight problem), the amount
of variance explained significantly increased: R2 =
.33, ΔR2 = .10, F = 5.15, p = .029. Cognitive
flexibility retained significant predictive power
over insight problem performance (β = .33,
p = .029); however, affect condition had reduced
its predictive power over insight problem performance (β = .34, p = .063),5 as Figure 1 shows.
Furthermore, a Sobel test (Holmbeck, 2002) for
examining the mediating effect showed significant
results, z = 2.00, p = .04.
The above results demonstrated that, after
including cognitive flexibility, the direct effect of
affect on insight problem-solving performance
significantly decreased, indicating the full mediating effect of cognitive flexibility for positive affect
compared to neutral affect. Because the variables
mentioned above were not correlated under the
contrast of negative and neutral affect conditions,
the mediation effect of cognitive flexibility could
not exist for the contrast between negative and
neutral affect conditions on insight problem solving. These results therefore demonstrate that, as
we predicted, cognitive flexibility could only
mediate the effect of positive affect, but not affect
per se, on insight problem performance.
Divergent thinking and arousal
As mentioned earlier, the cognitive flexibility
index and divergent thinking performances (for
both verbal and figural subtests) did not correlate
with each other. Cognitive flexibility was also not
a mediator between affect (for both positive vs.
neutral and negative vs. neutral contrasts) and
divergent thinking performance. Previous studies
have shown that arousal can effectively predict
divergent thinking, no matter whether in positive
or negative emotional states (e.g., Tsai et al.,
2013). Did arousal mediate the relationships
between emotional states and divergent thinking
performance? Because divergent thinking scores
and arousal were not different in the positive and
negative affect conditions and were both higher
than the neutral condition, we computed a contrast to code the affect conditions: positive =
negative = 1 (n = 40); and neutral = 0 (n = 20).
Correlational analyses (n = 60) showed that the
affect condition positively correlated with verbal
divergent thinking performance (r = .33, p = .005)
and with arousal (r = .42, p < .001), and that
arousal significantly corresponded to verbal divergent thinking (r = .28, p = .015). There was no
5
The β value of cognitive flexibility for predicting insight problem performance seemed smaller than that of positive
affect. However, the t-test showed a significant result for the former and a marginally significant result for the latter, possibly
because of different variances.
COGNITION AND EMOTION, 2014, 28 (5)
841
Downloaded by [National Taiwan Normal University] at 20:31 14 December 2015
LIN ET AL.
significant correlation between arousal and figural
divergent thinking. Therefore, we conducted a
mediation analysis of arousal only on verbal
divergent thinking.
The regression analyses of Model 1 (affect
condition → arousal) and Model 2 (affect condition → verbal divergent thinking) showed that the
affect condition could boost arousal and verbal
divergent thinking (β = .42, p = .001, R2 = .17; β =
.33, p = .009, R2 = .11). When both affect
condition and arousal were included in Model 3
(affect condition/arousal → verbal divergent thinking), arousal’s power to predict verbal divergent
thinking failed to reach significant levels (β = .17,
p = .21); however, affect’s still did (β = .26, p = .05),
F = 4.47, p = .016, R2 = .13, ΔR2 = .02. These
results indicate that arousal also could not mediate
the effect of emotions on divergent thinking.
DISCUSSION
The present study independently measured cognitive flexibility with a switch task and distinguished open-ended (divergent thinking) versus
closed-ended (insight problem solving) creativity
measures in order to investigate cognitive flexibility’s mediating role in the relationships between
positive emotional states and different creative
performances. The results of this study showed
that induced positive affect, but not negative affect,
could reduce switch costs—that is, promote cognitive flexibility, and further enhance participants’
insight problem-solving performance. Cognitive
flexibility played a full mediating role in the
positive affect—closed-ended creativity relationship. On the other hand, although positive affect
could also improve divergent thinking, this influence was not mediated by cognitive flexibility.
These results met our expectations. With
respect to Isen’s (2008) notions and the dual
process account of creativity (Lin et al., 2012; Lin
and Lien, 2013), open-ended divergent thinking
relies mainly on System 1 processing (expansive
thinking), while closed-ended insight problem
solving involves a reciprocal process switching
between Systems 1 and 2 (constrained thinking)
842
COGNITION AND EMOTION, 2014, 28 (5)
processing modes. Cognitive flexibility is more
crucial to successfully solving insight problems
than to divergent thinking.
The results also agreed with physiological
findings. For example, evidence has shown that
positive affect can affect the brain’s flexibilityrelated anterior cingulate cortex, increasing its
activation, raising its dopamine levels and promoting insight problem solving (Ashby, Valentin,
& Turken, 2002; Subramaniam, Kounios, Parrish,
& Jung-Beeman, 2009). Studies also found that
individuals who are better at solving insight
problems exhibited more transformations between
α and β brainwaves when solving closed-ended
creativity problems (Tseng, Tsai, Lin, & Huang,
2012) and were better at controlling their own
brainwave transformations (Li, Lin, Huang, &
Tseng, 2012) which might represent greater mental flexibility; however, individuals with better
divergent thinking abilities did not show these
patterns (Li et al., 2012; Tseng et al., 2012).
These results support our findings that the mechanism of positive affect on insight problem solving
is through the function of cognitive flexibility.
Our data did not demonstrate a mediating role
for cognitive flexibility between affect and divergent
thinking performance. Arousal level did not mediate this relationship either. However, arousal still
coincided with verbal divergent thinking performance (r = .28). Some researchers have proposed that
arousal caused by emotional state may be a possible
mechanism through which emotions can influence
creativity. For example, Adaman and Blaney (1996)
hypothesised that individuals feel uncomfortable
with a high arousal level and use participation in
creative activities to reduce such tension. Previous
studies have also shown that arousal effectively
predicts divergent thinking (e.g., Tsai et al., 2013).
A possible reason for arousal’s lack of mediating
effect between emotions and divergent thinking is
that both positive and negative conditions in this
study triggered only moderate levels of arousal.
More arousing films (see Tsai et al., 2013) may
reveal more significant results. In addition, we
found a difference between the verbal and figural
subtests of divergent thinking in relation to emotional state and arousal level. These results were not
Downloaded by [National Taiwan Normal University] at 20:31 14 December 2015
EMOTION, COGNITIVE FLEXIBILITY AND CREATIVITIES
consistent with Tsai et al. (2013). Researchers have
found that verbal and figural creativity are separately related to verbal and visual abilities (e.g.,
Palmiero, Nakatani, Raver, Belardinelli, & Leeuwen, 2010). The issue of domain-general versus
domain-specific creativity calls for further debate
and exploration.
In summary, although various manipulations of
emotional states, measurements of cognitive flexibility and open- and closed-ended creativity can
still be further investigated, this study demonstrated that cognitive flexibility only mediated the
link between positive emotional states and one
type of creative performance: closed-ended insight
problem solving. These results suggest that emotions influence different creative performances
through distinct mechanisms.
Manuscript received 23 April
Revised manuscript received 7 October
Manuscript accepted 7 October
First published online 15 November
2013
2013
2013
2013
REFERENCES
Adaman, J. E., & Blaney, P. H. (1996). The effects of
musical mood induction on creativity. Journal of
Creative Behavior, 22, 95–108.
Ashby, F. G., Valentin, V. V., & Turken, A. U. (2002).
The effects of positive affect and arousal on working
memory and executive attention: Neurobiology and
computational models. In S. Moore & M. Oaksford
(Eds.), Emotional cognition: From brain to behaviour
(pp. 245–287). Amsterdam: John Benjamins.
Baron, R. M., & Kenny, D. A. (1986). The moderatormediator variable distinction in social psychological
research: Conceptual, strategic and statistical considerations. Journal of Personality and Social Psychology, 51, 1173–1182. 10.1037/0022-3514.51.6.1173
Baumann, N., & Kuhl, J. (2005). Positive affect and
flexibility: Overcoming the precedence of global over
local processing of visual information. Motivation
and Emotion, 29(2), 123–134. 10.1007/s11031-0057957-1
De Dreu, C. K. W., Baas, M., & Nijstad, B. A. (2008).
Hedonic tone and activation level in the moodcreativity link: Toward a dual pathway to creativity
model. Journal of Personalityand Social Psychology, 94,
739–756.
Djamasbi, S. (2007). Does positive affect influence the
effective usage of a decision support system?Decision
Support Systems, 43, 1707–1717. doi:10.1016/j.
dss.2006.09.002
Dreisbach, G., & Goschke, T. (2004). How positive
affect modulates cognitive control: Reduced perseveration at the cost of increased distractibility. Journal
of Experimental Psychology: Learning, Memory, and
Cognition, 30, 343–353.
Estrada, C. A., Isen, A. M., & Young, M. J. (1997).
Positive affect facilitates integration of information and
decreases anchoring in reasoning among physicians.
Organizational Behavior and Human Decision Processes,
72(1), 117–135. doi:10.1006/obhd.1997.2734
Fredrickson, B. L. (2001). The role of positive emotions
in positive psychology: The broaden-and-build theory of positive emotions. The American Psychologist,
56, 218. doi:10.1037/0003-066X.56.3.218
Fredrickson, B. L. (2003). The value of positive
emotions. American Scientist, 91, 330–335.
Guilford, J. P. (1956). The structure of intellect.
Psychological Bulletin, 53, 267–293. doi:10.1037/
h0040755
Guilford, J. P. (1967). The nature of human intelligence.
New York: McGraw-Hill.
Hirt, E. R., Devers, E. E., & McCrea, S. M. (2008). I
want to be creative: Exploring the role of hedonic
contingency theory in the positive mood-cognitive
flexibility link. Journal of Personality and Social
Psychology, 94, 214–230. doi:10.1037/0022-3514.94.
2.94.2.214
Hirt, E. R., Levine, G., McDonald, H., Melton, R., &
Martin, L. L. (1997). The role of mood in quantitative and qualitative aspects of performance: Single or
multiple mechanisms? Journal of Experimental Social
Psychology, 33, 602–629. doi:10.1006/jesp.1997.1335
Holmbeck, G. N. (2002). Post-hoc probing of significant moderational and mediational effects in studies
of pediatric populations. Journal of Pediatric Psychology, 27(1), 87–96. doi:10.1093/jpepsy/27.1.87
Hsu, C. C. (2009). The multi-component model of mood
and creative thinking (MCMC): Impact of regulatory
focus, valence and activation components of mood on the
creative thinking, and mediating effects (Unpublished
doctoral dissertation). National Taiwan Normal
University, Taipei, Taiwan.
Isen, A. M. (2008). Some ways in which positive affect
influences decision making and problem solving.
In Lewis, M., Haviland-Jones, J. M., & Barrett, L. F.
(Eds.), Handbook of emotions (pp. 548–573). New
York, NY: Guilford Press.
COGNITION AND EMOTION, 2014, 28 (5)
843
Downloaded by [National Taiwan Normal University] at 20:31 14 December 2015
LIN ET AL.
Isen, A. M., & Daubman, K. A. (1984). The influence
of affect on categorization. Journal of Personality and
Social Psychology, 47, 1206–1217. doi:10.1037/00223514.47.6.1206
Isen, A. M., Daubman, K. A., & Nowicki, G. P.
(1987). Positive affect facilitates creative problem
solving. Journal of Personality and Social Psychology,
52, 1122–1131. doi:10.1037/0022-3514.52.6.1122
Isen, A. M., Johnson, M. M., Mertz, E., & Robinson, G.
F. (1985). The influence of positive affect on the
unusualness of word associations. Journal of Personality
and Social Psychology, 48, 1413–1426. doi:10.1037/
0022-3514.48.6.1413
Li, Y. H., & Lin, W. L. (2013). How to generate creative
hypotheses? Investigation on factors influencing the
generation of new-perspective hypothesis in a rulediscovery task (Unpublished master’s dissertation).
Fo Guang University, Yilan, Taiwan.
Li, Y. H., Lin, W. L., Huang, A. C. W., & Tseng, C.
Y. (2012, October). The creative individuals let
themselves be creative: Cortical control of creative
individuals with different creative potentials. Poster
presented in the 42nd annual meeting of Society for
Neuroscience, New Orleans, LA, USA.
Lin, W. L., Hsu, K. Y., Chen H. C., & Wang J. W.
(2012). The relations of gender and personality traits
on different creativities: A dual-process theory
account. Psychology of Aesthetics, Creativity, and the
Arts, 6(2), 112–123. doi:10.1037/a0026241
Lin, W. L., & Lien, Y. W. (2013). The different role of
working memory in open-ended versus closed-ended
creative problem solving: A dual-process theory
account. Creativity Research Journal, 25(1), 85–96.
doi:10.1080/10400419.2013.752249
Lin, W. L., Lien, Y. W., & Jen, C. H. (2005). Is the
more the better? The role of divergent thinking in
creative problem solving. Chinese Journal of Psychology, 47, 211–227.
Miyake, A., Friedman, N. P., Emerson, M. J., Witzki,
A. H., & Howerter, A. (2000). The unity and
diversity of executive functions and their contributions to complex “frontal lobe” tasks: A latent
variable analysis. Cognitive Psychology, 41, 49–100.
Mumford, M. A. (2003). Where have we been, where
are we going? Taking stock in creativity research.
Creativity Research Journal, 15, 107–120.
Palmiero, M., Nakatani, C., Raver, D., Belardinelli, M.
O., & Leeuwen, C. (2010). Abilities within and across
visual and verbal domains: How specific is their
844
COGNITION AND EMOTION, 2014, 28 (5)
influence on creativity? Creativity Research Journal,
22, 369–377. doi:10.1080/10400419.2010.523396
Russell, J. A., & Carroll, J. M. (1999). On the bipolarity
of positive and negative affect. Psychological Bulletin,
125(1), 3–30. doi:10.1037/0033-2909.125.1.3
Smillie, L. D., Cooper, A. J., Tharp, I. J., & Pelling, E.
L. (2009). Individual differences in cognitive control:
The role of psychoticism and working memory in
set-shifting. British Journal of Psychology, 100, 629–
643. doi:10.1348/000712608X382094
Stanovich, K. E., & West, R. F. (2000). Individual
differences in reasoning: Implications for the rationality debate? Behavioral and Brain Sciences, 23, 645–
726. doi:10.1017/S0140525X00003435
Subramaniam, K., Kounios, J., Parrish, T. B., & JungBeeman, M. (2009). A brain mechanism for facilitation of insight by positive affect. Journal of
Cognitive Neuroscience, 21, 415–432. doi:10.1162/
jocn.2009.21057
Torrance, E. P. (1974). The Torrance tests of creative
thinking: Norms—technical manual [Research edition]. Princeton, NJ: Personnel Press.
Tsai, P. S., Lin, W. L., & Lin, H. Y. (2013). Right
moods, right creativities: The different effects of
emotions on divergent thinking versus insight problem
solving. Bulletin of Educational Psychology, 45, 19–38.
Tseng, C. Y., Tsai C. H., Lin, W. L., & Huang, A. C.
W. (2012, October). Investigation of the different brain
wave patterns on two groups of creative individuals
selected by open-ended vs. closed-ended creative problems.
Poster presented in the 42nd annual meeting of
Society for Neuroscience, New Orleans, LA, USA.
Vartanian, O. (2009). Variable attention facilitates
creative problem solving. Psychology of Aesthetics,
Creativity, and the Arts, 3(1), 57–59.
Wakefield, J. F. (1989). Creativity and cognition: Some
implications for arts education. Creativity Research
Journal, 2(1–2), 51–63. doi:10.1080/1040041890
9534300
Weisberg, R. W. (1995). Prolegomena to theories of
insight in problem solving: A taxonomy of problems.
In R. J. Sternberg & J. E. Davidson (Eds.), The nature
of insight (pp. 157–196). Boston, MA: MIT Press.
Wu, C. C. (1998). The Chinese version of creative thinking
test. Taiwan: Foundation for Scholarly Exchange.
Yang, H., Yang, S., & Isen, A. M. (2013). Positive affect
improves working memory: Implications for controlled cognitive processing. Cognition and Emotion,
27, 474–482. doi:10.1080/02699931.2012.713325