Content uploaded by Sasha Y. Kimel
Author content
All content in this area was uploaded by Sasha Y. Kimel on Jul 23, 2018
Content may be subject to copyright.
Physiological Correlates of Choice-Induced Dissonance: An Exploration
of HPA-Axis Responses
SASHA Y. KIMEL
1
*, NESTOR LOPEZ-DURAN
2
** AND SHINOBU KITAYAMA
2
1
Harvard University, Cambridge, MA, USA
2
University of Michigan, Ann Arbor, MI, USA
ABSTRACT
Choice can produce a negatively arousing cognitive conflict (called dissonance), which is thought to motivate the chooser to spread their
preferences for the relevant options (called Spreading of Alternatives, or SA). The current work aimed to determine the relationship between
HPA-axis activity and both choice-induced dissonance and its reduction (i.e. SA) among individuals with varying cultural backgrounds.
European–Americans and Asians made a choice between two equally attractive CDs either in the presence of a cue indicative of social eyes
(i.e. public-choice condition) or in the absence thereof (i.e. private-choice condition). As predicted, European–Americans and Asians showed a
reliable SA primarily in the private and public choice conditions, respectively. Importantly, a sharp decline of salivary cortisol was observed
over the span of 30 min, and, moreover, this decline was reliably predicted by the magnitude of SA regardless of either culture or the choice
being private vs. public. These results suggest that although choice-induced dissonance is too weak to elicit an HPA-axis stress response, SA is
associated with variability in the decline of salivary cortisol during the laboratory task. Copyright © 2015 John Wiley & Sons, Ltd.
key words cognitive dissonance; social eyes; culture; spreading of alternatives (SA); cortisol; hypothalamic–pituitary–adrenal (HPA)
axis; choice; decision making
INTRODUCTION
Psychological research over the last half century shows the po-
tential role of dissonance arousal in evoking a choice justifica-
tion effect (Festinger, 1957; Brehm, 2007; see Harmon-Jones,
Amodio & Harmon-Jones, 2009 and Kitayama, Tompson &
Chua, 2014, for recent reviews). When making a decision
between two equally attractive options, individuals are assumed
to experience a negatively arousing cognitive conflict (Cooper
& Fazio, 1984; Higgins, Rhodewalt & Zanna, 1979). To reduce
this dissonance, they subsequently justify their choice by
spreading their original preferences such that the chosen option
becomes more attractive and the rejected option becomes less
so. Importantly, this “Spreading of Alternatives”(i.e. SA) is quite
robust when the choice is self-threatening (Steele, 1988) and,
thus, likely negatively arousing. The goal of the current investiga-
tion was to test the degree to which the hypothalamic–pituitary–
adrenal axis (HPA axis), the body’s major stress response system
(Dickerson & Kemeny, 2004), may be implicated in both
choice-induced dissonance arousal and its reduction (i.e. SA)
among individuals with varying cultural backgrounds.
Culture, choice, and self-threat
Evidence is growing that there is cross-cultural variability in
the conditions in which the act of choosing threatens the self.
Specifically, in European-American cultures, the self is
assumed to be independent (Markus & Kitayama, 1991),
and, thus, behaviors that are performed in private, in the ab-
sence of any social constraint, are considered self-relevant
and self-expressive. These behaviors become potentially
self-threatening by, for example, raising questions about
one’s own competence or moral integrity. In support of this
analysis, previous work shows that the SA effect is quite ro-
bust when the choice is made in private (Cooper & Fazio,
1984). Importantly, however, when European–Americans
are exposed to face-like pictures that appear to be watching
them while making a choice, they perceive their choice to
be socially constrained (Imada & Kitayama, 2010) and, as
a result, the threat to the self is likely mitigated. As may be
expected, under this condition (called the “public-choice”
condition because the choice is made in the presence of a
cue indicating “social eyes”), the SA becomes attenuated
for this group (Imada & Kitayama, 2010; Kitayama, Snibbe,
Markus & Suzuki, 2004).
In contrast, in Asian cultures, the self is assumed to be in-
terdependent (Markus & Kitayama, 1991), and, thus, behav-
iors that are performed in public, in the presence of social
eyes, are considered self-relevant and revealing of the self.
These behaviors become potentially self-threatening by, for
example, evoking concerns about what others may think
about one’s own competence or moral integrity. As may be
expected, Asians show a sizable SA effect when they make
choices while thinking about the preferences of those that
they care about—a manipulation designed to make social
eyes cognitively more salient (Kitayama et al., 2004)—or
when they make a choice for someone they care about
(Hoshino-Browne et al., 2005; Kimel, Grossmann &
Kitayama, 2012). Furthermore, they also show a sizable SA
effect when making a choice while exposed to face-like pic-
tures that appear to be watching them (Imada & Kitayama,
*Correspondence to: Sasha Y. Kimel, Harvard University, Psychology
Department, 33 Kirkland Street, Cambridge, MA 02138, USA. E-mail:
skimel@fas.harvard.edu
**Correspondence to: Nestor Lopez-Duran, University of Michigan, Psychology
Department, 530 Church Street, Ann Arbor, MI 48109-1109, USA. E-mail:
nestorl@umich.edu
Copyright © 2015 John Wiley & Sons, Ltd.
Journal of Behavioral Decision Making,J. Behav. Dec. Making (2015)
Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/bdm.1845
2010; Kitayama et al., 2004). Importantly, Asians show little
or no SA when the choice is made in private, with complete
anonymity, as these actions are seen as “merely personal”
and, thus, of little significance to interdependent selves
(Heine & Lehman, 1997; Kitayama et al., 2004).
One potential caveat comes from a recent argument that
the paradigm used—the free-choice dissonance paradigm—
is fraught with an artifact due to measurement error (Chen
& Risen, 2010). Specifically, in this paradigm, participants
are typically asked to rank a limited set of items (e.g. ten).
They are then given a choice between two items that are sim-
ilarly liked (e.g. their fifth and sixth favorites) and are later
asked to rank these same items again. It is assumed that the
pre-choice ranking is error-free and, thus, that these two
items are truly close in preference. However, these options
may be similarly ranked simply due to measurement error.
That is, even though the two items appear very close in rank-
ing, their true likability might, in fact, be further away from
each another than expected. In this case, the truly more fa-
vored item may indeed be selected and ranked higher in the
subsequent measurement.
Depending on how reliable the measurement of prefer-
ences is, this measurement error may be minimal (Kitayama
et al., 2014). That is, the more error-prone the measurement
procedure is in a given study, the more likely this artifact is
to compromise the validity of the SA measure. One impor-
tant way to guard against this artifact is to make sure that par-
ticipants take sufficient care in the ranking task. Moreover, it
is important to include a control condition in which no self-
threat is expected. If the artifact should come into play, the
SA should be observed in all conditions of the experiment re-
gardless of whether self-threat is induced. Conversely, if lit-
tle or no SA were observed in at least one condition, it would
suggest that the measurement error in the study was negligi-
ble. Thus, the artifact due to measurement error could be
ruled out as a viable explanation for non-zero SAs that might
be observed in other conditions of the study. In the current
work, we therefore included the specific conditions in which
little or no self-threat is anticipated (i.e. the public-choice
condition for European-Americans and private-choice condi-
tion for Asians) as well as allowed participants sufficient
time for the ranking task.
Dissonance and physiological arousal
In the current literature, the presence of the negatively arous-
ing state of cognitive dissonance is typically inferred from
SA or other comparable measures of attitude change (Cooper
& Fazio, 1984; Kitayama et al., 2014). Thus, little is cur-
rently known about whether the self-threat assumed in this
literature may be reflected in physiological stress responses
including those linked to hypothalamic–pituitary–adrenal
axis (HPA-axis) activity (Dickerson & Kemeny, 2004).
Previous research examining physiological correlates of
dissonance has only used measures of sympathetic nervous
system (SNS) activity (e.g. heart rate, pulse-amplitude).
One study examined its relationship with choice-induced dis-
sonance and failed to obtain any evidence of SNS activation.
However, this study was also unsuccessful in obtaining a
reliable SA effect (Gerard, 1967), suggesting certain method-
ological problems (see Croyle & Cooper, 1983). Other stud-
ies have used an induced compliance paradigm in which
participants are asked to act in a way that is contrary to their
own attitudes (e.g. writing a counter-attitudinal essay). These
studies showed an increase in SNS activity (i.e. skin conduc-
tance) when dissonance was presumed high (Croyle & Coo-
per, 1983; Elkin & Leippe, 1986; Losch & Cacioppo, 1990).
However, there was an inconsistent relationship between
SNS activity and the reduction of dissonance (Elkin &
Leippe, 1986; Elliot & Devine, 1994). To date, physiological
correlates of any kind have not been reliably examined
within the free-choice dissonance paradigm. Moreover, no
study—with either of the paradigms—has tested HPA-axis
activity.
The HPA axis can be activated independently of the SNS,
as the two systems are linked to separate networks originat-
ing in the hypothalamus and the locus ceruleous, respectively
(Tsigos & Chrousos, 2002). Although differential activation
of both systems can be stimulus-specific, under conditions
of stress, they function in a continuum, with the HPA axis
having a higher activation threshold than the SNS (Lovallo
& Thomas, 2000; Tsigos & Chrousos, 2002). Moreover,
while SNS activity can measure arousal but not necessarily
stress per se (Bjorklund, Hokfelt & Owman, 2010), the
HPA axis is responsive to more specific stressors, especially
to self-threat and threats that are socially evaluative (Mason,
1968; Dickerson & Kemeny, 2004). Since cognitive disso-
nance likely involves a threat to the positive evaluation of
the self (Steele, Spencer & Lynch, 1993), it may implicate
the HPA axis. Moreover, this effect may be particularly pro-
nounced when self-threat is evoked by social evaluation (e.g.
for Asians in public-choice condition).
Present study
The current work explored the degree to which the HPA axis
is involved in cognitive dissonance within the free-choice
paradigm. Both European–American and Asian participants
made a choice either in private (i.e. in the absence of “social
eyes”) or in public (i.e. in the presence of “social eyes”). We
anticipated that European–Americans would show a reliable
SA primarily in the private (vs. public) choice condition
while Asians would show the effect in the public (vs. private)
choice condition. We assessed salivary cortisol levels prior
to, as well as 5 and 20 min after, the choice. Our effort to
examine HPA-axis activity as a function of culture and the
choice conditions was guided by two potentially competing
hypotheses regarding the relative strength of the physiologi-
cal effects associated with choice-induced dissonance arousal
and its reduction (i.e. SA).
First, if the stress generated by choice-induced dissonance
is sufficiently strong to activate the HPA axis and is not
mitigated by SA, we should observe increases in salivary
cortisol when self-threat is assumed to be occurring (e.g.
public-choice condition for Asians and in the private-choice
condition for European–Americans). Previous work examin-
ing SNS activity in dissonance-arousing conditions within
other paradigms (i.e. induced compliance) showed that the
Journal of Behavioral Decision Making
Copyright © 2015 John Wiley & Sons, Ltd. J. Behav. Dec. Making (2015)
DOI: 10.1002/bdm
SNS arousal accompanying dissonance is not always related
to subsequent attitude change (Croyle & Cooper, 1983;
Elkin & Leippe, 1986; Losch & Cacioppo, 1990; Elliot &
Devine, 1994). Given this, increased cortisol levels may be
expected in the critical conditions regardless of the magnitude
of dissonance reduction. Alternatively, the stress reducing effect
of SA may be quite potent. If this were the case, even though
dissonance is aroused, the stress involved should quickly dissi-
pate once the choice is justified. Cortisol levels may not be
necessarily high in conditions in which dissonance is aroused
among those who engage in SA. Thus, the magnitude of SA
may mitigate HPA-axis activity in the critical conditions.
METHODS
Participants
Thirty-seven Asian/Asian–American undergraduates (16 males,
21 females, M
Age
= 19.25) and 40 European–American
undergraduates (22 males, 18 females, M
Age
=18.95) from
the University of Michigan, Ann Arbor participated in
exchange for a course credit or $12. Exclusion criteria were
acute or chronic infections, acute allergy, and diseases of the
adrenal gland, regular use of any medication, gravidity, and
actual or past mental illness. The study was approved by the
Institutional Review Board at the University of Michigan,
Ann Arbor.
Procedure and materials
To control for the circadian rhythm of cortisol (Kudielka,
Schommer, Hellhammer & Kirschbaum, 2004), experimental
sessions were scheduled from 2 pm to 6:30 pm. Participants
received an e-mail with guidelines to follow during the 24 h
prior to their participation. This included not engaging in
daily activities that could influence basal cortisol levels such
as consuming dairy products during the 3 h prior to the ses-
sion or eating and drinking in the 30 min beforehand.
Upon arrival, the study was introduced as a consumer
marketing survey sponsored by a CD retailers’association
that was assessing the contributions of nature versus nurture
to music preferences. As shown in Figure 1, participants
were first told to relax for 20 min in a room in which various
magazines were present. At the end of this period, they
rinsed out their mouth with water in order to eliminate con-
taminates from blood and residuals from food. They then
completed a questionnaire assessing daily activities and gen-
eral health conditions that might affect basal cortisol levels.
Right after this (approximately 30 min after their arrival),
the first saliva sample (time 1) was collected to measure
baseline cortisol levels.
Participants then completed the standard free-choice dis-
sonance paradigm (Brehm, 1956; Steele et al., 1993) under
procedures that are expected to reduce artifacts due to mea-
surement error (e.g. giving participants a relatively small
number of options and sufficient time to rank them;
Kitayama et al., 2014). Following the procedures used by
Kitayama et al. (2004), participants were first randomly
assigned to one of two conditions: public-choice or private-
choice. In the public-choice condition, participants were un-
obtrusively exposed to a poster that included schematic emo-
tional faces whose gazes were directed towards the
participant. The poster was identical to that used by
Kitayama et al. (2004) and others (Na & Kitayama, 2012;
Imada & Kitayama, 2010) and has been found to be a valid
proxy for public performance (Na & Kitayama, 2012; Imada
& Kitayama, 2010). In the private-choice condition, no
poster was present.
After random assignment, participants’viewed a binder of
30 popular CDs that had been pretested to reflect college stu-
dents’current preferences and asked to select 10, at their own
pace, that they would like to have but did not already own.
They then ranked them by preference. Next, during an al-
leged music preference survey, the experimenter interrupted
participants to say that the company was offering them a free
CD but that there were only two left in stock, from which
they could pick one to keep. In each case, the two CDs were
the ones that participants had rank-ordered as their fifth and
sixth favorites. Five minutes after the choice, and 15 min af-
ter cort time 1, the participant provided a second saliva sam-
ple (cort time 2) while completing various filler tasks. Next,
participants were told that the company was interested in mu-
sic preferences when customers leave a store and thus, when
CDs are no longer visually present. Participants then re-
Figure 1. Timeline for the experimental procedure. Abbreviations: Cort 1 refers to Baseline Cortisol, Cort 2 was 5 min post-choice and Cort 3
was 20 min post-choice
Cortisol and DissonanceS. Y. Kimel et al.
Copyright © 2015 John Wiley & Sons, Ltd. J. Behav. Dec. Making (2015)
DOI: 10.1002/bdm
ranked the 10 CDs again according to their current prefer-
ences. They were encouraged to indicate how they felt at that
very moment, regardless of their previously reported prefer-
ences. At the end of the task, participants remained in the
lab relaxing and provided the final saliva sample (cort time
3) 20 min after the choice and 30min after cort time 1. Partici-
pants then completed a demographic questionnaire and, be-
fore being debriefed, they were asked whether they noticed
“anything unusual”in the room. However, no participant
reported any suspicion about the poster.
Salivary cortisol procedures
Participants spit into 50-ml polypropylene collection tubes
(Crystalgen Inc.) at the times previously described (see
Figure 1). To stimulate saliva flow, Trident Original Flavor
chewing gum was used, which has been shown to yield the
least bias in steroid levels as compared to unaided saliva flow
(Dabbs, 1991; see also Shirtcliff, Granger, Schwartz &
Curran, 2001). Immediately following each subject’s session,
saliva samples were stored in a freezer at approximately
20 °C. Samples were freed from mucopolysaccarides and
other residuals by three freeze–thaw cycles with subsequent
centrifugation, and later assayed for cortisol. Salivary cortisol
levels were determined by solid-phase
125
I radioimmunoassays
(Coat-A-Count, Diagnostic Products Corp. Biermann, Bad
Nauheim, Germany). Cortisol was measured using 400-μl
saliva samples in combination with water-diluted standards
(analytical range: 0.5 to 50ng/ml) and overnight incubation
at room temperature. Averaged across four assays, intra-assay
coefficients of variation for these samples were 4.16%. Inter-
assay coefficients of variation were 4.04% and 2.92% for low
and high pools, respectively. Moreover, inter-assays coeffi-
cients of variation were 10.84% and 4.66% for low and high
lypochecks, respectively. Sensitivity of the assay (B0-3 SD)
was at 0.37 ng/ml. This assay protocol was developed and
analytically validated by previous researchers (Schultheiss &
Stanton, 2009).
Analytic strategy
In line with previous work (Hoshino-Browne et al., 2005;
Steele et al., 1993), we defined Spread of Alternatives (SA)
as the increase in liking of chosen CD plus the decrease in
liking of rejected CD [(pre-choice rank of selected
CD post-choice rank of selected CD) + (post-choice rank
of rejected CD pre-choice rank of rejected CD)]. Consis-
tent with other researchers, ranking was used as the measure
of liking (e.g. Kitayama et al., 2004). We first used a 2 (Asian
vs. European-Americans) × 2 (public-choice vs. private-
choice) ANOVA and tested the effects of culture and condi-
tion on the magnitude of SA.
We then examined the effects of SA, culture, and condi-
tion on cortisol reactivity. In this analysis, we used a multi-
level growth curve modeling approach (GCM) via SAS
PROC MIXED with random slopes and intercepts. We used
GCM modeling instead of a Repeated Measures ANOVA
because GCM does not assume independence of cortisol
samples within individuals, allows us to model the correct
covariate matrix, and offers significantly more statistical
power when examining small samples (Gueorguieva, 2004;
Matthews, Altman, Campbell & Royston, 1990; Hruschka,
Kohrt & Worthman, 2005). Cortisol data was log-
transformed due to significant skewness.
RESULTS
Spread of alternatives (SA)
A 2 (culture: European-American, Asian) × 2 (condition:
public, private) ANOVA preformed on mean SAs showed
a significant interaction, F(1,72) = 7.62, p<.01. The pattern
is illustrated in Figure 2. As predicted, Asian participants
showed a reliable SA in the public-choice condition
(M= 1.52, SD = 1.67), t(18) = 3.96, p<.01, but not in the
private-choice condition (M= 0.11, SD = 2.37), t<1. In fact,
the SA was significantly greater in the former than in the
later, F(1,35 = 4.18, p<.05). In contrast, European–
Americans showed a reversed pattern. As predicted, they
showed a reliable SA in the private-choice condition
(M= 1.85, SD = 1.49), t(19) = 5.52, p<.001. Although the
SA in the public-choice condition was still sizable
(M= 1.10, SD = 1.04), t(18) = 4.59, p<.001, it was less,
albeit marginally, than the one observed in the private-choice
condition, F(1,37) = 3.20, p= .08. From a different angle, it is
evident that European–Americans showed a significantly
greater SA than Asians did during private choice, F(1,36)
= 7.45, p= .01. The pattern was non-significantly reversed
in the public-choice condition, F<1.
Cortisol stress response
We first tested several potential covariates on the stress
response using separate models. Preliminary analysis showed
that neither cortisol intercept nor diurnal slope was influ-
enced by gender, age, current oral contraceptive use, or men-
strual phase (among females). These variables were
subsequently dropped. The cortisol intercept decreased as a
function of the number of hours from wakening, β=0.09,
Figure 2. Spread of Alternatives (SAs) for European–American and
Asian participants in the private-choice and public-choice condi-
tions. Error bars represent standard errors of the means
Journal of Behavioral Decision Making
Copyright © 2015 John Wiley & Sons, Ltd. J. Behav. Dec. Making (2015)
DOI: 10.1002/bdm
t(62) = 3.71, p<.001, while increasing for those who had
eaten within the past 90 min, β= 0.20, t(74) = 1.97, p=.05.
These variables were entered as covariates in the subsequent
analysis.
Next, we examined whether there was significant within-
individual variability in an unconditional mean model to jus-
tify the examination of predictors. There was significant var-
iability of cortisol levels at both the intercept and the slope
(p<.001 for both parameters). We also examined condi-
tional linear and quadratic growth models of the cortisol tra-
jectory from time 0. The linear model was the best fit to the
data (linear model AIC = 44.5 vs. quadratic model
AIC = 46.3). We found a significant negative linear slope as
a function of time β=0.009, t(151) = 6.66, p<.001, indi-
cating that cortisol levels decreased linearly over the span of
30 min during the study. Thus, we subsequently tested pre-
dictors of the linear trend only.
In testing our primary predictions, we first tested whether
the cortisol slope would vary as a function of both culture
and condition. Our first hypothesis implies that, if the stress
generated by choice-induced dissonance is sufficiently strong
to activate the HPA axis and is not mitigated by SA, we
should observe increases in salivary cortisol in the public
(vs. private) choice condition for Asians and in the private
(vs. public) choice condition for European-Americans. How-
ever, neither of the main effects nor the interaction between
the two was significant (see Table 1).
Since we observed no impact of the critical conditions
(i.e. in the private-choice condition for European–Americans
and the public-choice condition for Asians) we could not
directly test our second hypothesis, that the effect of the
critical conditions would be partially mitigated by the effect
of SA. However, as shown in Figure 3, the cortisol level
steadily declined over the span of 30min during the experi-
mental procedure. This cortisol decline amounted to as much
as 30% from the peak response at a rate of .10μg/dl per hour.
This rate is nearly five-fold higher than expected from the di-
urnal decline (Ranjit, Young, Raghunathan & Kaplan, 2005)
and larger than might typically be expected for diurnal
change during the afternoon (Edwards, Clow, Evans &
Hucklebridge, 2001). Given that it is more consistent with
the rate observed during recovery to acute laboratory
stressors (see Yim, Quas, Cahill & Hayakawa, 2010;
Schwabe, Haddad & Schachinger, 2008), this decline likely
reflected a recovery from events that occurred prior to arriv-
ing at the laboratory (e.g. increased stress in anticipation of
the laboratory session). Therefore, we tested whether SA
would entail a greater cortisol decline. This analysis showed
asignificant effect of SA on the slope, F(1,124) = 3.90,
p=.05. Specifically, increases in SA were associated with
greater declines in the cortisol trajectory, b=.0013,
t(124) = 1.98, p= .05. Additional analysis that included
culture and condition showed that the effect of SA on
the slope was not qualified by culture, SA × culture × Time
F(1,120) = .30, p>.20, or private vs. public choice, SA x
Condition x Time F(1,120) = .02, p>.20.
Table 1. Individual growth curve models of cortisol (Log) trajectories during a cognitive dissonance task
Fixed effects βStd. error t
Model 1
Intercept 1.16 .176 6.62**
Eat (no) .216 .093 2.30*
Times since awaking .075 .022 3.28**
Time (slope) .010 .002 4.01**
Culture (Asian) effect on baseline .142 .133 1.07
Condition (Private) effect on baseline .226 .147 1.54
Culture × condition on baseline .261 .194 1.34
Culture effect on slope .001 .003 .55
Condition effect on slope .003 .003 .89
Culture × condition on slope .001 .004 .36
Model 2
Intercept 1.28 .155 8.28**
Eat (no) .217 .094 2.30*
Time since awaking .071 .022 3.20**
Time (slope) .007 .001 6.63**
SA .003 .100 .03
SA effect on slope .001 .001 1.98*
SA = spreading of alternatives.
Figure 3. Salivary cortisol levels in the free-choice dissonance
paradigm
Cortisol and DissonanceS. Y. Kimel et al.
Copyright © 2015 John Wiley & Sons, Ltd. J. Behav. Dec. Making (2015)
DOI: 10.1002/bdm
DISCUSSION
SA as a function of culture and social eyes
The primary goal of the current work was to investigate
whether HPA-axis activity would be involved in the pro-
cesses of choice-induced dissonance arousal and its reduc-
tion (i.e. SA). To set the stage for addressing this question,
we first identified specific conditions in which a substantial
self-threat and, thus, dissonance arousal would occur. Draw-
ing on previous work suggesting that private (vs. public)
choice is self-relevant for European–Americans whereas
public (vs. private) choice is more so for Asians (Kitayama
et al., 2004; Na & Kitayama, 2012), we found a significant
culture by condition interaction on SA. For European–
Americans, the SA was reliable in the private-choice condi-
tion (i.e. in the absence of “social eyes”) whereas, for Asians,
the SA was reliable in the public-choice condition (i.e. in the
presence of “social eyes”).
Two additional findings are relevant in interpreting the
HPA-axis activity findings. First, the SA was negligible
when Asians made a choice in private. This shows that, in
the current study, the error associated with the measurement
of preferences was negligible. Thus, the reliable SAs ob-
served in the other conditions are unlikely to be an artifact
produced by measurement error. Instead, they likely reflected
substantive psychological mechanisms. The observed pattern
is consistent with the hypothesis that a threat to the culturally
approved sense of self (i.e. independent for European–
Americans; interdependent for Asians) is instrumental in
fostering SA.
Second, our manipulation of public (vs. private) choice
involved no more than mere exposure to a set of schematic
faces that appeared to be watching each participant. At first
glance, the fact that such a subtle manipulation had substan-
tive psychological effects might seem surprising. Neverthe-
less, our finding is consistent with a growing body of
evidence that faces are processed both quickly and automati-
cally to produce a psychological readiness to cope with
the potentially important consequences of “being watched”
(Haley & Fessler, 2005; Na & Kitayama, 2012; Park &
Kitayama, 2014; Rigdon, Ishii, Watabe & Kitayama, 2009).
Moreover, because of the subtlety of this manipulation, de-
mand characteristics can be ruled out as a potential explana-
tion of our findings.
SA and cortisol stress response
We had two potentially competing hypotheses. Our first hy-
pothesis implies that cortisol levels should increase in condi-
tions in which self-threat is occurring (i.e. the private-choice
condition for European–Americans and the public-choice
condition for Asians). However, we did not find any evi-
dence for this hypothesis. Thus, the current data suggests
that, while the self-threat operative in the free-choice para-
digm is likely responsible for SA, it is not strong enough to
produce a significant HPA-axis stress response. Our second
hypothesis implies that SA should be associated with less
HPA activation only in conditions in which the choice is
self-threatening. While we observed a decline of cortisol as
a function of SA, contrary to our second prediction, this
was observed regardless of either culture or condition. There-
fore, our findings provided no support for our second
hypothesis, either.
Although our task did not result in the expected increase
in cortisol typical of laboratory stress tasks, the negative cor-
relation between SA and changes in cortisol levels during the
tasks suggests that justifying one’s choice is associated with
variability in “recovery”to events that occurred prior to ar-
riving to the laboratory (e.g. novelty of the laboratory visit).
This conjecture is based on the observation that cortisol
levels declined at a rate of 10 μg/dl per hour, which is in line
with the kinetics observed during the recovery to acute psy-
chosocial stressors (Yim et al., 2010; Schwabe et al., 2008).
The rate of decline and the total decline of nearly 30% of
maximum levels are also far greater than expected for diurnal
change that occurs during the afternoon (Edwards et al.,
2001; Ranjit et al., 2005). Thus, although there may be other
factors that led to such high rate of decline (e.g. pulsatility of
diurnal levels, post-wakening recovery among late sleepers;
Young, Abelson & Lightman, 2004), we suggest that SA
likely impacted factors that modulate the down-regulation
of the axis. This is consistent with research suggesting that
there are individual differences in dissonance reduction pro-
cesses (see Harmon-Jones, Amodio & Harmon-Jones,
2009). It is possible that greater SA reflects better coping
mechanisms or that other psychological factors (e.g. en-
hanced emotion regulation) facilitated the recovery to the
stress of entering the laboratory (or other sources of stress
leading to the decline). Our results may suggest then, that
greater justification of one’s choice facilitated this regulation
or, conversely, that failure to justify one’s choice elicited
enough stress to disrupt the down regulation of the HPA axis
during the study. Therefore, whether the observed effect re-
flects a recovery-facilitation by high SA or a recovery-
disruption by low SA, or both, is yet to be determined.
CONCLUSION
The contribution of the current work was two-fold. First, we
replicated previous behavioral findings on culture and
choice-induced dissonance by showing that SA occurs when
one’s cultural self is likely threatened (i.e. when European–
Americans make a private-choice and when Asians make a
public-choice). Second, we found that the relationship be-
tween this phenomenon and HPA-axis activity, as assessed
by salivary cortisol levels, is a complex and nuanced one.
Specifically, our data suggest that the self-threat involved
in choice-induced dissonance may be too subtle to produce
any evidence of HPA-axis stress reactivity. At the same time,
however, regardless of both culture and the choice being
private or public, the magnitude of SA was significantly
associated with the recovery of cortisol levels during the
visit, suggesting that either those who engage in greater
dissonance reduction following choice-making may have
enhanced coping strategies or that failure to engage in disso-
nance reduction disrupted the expected declining pattern
observed throughout the visit.
Journal of Behavioral Decision Making
Copyright © 2015 John Wiley & Sons, Ltd. J. Behav. Dec. Making (2015)
DOI: 10.1002/bdm
More research is required to clarify the neural and physio-
logical bases of cognitive dissonance. Existing evidence
suggests that the state of dissonance is likely neutrally repre-
sented in terms of certain brain regions such as the dorsal
anterior cingulate cortex (which is known to respond to cog-
nitive conflicts) and the anterior insula (which is known to
respond to negative arousal; Kitayama, Chua, Tompson &
Han, 2013). Future work should examine whether this neural
representation of dissonance may be linked to the activation of
neuroendocrine response systems including the HPA axis.
Although we did not find any evidence for HPA-axis
reactivity in the current study, such a link may exist in daily
conflicts that entail much stronger dissonance reactions or in
situations that involve higher levels of perceived social
evaluation (e.g. Curley, Yates & Abrams, 1986). Moreover,
participants in our study were likely still regulating to the
stress associated with coming to a laboratory study.
Although this is a common limitation of many studies and it
does not prevent HPA-axis activation to common laboratory-
based paradigms (see for example Abelson et al., 2014), it is
possible that elevated cortisol levels at the start of the task
could limit stress reactivity to a weak stressor. Therefore,
future work may benefit from a procedure that allows the
HPA axis to habituate more fully in the experimental setting
before dissonance is induced.
REFERENCES
Abelson, J. L., Erickson, T. M., Mayer, S. E., Crocker, J., Briggs,
H., Lopez-Duran, N. L., & Liberzon, I. (2014). Brief cognitive
intervention can modulate neuroendocrine stress responses to
the trier social stress test: Buffering effects of a compassionate
goal orientation. Psychoneuroendocrinology,44,60–70.
doi:10.1016/j.psyneuen.2014.02.016
Bjorklund, A., Hokfelt, T., & Owman, C., (2010). Sympathetic ner-
vous system. In George Fink (Ed.), Stress science:neuroendocri-
nology (Vol. 47, pp. 290). Oxford, UK: Academic Press.
Brehm, J. W. (2007). A brief history of dissonance theory. Social
and Personality Psychology Compass,1(1), 381–391.
Brehm, J. W. (1956). Post-decision changes in the desirability of
alternatives. Journal of Abnormal and Social Psychology,52
(2), 384–389.
Chen, M. K., & Risen, J. L. (2010). How choice affects and reflects
preferences: Revisiting the free-choice paradigm. Journal of
Personality and Social Psychology,99(4), 573–594.
Cooper, J., & Fazio, R. (1984). A new look at dissonance theory.
Advances in Experimental Social Psychology,17, 229–264.
Croyle, R. T., & Cooper, J. (1983). Dissonance arousal: Physiolog-
ical evidence. Journal of Personality and Social Psychology,45,
782–791.
Curley, S. P., Yates, J. F., & Abrams, R. A. (1986). Psychological
sources of ambiguity avoidance. Organizational Behavior and
Human Decision Processes,38(2), 230–256.
Dabbs, J. M. (1991). Salivary testosterone measurement:
Collecting, storing and mailing salivary samples. Physiology &
Behavior,49, 815–817.
Dickerson, S. S., & Kemeny, M. E. (2004). Acute stressors and
cortisol responses: A theoretical integration and synthesis of
laboratory research. Psychological Bulletin,130(3), 355–391.
Edwards, S., Clow, A., Evans, P., & Hucklebridge, F. (2001). Ex-
ploration of the awakening cortisol response in relation to diur-
nal cortisol secretory activity. Life Sciences,68(18), 2093–2103.
Elkin, R. A., & Leippe, M. R. (1986). Physiological arousal, disso-
nance, and attitude change: Evidence for a dissonance-arousal
link and a "don’t remind me" effect. Journal of Personality
and Social Psychology,51,55–65.
Elliot, A. J., & Devine, P. G. (1994). On the motivation nature of
cognitive dissonance: Dissonance as psychological discomfort.
Journal of Personality and Social Psychology,67, 382–394.
Festinger, L. (1957). A theory of cognitive dissonance. Evanston,
IL: Row, Peterson.
Gerard, H. B. (1967). Choice difficulty, dissonance, and the deci-
sion sequence. Journal of Personality,35,91–108.
Gueorguieva, R. (2004). Move over ANOVA. Archives of General
Psychiatry,61(3), 310.
Haley, K. J., & Fessler, D. M. T. (2005). Nobody’s watching? Evo-
lution and Human Behavior,26(3), 245–256. doi:10.1016/j.
evolhumbehav.2005.01.002
Harmon-Jones, E., Amodio, D. M., & Harmon-Jones, C. (2009).
Action-based model of dissonance: A review, integration, and
expansion of conceptions of cognitive conflict. Advances in Ex-
perimental Social Psychology,41, 119–166.
Harmon-Jones, E., & Harmon-Jones, C. (2009). Action-based
model of dissonance: A review of behavioral, anterior cingulate,
and prefrontal cortical mechanisms. Social and Personality Psy-
chology Compass,2(3), 1518–1538.
Heine, S. J., & Lehman, D. R. (1997). The cultural construction of
self-enhancement: an examination of group-serving biases.
Journal of Personality and Social Psychology,72(6), 1268.
Higgins, E. T., Rhodewalt, F., & Zanna, M. P. (1979). Dissonance
motivation: Its nature, persistence, and reinstatement. Journal
of Experimental Social Psychology,15,16–34.
Hoshino-Browne, E., Zanna, A. S., Spencer, S. J., Zanna, M. P.,
Kitayama, S., & Lackenbauer, S. (2005). On the cultural
guises of cognitive dissonance: The case of easterners and
westerners. Journal of Personality and Social Psychology,
89,294–310.
Hruschka, D. J., Kohrt, B. A., & Worthman, C. M. (2005). Estimat-
ing between- and within individual variation in cortisol levels
using multilevel models. Psychoneuroendocrinology,30(7),
698–714
Imada, T., & Kitayama, S. (2010). Social eyes and choice justifica-
tion: Culture and dissonance revisited. Social Cognition,28(5),
589–608.
Kimel, S. Y., Grossmann, I., & Kitayama S. (2012). When gift-
giving produces dissonance: Effects of subliminal affiliation
priming on choices for one’s self versus close others. Journal
of Experimental Social Psychology,48(5), 1221–1224.
Kitayama, S., Chua, H. F., Tompson, S., & Han, S. (2013). Neural
mechanisms of dissonance: An fMRI investigation of choice jus-
tification. NeuroImage,13, 206–212.
Kitayama, S., Snibbe, A. C., Markus, H. R., & Suzuki, T. (2004). Is
there any "free" choice? self and dissonance in two cultures.
Psychological Science,15(8), 527–533.
Kitayama, S., Tompson, S., & Chua, H. F. (2014). Cultural
neuroscience of choice justification. In J. Forgas, & E.
Harmon-Jones (Eds.), Control within: Motivation and its regula-
tion (pp. 313–330). Sydney Symposium: Psychology Press.
Kudielka, B. M., Schommer, N. C., Hellhammer D. H., &
Kirschbaum, C. (2004). Acute HPA axis responses, heart
rate, and mood changes to psychosocial stress (TSST) in
humans at different times of day. Psychoneuroendocrinology,
29,983–992.
Losch, M. E., & Cacioppo, J. T. (1990). Cognitive dissonance may
enhance sympathetic tonus, but attitudes are changed to reduce
negative affect rather than arousal. Journal of Experimental
Social Psychology,26, 289–304.
Lovallo, W. R., & Thomas, T. L. (2000). Stress hormones in psy-
chophysiological research: emotional, behavioral, and cognitive
implications. In J. T. Cacioppo, L. G. Tassinary, & G.G.
Bernston (Eds.) Handbook of psychophysiology (Second ed.,
pp. 342–367). New York: Cambridge University Press.
Cortisol and DissonanceS. Y. Kimel et al.
Copyright © 2015 John Wiley & Sons, Ltd. J. Behav. Dec. Making (2015)
DOI: 10.1002/bdm
Mason, J. W. (1968). A review of psychoendocrine research on the
pituitary–adrenal corticalsystem. Psychosomatic Medicine,30,
576–607.
Markus H. R., & Kitayama, S. (1991). Culture and the self: Implica-
tions for cognition, emotion and motivation. Psychological Re-
view,98, 224–253.
Matthews, J. N., Altman, D. G., Campbell, M. J., & Royston, P.
(1990). Analysis of serial measurements in medical research.
British Medical Journal,300(6719), 230–235.
Na, J., & Kitayama, S. (2012). Will people work hard on a task they
choose? Social-eyes priming in different cultural contexts. Jour-
nal of Experimental Social Psychology,48(1), 284–290.
Park, J., & Kitayama, S. (2014). Interdependent selves show face-induced
facilitation of error processing: Cultural neuroscience of self-threat.
Social Cognitive and Affective Neuroscience,9(2), 201–208.
Ranjit, N., Young, E. A., Raghunathan, T. E., & Kaplan, G. A.
(2005). Modeling cortisol rhythms in a population-based study.
Psychoneuroendocrinology,30, 615–624.
Rigdon, M., Ishii, K., Watabe, M., & Kitayama, S. (2009). Minimal
social cues in the dictator game. Journal of Economic Psychology,
30,358–367.
Schultheiss, O. C., & Stanton, S. J. (2009). Assessment of sali-
vary hormones. In E. Harmon-Jones, & J. S. Beer (Eds.),
Methods in the neurobiology of social and personality psy-
chology (pp. 17–44). New York, NY: Guilford.
Schwabe, L., Haddad, L., & Schachinger, H. (2008). HPA axis ac-
tivation by a socially evaluated cold-pressor test. Psychoneuro-
endocrinology,33, 890–895.
Shirtcliff, E. A., Granger, D. A., Schwartz, E., Curran, M. J. (2001). Use
of salivary biomarkers in bio-behavioral research: Cotton-based
sample collection methods can interfere with salivary immunoassay
results. Psychoneuroendocrinology,26(2), 165–173.
Steele, C. M. (1988). The psychology of self-affirmation: Sustaining
the integrity of the self. Advances in Experimental Social Psy-
chology,21(2), 261–302.
Steele, C. M., Spencer, S. J., & Lynch, M. (1993). Self-image
resilience and dissonance: The role of affirmational
resources. Journal of Personality and Social Psychology,
64(6), 885–896.
Tsigos, C., & Chrousos, G. P., (2002). Hypothalamic-pituitary-
adrenal axis, neuroendocrine factors and stress. Journal of
Psychosomatic Research,53(4), 865–871.
Yim, I. S., Quas, J. A., Cahill, L., & Hayakawa, C. M. (2010). Chil-
dren’s and adults’salivary cortisol responses to an identical psy-
chosocial laboratory stressor. Psychoneuroendocrinology,35,
241–248.
Young, E. A., Abelson, J. L., & Lightman, S. L. (2004). Cortisol
pulsatility and its role in stress regulation and health. Frontiers
in Neuroendocrinology,25,69–76.
Authors’biographies:
Sasha Kimel is a post-doctoral scholar at Harvard University. She
earned her PhD in Social Psychology from University of Michigan.
Her work focuses on understanding the psychological underpin-
nings of interethnic conflict.
Nestor Lopez-Duran is an Assistant Professor in Clinical Psycho-
logy at University of Michigan, where he also received his PhD. His
work focuses on understanding of the role of stress regulation in
childhood mood disorders.
Shinobu Kitayama is a Professor in Social Psychology at Univer-
sity of Michigan, where he also received his PhD. His current work
focuses on (i) gene × culture interactions, (ii) culture and the self,
and (iii) culture and biological health.
Authors’addresses:
Sasha Y. Kimel, Harvard University, Cambridge, MA, USA
Nestor Lopez-Duran and Shinabu Kitayama, University of
Michigan, Ann Arbor, MI, USA
Journal of Behavioral Decision Making
Copyright © 2015 John Wiley & Sons, Ltd. J. Behav. Dec. Making (2015)
DOI: 10.1002/bdm
