Cognitive emotion regulation fails the stress test
Candace M. Raioa, Temidayo A. Orederub, Laura Palazzoloc, Ashley A. Shurickd, and Elizabeth A. Phelpsa,e,f,1
aPsychology Department andeCenter for Neural Science, New York University, New York, NY 10003;bPsychology Department, Hunter College, New York,
NY 10065;cState University of New York Downstate College of Medicine, Brooklyn, NY 11203;dPsychology Department, Stanford University, Stanford,
CA 94305; andfEmotional Brain Institute, Nathan Kline Institute for Psychiatric Research, Orangeburg, NY 10962
Edited by Bruce S. McEwen, The Rockefeller University, New York, NY, and approved July 9, 2013 (received for review March 29, 2013)
Cognitive emotion regulation has been widely shown in the
laboratory to be an effective way to alter the nature of emotional
responses. Despite its success in experimental contexts, however,
we often fail to use these strategies in everyday life where stress
is pervasive. The successful execution of cognitive regulation relies
on intact executive functioning and engagement of the prefrontal
cortex, both of which are rapidly impaired by the deleterious
effects of stress. Because it is specifically under stressful conditions
that we may benefit most from such deliberate forms of emotion
regulation, we tested the efficacy of cognitive regulation after
stress exposure. Participants first underwent fear-conditioning,
where they learned that one stimulus (CS+) predicted an aversive
outcome but another predicted a neutral outcome (CS−). Cogni-
tive regulation training directly followed where participants were
taught to regulate fear responses to the aversive stimulus. The
next day, participants underwent an acute stress induction or
a control task before repeating the fear-conditioning task using
these newly acquired regulation skills. Skin conductance served as
an index of fear arousal, and salivary α-amylase and cortisol con-
centrations were assayed as neuroendocrine markers of stress re-
sponse. Although groups showed no differences in fear arousal
during initial fear learning, nonstressed participants demonstrated
robust fear reduction following regulation training, whereas
stressed participants showed no such reduction. Our results sug-
gest that stress markedly impairs the cognitive regulation of emo-
tion and highlights critical limitations of this technique to control
affective responses under stress.
intensely with a loved one, or having a rough day at work,
controlling our emotions when circumstances become stressful
can be challenging. Although extensive work in the laboratory
has demonstrated that we can cognitively alter emotional
responses to foster more adaptive behavior (1–3), in real-world
emotional contexts we often fail to do so. One potential reason
for this regulatory failure might be that the pervasive presence of
stress in daily life compromises our ability to effectively regulate
emotions. Indeed, negative affect has long been proposed to play
a key role in the failure to exert self-regulatory control over our
thoughts and behavior (4, 5). However, as of yet, a direct re-
lationship between the physiological stress response and the
cognitive control of emotion has not been examined. Here, we
sought to explore how the cognitive regulation of emotion is
affected by an acute stress induction.
A large body of work has shown that responses to emotionally
salient stimuli can be flexibly changed and controlled through
cognitive emotion regulation (for review, see refs. 1–3 and 6). By
targeting what has been described to be the initial appraisal of
a salient cue or event (7, 8), cognitive regulation allows an in-
dividual to alter the relevance and meaning of a stimulus, sub-
sequently shaping its emotional response (2, 3). Recruiting
cognitive strategies to deliberately change the way a stimulus is
evaluated, either by reinterpreting (i.e., reappraising) its mean-
ing or focusing on its more positive aspects, has proven effective
at reducing the subjective (2, 9, 10), physiological (2, 10, 11), and
neural components (3, 9, 11, 12) of emotional arousal.
The cognitive regulation of emotion, however, is generally a
complex and goal-directed process that depends on a number of
higher cognitive functions, such as attention, cognitive flexibility,
hether we are running late to an appointment, arguing
motivation, and working memory, which all facilitate the online
maintenance of information needed to override initial affective
reactions (2, 3, 13, 14). This regulatory capacity is critical to
mental (15) and physical (16) health and its impairment strongly
predicts vulnerability to an array of affective disorders (17, 18).
Importantly, the principles underlying cognitive regulation also
form the basis of cognitive-behavioral therapy (CBT), a tool
widely used in the clinic to treat affective psychopathology. Like
cognitive regulation, CBT relies on the tightly coupled relation-
ship between thoughts and emotions and promotes the correction
of irrational or distorted cognitive appraisals to engender more
adaptive emotional responses (19). Although cognitive regulation
has emerged as a highly effective technique for controlling emo-
tional responses, its success relies on the availability of cognitive
resources and intact executive function (3–5, 7, 13, 14). Critically,
a growing body of work has revealed that exposure to acute stress
impairs many of these higher cognitive processes (20–22), in-
cluding cognitive flexibility (23, 24), goal-directed behavior (25),
working memory (26–30), and self-control (5). These rapid cogni-
tive effects of stress are thought to be mediated by neuroendocrine
responses to acute stress exposure that impacts the functional
integrity of the prefrontal cortex (PFC) (20, 21), which supports
these processes (31, 32). Importantly, acute stress effects appear
to preferentially target the dorsolateral PFC (20, 21, 33, 34), which
has consistently been implicated as playing a key role in the suc-
cessful execution of cognitive emotion regulation (3, 6, 12).
These findings suggest an important, yet unexplored, paradox:
in the stressful situations in which we might benefit most from
deliberate, active forms of emotional control, the mechanisms
required to support such cognitive regulation may be impaired.
Thus, cognitive regulation may be ineffective at controlling
emotional responses precisely when such control is needed most.
Here, we test whether acute stress influences the ability to use
cognitive regulation to diminish conditioned fear responses.
Given the pervasive nature of stress in daily life, characterizing
how stress influences our ability to modify emotional responding
is critical for understanding the boundaries within which existing
regulation techniques are effective, and offer insight into treatment
options for those who suffer from stress-related psychological
We investigated the effect of stress on cognitive emotion
regulation using a 2-d protocol. On day 1 (learning session),
participants underwent a fear-conditioning paradigm using visual
cues as conditioned stimuli (CS) and a mild electric wrist-shock
as the unconditioned stimulus (US). Skin conductance response
(SCR) served as an index of physiological fear arousal. One
image (CS+) was paired with a shock on a subset of trials, but the
other (CS−) was never paired with shock and served as a base-
line measure of arousal. Directly after fear-conditioning, par-
ticipants reported three emotions that they associated with each
CS and rated the intensity of these emotions on a scale from 1
(least intense) to 10 (most intense). Participants were then
C.M.R., A.A.S., and E.A.P. designed research; C.M.R., T.A.O., and L.P. performed research;
C.M.R., T.A.O., and L.P. analyzed data; and C.M.R. and E.A.P. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
1To whom correspondence should be addressed. E-mail: email@example.com.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
www.pnas.org/cgi/doi/10.1073/pnas.1305706110PNAS Early Edition
| 1 of 6
trained to use a cognitive regulation strategy that incorporated
elements of reappraisal and was based on the principles of CBT.
This regulation strategy has been shown in previous work to
result in the persistent attenuation of conditioned fear com-
pared with sustained fear seen in those participants who un-
derwent a control task (10) (Methods). After cognitive regulation
training (CRT), participants rerated the intensity of their self-
reported emotions and were asked to return the next day to un-
dergo the same fear-conditioning task using their newly acquired
On day 2 (regulation session), participants who showed ade-
quate fear-learning the previous day (Methods) were randomly
assigned to the stress or control group. Critically, we manipu-
lated stress levels by having participants undergo either the cold-
pressor (CP) task (35), an acute stress induction in which par-
ticipants submerged their arms in ice-cold water for 3 min, or
a control task using room-temperature water, before repeating
the fear-conditioning session. To ensure that our stress manip-
ulation elicited hypothalamic-pituitary-adrenal (HPA) axis ac-
tivity, the hallmark of an acute stress response, we measured
cortisol concentrations in participants’ saliva throughout the
experiment. HPA-axis activation triggers the release of stress
hormones that peak ∼10–20 min after a stressor (20, 36–39).
Therefore, we collected salivary samples at baseline (immedi-
ately before the CP/control task), again 10 min later (after the
CP/control task but before the fear-conditioning task), and,
finally, 20 min after the CP/control task (immediately after the
fear-conditioning task) (see Fig. 1 for a schematic of experi-
mental protocol). Cortisol release is preceded and triggered by
an earlier wave of catecholamines (i.e., noradrenaline) that
reflects autonomic nervous system arousal and is released rapidly
after a stressor to facilitate preparatory responses to stress
(20, 21, 39). To measure this response, we also assayed α-amylase,
a salivary enzyme that serves as an index of noradrenergic ac-
tivity (40–42) in each of these samples. Participants reported
subjective levels of stress after the CP/control task on a scale
from 1 (least stressful) to 10 (most stressful). Self-reported
emotions and intensity ratings were collected on day 2 after
the CP/control manipulation, but before participants repeated
the fear-conditioning task. We hypothesized that after undergoing
CRT nonstressed participants would successfully regulate fear
to the CS+ on day 2 (10), whereas fear responses in stressed
participants would persist.
Physiological Fear Responses. To ensure that we could measure
the influence of stress on fear regulation the following day,
only participants who showed evidence of acquiring a condi-
tioned fear response (CS+ > CS− by 0.1 μS) were included
(see Methods for detailed exclusion criteria). To confirm that
fear acquisition on day 1 did not differ between conditions,
we conducted a repeated-measures ANOVA using a within-
subject factor of CS (CS+, CS−), and a between-subjects factor
of condition, using mean SCRs from the fear-conditioning ses-
sion. As expected, given our exclusion criteria, we found a sig-
nificant main effect of CS [F(1, 78)= 190.20, P < 0.000001].
Importantly, there was no effect of condition and no interaction.
Independent t tests confirmed that participants in both groups
showed greater mean SCRs to the CS+ than CS− [stress: t(35)=
9.24, P < 0.000001; control: t(43)= 10.24, P < 0.0000001] and that
fear responses did not differ between groups for either CS [CS+:
t(78)= −0.28, P = 0.78; CS−: t(78)=0.62 P = 0.54].
Our primary analysis of interest was how fear responses to the
CS+ decreased across sessions for each condition. We conducted
a repeated-measures ANOVA using condition (stress, control)
as a between-subject factor and session (day 1, day 2) as a within-
subject factor. We observed a main effect of session [F(1, 78)=
8.45, P = 0.005], no effect of condition [F(1, 78)= 1.32, P = 0.25],
and a significant session X condition interaction [F(1, 78)= 4.55,
P = 0.03]. Paired samples t-tests confirmed that the control group
showed a significant decrease in SCR to the CS+ across sessions
[t(43)= 3.65, P = 0.001], whereas the stress group showed no such
reduction [t(35)= 0.546, P = 0.59]. Mean SCR to the CS+ did not
differ between groups on day 1 [t(78)= −0.28, P = 0.78, one-
tailed]; however, on day 2 stressed participants showed signifi-
cantly stronger SCR to the CS+ than did the control group,
despite both groups undergoing the regulation training [t(78)=
−1.76, P = 0.04, one-tailed] (Fig. 2). An additional analysis of
our baseline stimulus (CS−) yielded no interaction or group
differences across sessions (Fig. S1 and SI Results).
Self-Reported Fear Responses. Subjective reports of emotional
experience were also consistent with our physiological findings.
We examined the influence of stress on self-reported fear by
assessing the proportion of fear-related emotions that participants
assigned to the CS+ across sessions (Methods). A session X con-
dition repeated-measures ANOVA revealed a main effect of
session [day 1, day 2; F(1, 78)= 28.53, P = 0.0000002], a marginally
significant main effect of condition [stress, control; F(1, 78)= 3.94,
P = 0.050], and a trend toward an interaction [F(1, 78)= 2.57,
P = 0.11]. Planned comparisons confirmed that the proportion
of fear-related words assigned to the CS+ were equivalent after
fear conditioning on day 1 [t(78)= −0.372, P = 0.71], verifying
that both groups initially considered the CS+ equally fearful. On
day 2, although both groups assigned fewer fear-related words
overall to the CS+ [stress: t(35)= 2.38, P = 0.02; control: t(43)= 5.44,
P = 0.00001], stressed participants reported a higher number of
fear-related emotions for the CS+ than did controls [t(78)= −2.57,
P = 0.01], indicating that stress influenced self-reported fear in
addition to physiological fear arousal (Fig. 3). The stress group
not only assigned significantly more fear-related emotions to
the CS+ on day 2, but the average intensity rating for those
perimental procedure and timeline of neuroendo-
crine assessments for both day 1 (learning) and
day 2 (regulation).
Experimental timeline. Schematic of ex-
2 of 6
| www.pnas.org/cgi/doi/10.1073/pnas.1305706110 Raio et al.
self-reported fear emotions was also marginally higher (Fig. S2
and SI Results).
Stress Analyses. Self-reported stress. As expected, stressed partic-
ipants rated the CP task with significantly higher levels of dis-
comfort than those participants who underwent the control
manipulation [t(76)= 10.42, P < 0.00001, two-tailed] (Fig. S3). To
assess whether these subjective ratings translated to larger
increases in cortisol after the CP/control task, we conducted
a correlation analysis between increases in cortisol relative to
baseline and self-reported stress ratings. We found a positive cor-
relation between the increase in cortisol and stress ratings both 10
min (r = 0.43, P = 0.0002) and 20 min (r = 0.48, P = 0.00001) after
the CP/control task. (Fig. S4). This relationship did not emerge
between increases in α-amylase and stress ratings at either time-
point (+10 min: r = 0.13, P = 0.28; +20 min: r = 0.04, P = 0.68). No
relation was found between these ratings and subsequent fear
arousal responses (i.e., SCR) to either the CS+ (r = 0.14; P = 0.20),
or the CS− (r = −0.04, P = 0.72).
Neuroendocrine responses. Our two neuroendocrine measures of
stress response included cortisol, a reliable measure of HPA-axis
activity, and α-amylase, which reflects noradrenergic response.
Groups did not differ in α-amylase or cortisol concentrations on
day 1 (SI Results). On day 2 we examined how cortisol concen-
trations differed between groups to confirm the efficacy of our
stress manipulation. (Any samples that contained insufficient
saliva could not be analyzed and were excluded from this anal-
ysis.) A repeated-measures ANOVA with a between-subject
factor of condition (stress, control) and within-subject factor of
time (baseline, +10 min, +20 min) revealed a main effect of time
[F(2, 134)= 5.16, P = 0.007] and condition [F(1, 67)= 16.36, P =
0.0001], as well as a time X condition interaction [F(2, 134)=
13.23, P = 0.000005]. Independent samples t tests confirmed that
cortisol levels differed between groups at each time-point mea-
sured after the CP/control task [+10 min: t(69)= −4.19, P =
0.00008; +20 min: t(72)= −4.74, P = 0.00001], but not at baseline.
Only the stress group showed significantly higher cortisol com-
pared with baseline 10 min [t(34)= −3.69, P = 0.001] and 20 min
[t(34)= −3.04, P = 0.005] after the stressor. Participants in the
control condition showed no cortisol change 10 min after the
control task [t(34)= 1.62, P = 0.11], and demonstrated a significant
decrease in cortisol relative to baseline 20 min after the task
[t(37)= 5.93, P = 0.0000007] (Fig. 4). A similar analysis using
alpha-amylase concentrations yielded no group differences or
interaction (SI Results).
Neuroendocrine responses and fear regulation. To explore whether in-
creases in glucocorticoids influenced fear regulation on day 2, we
conducted a linear regression using participants’ mean cortisol
level 10 and 20 min after the CP/control task as independent
variables and physiological fear arousal to the CS+ (i.e., SCR) as
our dependent variable. We found no relation between cortisol
increase and fear arousal responses across participants either
10 min (β = −0.03, P = 0.77), or 20 min (β = 0.09, P = 0.43),
after the CP/control manipulation.
We then examined whether α-amylase levels, indicative of
noradrenergic activity, were related to fear regulation responses
across participants by using participants’ mean α-amylase level
after the CP/control manipulation. Because α-amylase is char-
acterized by a rapid onset and exerts effects more quickly than
does cortisol (36–38), we hypothesized that only α-amylase
samples taken 10 min after the CP/control task (but not 20 min
after) would potentially influence fear arousal. We conducted
a linear regression using participants’ mean α-amylase level both
10 and 20 min after the CP/control task on mean SCR during the
regulation session. α-Amylase levels assessed 10 min after the
CP/control task predicted fear arousal responses across partic-
ipants (β = 0.25, P = 0.03) (Fig. 5). However, for α-amylase levels
20 min after the CP/control task, this correlation was only
trending (β = 0.17, P = 0.13).
These results suggest that acute stress impairs the ability to recruit
cognitive regulation to control fear responses to aversive cues.
Although both groups underwent CRT following initial fear
acquisition, only the nonstressed group was able to successfully
use these regulation techniques to diminish fear arousal on a
subsequent test. In contrast, fear responses in the stress group
were comparable to those during the initial fear acquisition, before
any regulation training.
Although our data cannot speak to the precise neural mech-
anisms underlying these fear regulation impairments, we can
take advantage of our understanding of the neurobiology of the
stress response to hypothesize as to why these effects might
emerge. Regulated fear responses on day 2 were correlated with
salivary α-amylase, a marker of noradrenergic activity. This
finding suggests that early catecholamine responses driven by
sympathetic nervous system arousal may be one mechanism by
which cognitive control over fear responses is impaired. This
finding is consistent with work both in animals (20, 21, 29) and
humans (23, 27, 29, 30, 33, 43) showing that elevated noradren-
aline levels during stress lead to rapid alterations in executive
functioning and impair the PFC. Indeed, a recent pharmacological
study conducted in humans showed that exogenously adminis-
tering a noradrenergic antagonist reduced stress-related neural
each group across sessions. On day 1, groups demonstrated equivalent levels
of SCR. On day 2, fear arousal to the CS+ was successfully diminished for the
control group only; no such reduction was shown in the stress group. *P <
0.05; **P < 0.01; error bars denote SEM.
Conditioned fear response for each group. Mean SCR for the CS+ for
sessions. Groups did not differ on day 1 before regulation training; however,
after the CP/control manipulation on day 2, stressed participants reported
a higher proportion of fear-related words for the CS+ than did controls.
Both groups significantly reduced the proportion of fear-related words
reported across sessions. *P < 0.05, **P < 0.00001; error bars denote SEM.
Average proportion of fear-related words assigned to the CS+ across
Raio et al. PNAS Early Edition
| 3 of 6
responsiveness and interconnectivity, whereas a cortisol an-
tagonist had no such effect (35). The finding is also consistent
with clinical investigations showing successful treatment of post-
traumatic stress disorder symptoms after administration of prazosin,
which blocks α-1-adrenoreceptor activity, enhancing prefrontal
functioning and dampening amygdala activity (44, 45).
We did not observe a similar correlation between cortisol and
fear responses during regulation. The elevations we show in
cortisol are indicative of HPA-axis activity and are widely known
to facilitate an organism’s recovery from a stressor, but cortisol
can also influence prefrontal function by potentiating earlier
catecholamine release (20, 29, 46). It has been suggested that
cortisol may preserve or exaggerate the effects of catecholamines
by inhibiting their clearance from the PFC, thus altering their
brain concentration (20, 29, 33, 46). Although only α-amylase
levels correlated with fear arousal 10 min after the stressor, it is
possible that the lasting elevated cortisol responses in the stress
group exacerbated the effects of these noradrenergic inputs in
the brain, allowing their effects on PFC function to persist even
after detectable differential salivary levels of alpha amylase be-
tween the groups dissipated. Therefore, it is possible that both
noradrenergic and glucocorticoid responses to stress, and the
interacting influence they exert on one another in the brain,
serve as a potential mechanism for the impact of stress on the
cognitive control of fear. Nonetheless, a clearer understanding of
the neural mechanisms that underlie this stress-induced cognitive
regulation impairment requires further study.
Although preliminary evidence has shown that individuals
with high trait anxiety are impaired at regulating emotions
(47), this investigation is unique in explicitly manipulating stress
levels in healthy humans during a cognitive emotion regulation
task, precisely when this capacity may be compromised. The
regulatory deficits described here are consistent with a growing
body of evidence suggesting that stress impairs cognitive control
and flexibility, presumably by disrupting PFC functioning (20,
21). It is also consistent with theories of self-regulation failure,
most notably, those derived from Baumeister and Heatherton
et al. (4, 5) that describe self-regulatory capacities as a limited
resource that may become weak and depleted either over time
(48) or when exposed to negative emotions (4, 5). This influential
model of self-regulation asserts that regulatory capacities rely on
top-down prefrontal control and may be weakest when the PFC
is impaired, or when subcortical regions involved in the auto-
matic emotional response behavior are enhanced (5). Our find-
ings are consistent with this model in a broader context of self-
regulation failure and provide a unique demonstration of one
underlying reason such self-regulation failure may be common in
Considering that situations that require such deliberate effort
to control emotions are typically those characterized by stress,
the present results highlight important limitations on the use of
cognitive strategies to change emotional responses, suggesting that
such strategies may be largely ineffective in the face of stress.
Asserting control over our emotional responses to subsequently
alter behavior has been of considerable interest to a number of
research domains. The stress-induced impairments seen here
highlight significant limitations in the real-world application
of cognitive regulation in the domains of reward processing
(4, 5, 49, 50), decision-making (51–53), and intergroup atti-
tude and bias change (54, 55), all of which have effectively
used cognitive regulation to change or influence behavior in
That cognitive emotion regulation techniques are rendered in-
effective under stress has broad implications for the efficacy of
cognitive regulation to change behavior in everyday life. Impor-
tantly, these results offer insight into why strategies taught in the
clinic may not always generalize to the real world, where stress
exposure is ubiquitous. Alternative forms of emotion regulation
that are less reliant on the PFC may be more suitable for changing
emotion responses under stress (6). Additionally, under longer
durations of training or practice, the recruitment of cognitive
regulation strategies might become easier and more habitual,
thus relying less on the top-down cognitive control and executive
functioning that are compromised by stress (56). Furthering our
understanding of how emotion regulation malfunctions under
stress may lead to better interventions that foster resistance to
stress-induced regulatory impairments and offer better treatment
options for clinical populations.
Participants. Participants were ineligible for the study if they were pregnant,
had experienced heart or blood pressure problems in the past, or were taking
any antidepressant or antianxiety medication. All participants signed a con-
sent form approved by New York University’s Committee on Activities In-
volving Human Subjects and were compensated $30 for their participation.
Participants were assigned either two images of spiders or two images of
snakes to serve as the CS+ and CS−. To ensure participants were not phobic
of these stimuli, they completed the Snake Phobic Questionnaire (SNAQ)
and Spider Phobic Questionnaire (SPQ) (57) on-line before participating in
the study. Those who scored above 15 on either of these questionnaires
(scores range from 0 to 31; higher score indicates increased fear) were
considered “phobic” and were thus ineligible to participate. This cut-off is
the CP/control task. *P < 0.01; **P < 0.001; error bars denote SEM.
Mean cortisol levels at baseline, as well as 10 min and 20 min after
responses to the CS+ during the regulation session as a function of α-amy-
lase levels 10 min after the CP/control manipulation.
α-Amylase predicts regulated fear arousal. Mean fear arousal
4 of 6
| www.pnas.org/cgi/doi/10.1073/pnas.1305706110Raio et al.
less than that which has been designated for phobic individuals (SNAQ: M =
24.44, SD = 2.95; SPQ: M = 23.76; SD = 3.8) (58). The stimulus category with
the higher score was used to ensure the stimuli were emotionally arousing.
Participants included in our final analysis had a mean SNAQ and SPQ score of
8.4 (SD = 6.68) and 6.95 (SD = 5.44), respectively, both of which are com-
parable to that of healthy populations (57).
Because SCRsserved as our index of fear arousal, participants who failed to
show measurable electrodermal signal to the shock (US) or on >75% of
nonreinforced CS+ trials were classified as nonresponders and excluded
before the second session (n = 25). Participants who failed to demonstrate
adequate differential conditioning (mean CS+ > mean CS− by at least 0.1 μS)
across the acquisition session were categorized as nonacquirers and ex-
cluded before the second session (n = 44). This criterion was necessary be-
cause we could not assess fear regulation without first confirming that
participants acquired differential fear to the CS+ vs. CS−; similar criteria
have been used in previous fear-conditioning studies (10, 59). Four partic-
ipants were excluded because of technical errors with data acquisition
software and two were excluded for failing to follow experimental instruc-
tions. Additionally, four participants from the stress group were unable to
complete the CP task and were thus excluded. Two final participants were
removed from the stress group before analysis because their cortisol con-
centrations at baseline were greater than 3 SDs from the mean. Our final
analysis included a total of 78 healthy participants (39 females) with a mean
age of 23.2 (SD = 8.18; range = 18–54).
Fear Conditioning. A delay, discrimination fear-conditioning procedure with
partial reinforcement was used. CSs were either two distinct images of
snakes or two distinct images of spiders (Fig. S5). CS category assignment
(i.e., snakes vs. spiders) was counterbalanced across participants, as was the
assignment of images within each category as CS+ or CS−. These “prepared
stimuli” were used because they are emotionally engaging without eliciting
fear, as verified by the SNAQ and SPQ. Two distinct, pseudorandomized
trial orders were used such that no trial of the same kind occurred more
than two times consecutively; these trial orders were also counterbalanced
A mild electric wrist-shock (200 ms) served as the US. One image (CS+)
coterminated with the US on a subset of trials, whereas the other image
(CS−) was never paired with shock (42 trials total; 17 CS+, 17 CS−, 8 CS-US
pairings). Partial reinforcement (∼33%) was used to allow unreinforced trials
to be analyzed for anticipatory fear arousal without contamination from US
responses. On each trial, the CS was presented for 4 s, followed by an 8- to
12-s intertrial interval, during which a fixation cross was presented at the
center of the screen. Participants were instructed to pay attention to the
relationship between the image and shock, and that this relationship would
be discussed with the experimenter afterward. This fear-conditioning pro-
tocol was repeated on day 2 (regulation phase) with instructions to use the
CRT from the previous day.
Psychophysiological Stimulation and Assessment. Mild electric shocks (50
pulses per second) were delivered through a bar electrode (Biopac Systems)
attached with a Velcro strap to participants’ wrist. The electrode wells were
manually filled with NaCl electrolyte gel to enhance conductance and at-
tached to a SD9 Square Pulse Stimulator (Grass Technologies). To identify
each individual’s shock level, mild shocks were manually triggered and in-
creased in increments of 5 V until the participant reported the shock to be
uncomfortable, but not painful, or 60 V were reached. Our index of fear
arousal was SCR, an assay of sympathetic nervous system arousal that reflects
changes over discrete intervals consistent with the autonomic arousal charac-
teristic of fear (12, 60). To record SCR, shielded Ag-AgCl electrodes filled with
standard NaCl electrolyte gel were first applied between the first and second
phalanges of the participant’s left index and middle fingers. SCR was sampled
at a rate of 200 Hz and recorded using an MP100 Data Acquisition module
(Biopac Systems) connected to an Apple computer. Raw SCR amplitudes were
square root transformed to reduce skew and were subsequently divided by
individual mean US responses to account for individual differences in shock
reactivity (61). We assessed conditioned responding off-line using Acq-
Knowledge software (Biopac Systems) by analyzing unreinforced trials
only. The level of skin conductance was assessed for each trial by taking
largest base-to-peak waveform amplitude response (in microsiemens, μS)
within the 0.5- to 4.5-s interval after stimulus onset. Responses lower than a
predetermined criterion of 0.02 μS were recorded as zero. Each individual’s
SCR data were preprocessed using AcqKnowledge software (Biopac Systems)
before analysis by low-pass filtering (cut-off frequency 25 Hz) and mean-value
smoothing using a three-sample window.
CRT Session. The CRT session incorporated elements of reappraisal (1–3) and
CBT and has been shown previously to persistently attenuate conditioned
fear responses (10) (see SI Methods and Figs. S6–S8 for details regarding this
protocol). The primary objective of the CRT session was to illustrate how
thoughts directly influence emotional responses. Participants were trained
to recognize this relationship as well as cognitive errors people typically
make when dealing with emotionally charged stimuli or events (i.e., cata-
strophizing). The participants were then instructed to reappraise the CS+ as
less threatening by generating alternative, more positive ways in which they
could think about the image, and the fear-conditioning session overall. After
this training, participants rerated the intensity of the emotions previously
reported to be associated with the CS+. All participants were informed that
they would return in 24 h to undergo the same fear-conditioning protocol,
during which they should consider and try to incorporate the strategies
reviewed during the CRT session.
Stress Manipulation. On day 2 (regulation phase), participants were randomly
underwent the CP task,during which theysubmergedtheir right hand to elbow
in a 0–4 °C ice-water bath for 3 min. The CP task is used widely in the labo-
ratory to model the effects of mild to moderate acute stress that participants
might encounter in everyday life (35–37) and has been shown to reliably ac-
tivate sympathetic nervous system and HPA axis arousal as measured by in-
creased physiological, endocrine (i.e., cortisol), and subjective levels of stress
(35–38). If a participant was unable to keep their arm in the water, the ex-
periment was terminated (n = 4; see Participants). The control participants
submerged their right arms in room-temperature water for 3 min. To assess
subjective levels of stress after the task, all participants rated how stressful
they found the stress or control task on a scale from 1 (not stressful) to 10
(most stressful) (SI Results). Participants waited 10 min after the CP/control task
before repeating the fear-conditioning task on day 2 to allow time for cortisol
levels to rise and to ensure participants’ arms in the CP condition were not cold
before the fear-conditioning session. Shock levels were recalibrated to ensure
that participants still found the US to be uncomfortable but not painful.
Neuroendocrine Analysis. Cortisol levels were measured using an oral swab
that participants placed under their tongue for 2 min. A baseline sample was
collected 10 min after participants’ arrival each day to allow for contextual
acclimation of the laboratory. On day 1, samples were also collected directly
after the fear-conditioning and CRT session. On day 2, samples were collected
10 min after the CP/control manipulation (+10 min) and directly after the fear-
conditioning session (+20 min). All samples were stored immediately in a sterile
tube in a freezer set to −20 °C for preservation. Samples were analyzed using
Salimetrics Testing Services (State College, PA). Samples were brought to room
temperature and centrifuged at ∼1500 × g for 15 min before testing.
Self-Reported Fear. Self-reported emotions and intensity ratings were col-
lected on day 1 after fear-conditioning. To confirm that regulation training
reduced the intensity of these reported emotions, participants rerated each
participants filled out a worksheet that reviewed the CRT session from the
previous day. Here, participants again reported and rated the intensity of
three emotions related to each CS. The participants then adjusted any re-
sidually fear-related emotions or thoughts to those more positive and
adaptive ones that were generated during the CRT session the previous day.
Self-reported emotions were categorized into one of six basic affective cate-
gories (i.e., fear, disgust, sadness, anger, surprise, happiness), or as neutral (10).
Because our objective was to examine how stress affects the regulation of fear,
only those emotions categorized under “fear” were analyzed. The proportion
of reported emotions categorized as fearful was used to confirm that par-
ticipants found the CS+ subjectively fearful. Intensity ratings for these fear-
related words were averaged across participants and served as an estimate
of subjective fear intensity.
Questionnaires. Participants completed self-report questionnaires at the start
of day 1 that included the State and Trait Anxiety Inventory (STAI-S, STAI-T)
(62), the Intolerance of Uncertainty Scale (IUS) (63), the Need for Closure
Scale (NFCS) (64), the Emotional Regulation Questionnaire (ERQ) (65), and
the Pain Catastrophizing Scale (PCS) (66). On day 2, participants repeated the
STAI-S, IUS, and PCS. None of these questionnaires yielded any group dif-
ferences on either day (SI Results).
ACKNOWLEDGMENTS. This work was supported by the National Institute of
Health Grants R01 AG039283, MH080756, and MH097085 (to E.A.P.), and a
Henry M. MacCracken Graduate Fellowship (to C.M.R.).
Raio et al. PNAS Early Edition
| 5 of 6
1. Gross JJ (1998) The emerging field of emotion regulation: An integrative review. Rev Download full-text
Gen Psychol 2(3):271–299.
2. Gross JJ, Thompson RA (2007) Emotion regulation: Conceptual foundations. Hand-
book of Emotion Regulation, ed Gross JJ (Guilford, New York), pp 3–24.
3. Ochsner KN, Silvers JA, Buhle JT (2012) Functional imaging studies of emotion regu-
lation: A synthetic review and evolving model of the cognitive control of emotion.
Ann N Y Acad Sci 1251:E1–E24.
4. Baumeister RF, Heatherton TF (1996) Self-regulation failure: An overview. Psychol Inq
5. Heatherton TF, Wagner DD (2011) Cognitive neuroscience of self-regulation failure.
Trends Cogn Sci 15(3):132–139.
6. Hartley CA, Phelps EA (2010) Changing fear: The neurocircuitry of emotion regula-
tion. Neuropsychopharmacology 35(1):136–146.
7. Frijda NH (1988) The laws of emotion. Am Psychol 43(5):349–358.
8. Smith CA, Lazarus RS (1988) Appraisal components, core relational themes, and the
emotions. Cogn Emotion 7(3):233–269.
9. Ochsner KN, Bunge SA, Gross JJ, Gabrieli JDE (2002) Rethinking feelings: An FMRI
study of the cognitive regulation of emotion. J Cogn Neurosci 14(8):1215–1229.
10. Shurick AA, et al. (2012) Durable effects of cognitive restructuring on conditioned
fear. Emotion 12(6):1393–1397.
11. Delgado MR, Nearing KI, Ledoux JE, Phelps EA (2008) Neural circuitry underlying the
regulation of conditioned fear and its relation to extinction. Neuron 59(5):829–838.
12. Schiller D, Delgado MR (2010) Overlapping neural systems mediating extinction, re-
versal and regulation of fear. Trends Cogn Sci 14(6):268–276.
13. Hofmann W, Schmeichel BJ, Baddeley AD (2012) Executive functions and self-regu-
lation. Trends Cogn Sci 16(3):174–180.
14. Schmeichel BJ, Volokhov RN, Demaree HA (2008) Working memory capacity and the
self-regulation of emotional expression and experience. J Pers Soc Psychol 95(6):
15. Gross JJ, Muñoz RF (1995) Emotion regulation and mental health. Clin Psychol Sci
16. Sapolsky RM (2007) Stress, stress-related disease, and emotional regulation. Hand-
book of Emotion Regulation, ed Gross JJ (Guilford, New York).
17. Cisler JM, Olatunji BO, Feldner MT, Forsyth JP (2010) Emotion regulation and the
anxiety disorders: An integrative review. J Psychopathol Behav Assess 32(1):68–82.
18. Johnstone T, van Reekum CM, Urry HL, Kalin NH, Davidson RJ (2007) Failure to reg-
ulate: Counterproductive recruitment of top-down prefrontal-subcortical circuitry in
major depression. J Neurosci 27(33):8877–8884.
19. Beck AT, Emery G (1985) Anxiety Disorders and Phobias: A Cognitive Perspective (Basic
Books, New York, NY).
20. Arnsten AF (2009) Stress signalling pathways that impair prefrontal cortex structure
and function. Nat Rev Neurosci 10(6):410–422.
21. Arnsten AFT, Wang MJ, Paspalas CD (2012) Neuromodulation of thought: Flexibilities
and vulnerabilities in prefrontal cortical network synapses. Neuron 76(1):223–239.
22. Roozendaal B, McEwen BS, Chattarji S (2009) Stress, memory and the amygdala. Nat
Rev Neurosci 10(6):423–433.
23. Alexander JK, Hillier A, Smith RM, Tivarus ME, Beversdorf DQ (2007) Beta-adrenergic
modulation of cognitive flexibility during stress. J Cogn Neurosci 19(3):468–478.
24. Plessow F, Fischer R, Kirschbaum C, Goschke T (2011) Inflexibly focused under stress:
Acute psychosocial stress increases shielding of action goals at the expense of reduced
cognitive flexibility with increasing time lag to the stressor. J Cogn Neurosci 23(11):
25. Plessow F, Kiesel A, Kirschbaum C (2012) The stressed prefrontal cortex and goal-di-
rected behaviour: Acute psychosocial stress impairs the flexible implementation of
task goals. Exp Brain Res 216(3):397–408.
26. Duncko R, Johnson L, Merikangas K, Grillon C (2009) Working memory performance
after acute exposure to the cold pressor stress in healthy volunteers. Neurobiol Learn
27. Elzinga BM, Roelofs K (2005) Cortisol-induced impairments of working memory re-
quire acute sympathetic activation. Behav Neurosci 119(1):98–103.
28. Luethi M, Meier B, Sandi C (2009) Stress effects on working memory, explicit memory,
and implicit memory for neutral and emotional stimuli in healthy men. Front Behav
29. Roozendaal B, McReynolds JR, McGaugh JL (2004) The basolateral amygdala interacts
with the medial prefrontal cortex in regulating glucocorticoid effects on working
memory impairment. J Neurosci 24(6):1385–1392.
30. Schoofs D, Wolf OT, Smeets T (2009) Cold pressor stress impairs performance on
working memory tasks requiring executive functions in healthy young men. Behav
31. Goldman-Rakic PS (1996) The prefrontal landscape: Implications of functional archi-
tecture for understanding human mentation and the central executive. Philos Trans R
Soc Lond B Biol Sci 351(1346):1445–1453.
32. Fuster JM (2008) The Prefrontal Cortex (Academic, London).
33. Qin S, Hermans EJ, van Marle HJ, Luo J, Fernández G (2009) Acute psychological stress
reduces working memory-related activity in the dorsolateral prefrontal cortex. Biol
34. Kern S, et al. (2008) Glucose metabolic changes in the prefrontal cortex are associated
with HPA axis response to a psychosocial stressor. Psychoneuroendocrinology 33(4):
35. Lovallo W (1975) The cold pressor test and autonomic function: A review and in-
tegration. Psychophysiology 12(3):268–282.
36. Lupien SJ, Maheu F, Tu M, Fiocco A, Schramek TE (2007) The effects of stress and stress
hormones on human cognition: Implications for the field of brain and cognition.
Brain Cogn 65(3):209–237.
37. Velasco M, Gómez J, Blanco M, Rodriguez I (1997) The cold pressor test: Pharmaco-
logical and therapeutic aspects. Am J Ther 4(1):34–38.
38. McRae AL, et al. (2006) Stress reactivity: Biological and subjective responses to the
cold pressor and trier social stressors. Hum Psychopharmacol 21(6):377–385.
39. Sapolsky RM, Romero LM, Munck AU (2000) How do glucocorticoids influence stress
responses? Integrating permissive, suppressive, stimulatory, and preparative actions.
Endocr Rev 21(1):55–89.
40. Thoma MV, Kirschbaum C, Wolf JM, Rohleder N (2012) Acute stress responses in
salivary alpha-amylase predict increases of plasma norepinephrine. Biol Psychol 91(3):
41. Nater UM, Rohleder N (2009) Salivary alpha-amylase as a non-invasive biomarker for
the sympathetic nervous system: Current state of research. Psychoneuroendocrinol-
42. van Stegeren A, Rohleder N, Everaerd W, Wolf OT (2006) Salivary alpha amylase as
marker for adrenergic activity during stress: Effect of betablockade. Psychoneur-
43. Hermans EJ, et al. (2011) Stress-related noradrenergic activity prompts large-scale
neural network reconfiguration. Science 334(6059):1151–1153.
44. Taylor F, Raskind MA (2002) The alpha1-adrenergic antagonist prazosin improves
sleep and nightmares in civilian trauma posttraumatic stress disorder. J Clinical Psy-
45. Raskind MA, et al. (2003) Reduction of nightmares and other PTSD symptoms in
combat veterans by prazosin: A placebo-controlled study. Am J Psychiatry 160(2):
46. Gründemann D, Schechinger B, Rappold GA, Schömig E (1998) Molecular identifica-
tion of the corticosterone-sensitive extraneuronal catecholamine transporter. Nat
47. Indovina I, Robbins TW, Núñez-Elizalde AO, Dunn BD, Bishop SJ (2011) Fear-condi-
tioning mechanisms associated with trait vulnerability to anxiety in humans. Neuron
48. Vohs KD, Heatherton TF (2000) Self-regulatory failure: A resource-depletion ap-
proach. Psychol Sci 11(3):249–254.
49. Delgado MR, Gillis MM, Phelps EA (2008) Regulating the expectation of reward via
cognitive strategies. Nat Neurosci 11(8):880–881.
50. Li CS, Sinha R (2008) Inhibitory control and emotional stress regulation: Neuroimaging
evidence for frontal-limbic dysfunction in psycho-stimulant addiction. Neurosci Bio-
behav Rev 32(3):581–597.
51. Hutcherson CA, Plassmann H, Gross JJ, Rangel A (2012) Cognitive regulation during
decision making shifts behavioral control between ventromedial and dorsolateral
prefrontal value systems. J Neurosci 32(39):13543–13554.
52. Mitchell DGV (2011) The nexus between decision making and emotion regulation: A
review of convergent neurocognitive substrates. Behav Brain Res 217(1):215–231.
53. Sokol-Hessner P, et al. (2009) Thinking like a trader selectively reduces individuals’ loss
aversion. Proc Natl Acad Sci USA 106(13):5035–5040.
54. Amodio DM, Devine PG (2010) Control in the regulation of intergroup bias. Self
Control in Society, Mind and Brain eds Hassin RR, Ochsner KN, Trope Y (Oxford Univ
55. Krendl AC, Kensinger EA, Ambady N (2012) How does the brain regulate negative
bias to stigma? Soc Cogn Affect Neurosci 7(6):715–726.
56. Hauner KK, Mineka S, Voss JL, Paller KA (2012) Exposure therapy triggers lasting
reorganization of neural fear processing. Proc Natl Acad Sci USA 109(23):9203–9208.
57. Klorman R, Weerts TC, Hastings JE, Melamed BG, Lang PJ (1974) Psychometric de-
scription of some specific-fear questionnaires. Behav Ther 5(3):401–409.
58. Fredrikson M (1983) Reliability and validity of some specific fear questionnaires.
Scand J Psychol 24(4):331–334.
59. Schiller D, et al. (2010) Preventing the return of fear in humans using reconsolidation
update mechanisms. Nature 463(7277):49–53.
60. Critchley HD (2002) Electrodermal responses: What happens in the brain. Neurosci-
61. Lykken DT, Venables PH (1971) Direct measurement of skin conductance: A proposal
for standardization. Psychophysiology 8(5):656–672.
62. Spielberger CD, Gorsuch RL, Lusthene RE (1970) Manual for the State-Trait Anxiety
Inventory (Consulting Psychologists Press, Palo Alto, CA).
63. Buhr K, Dugas MJ (2002) The Intolerance of Uncertainty Scale: Psychometric proper-
ties of the English version. Behav Res Ther 40(8):931–945.
64. Kruglanski AW, Webster DM, Klem A (1993) Motivated resistance and openness to per-
suasion in the presence or absence of prior information. J Pers Soc Psychol 65(5):861–876.
65. Gross JJ, John OP (2003) Individual differences in two emotion regulation processes:
Implications for affect, relationships, and well-being. J Pers Soc Psychol 85(2):348–362.
66. Sullivan MJL, Bishop S, Pivik J (1995) The Pain Catastrophizing Scale: Development
and validation. Psychol Assess 7(4):524–532.
6 of 6
| www.pnas.org/cgi/doi/10.1073/pnas.1305706110 Raio et al.