Cortisol-Induced Impairments of Working Memory Require Acute
Bernet M. Elzinga and Karin Roelofs
University of Leiden
The present study assessed whether the effects of cortisol on working memory depend on the level of
adrenergic activity (as measured by sympathetic activation) during memory performance. After exposure
to a psychosocial stress task, participants were divided into cortisol responders and nonresponders.
Cortisol responders showed working memory impairments during the psychosocial stress phase, when
cortisol and adrenergic activity were enhanced, whereas nonresponders did not. During recovery,
however, when cortisol levels were elevated but adrenergic activity was normalized, working memory of
responders did not differ from that of nonresponders. Among several stress measures, cortisol was the
only significant predictor for working memory performance during stress. These findings suggest that
adrenergic activation is essential for the impairing effects of stress-induced cortisol on working memory.
It is well established that glucocorticoid (GC) hormones re-
leased from the adrenal cortex during stressful or emotionally
arousing experiences play an important role in the regulation of
cognitive functions (see De Kloet, Oitzl, & Joe ¨ls, 1999; Lupien &
Lepage, 2001; Roozendaal, 2002). Several studies in humans have
documented a deleterious effect of GCs on the retrieval of previ-
ously learned material, although they may enhance long-term
consolidation of (emotional) material (Buchanan & Lovallo, 2001;
Cahill, Gorski, & Le, 2003; Kirschbaum, Wolf, May, Wippich, &
Hellhammer, 1996; Newcomer, Craft, Hershey, Askins, &
Bardgett, 1994; Newcomer et al., 1999; De Quervain, Roozendaal,
Nitsch, McGaugh, & Hock, 2000). Working memory, the cognitive
mechanism that allows us to keep a limited amount of information
active for a limited period of time (Baddeley, 1996), is one of the
memory functions that is particularly sensitive to GCs. Three
studies in humans have consistently shown cortisol administration
to impair working memory, without having effects on declarative
memory and sustained attention (Lupien, Gillin, & Hauger, 1999;
Wolf et al., 2001; Young, Sahakian, Robbins, & Cowen, 1999),
suggesting that cortisol specifically impairs the working memory
Evidence from lesion, pharmacological, and preclinical studies
indicates that the prefrontal cortex (PFC) is a key brain structure in
working memory (Arnsten, 1998; Fuster, 1997). This is supported
by neuroimaging studies showing a significant linear relationship
between the increase in working memory load and the degree of
activation observed in the dorsolateral PFC (Braver et al., 1997;
Jansma, Ramsey, Coppola, & Kahn, 2000; Rypma, Prabhakaran,
Desmond, Glover, & Gabrieli, 1999; Veltman, Rombouts, &
Dolan, 2003). The PFC is a significant target for the negative-
feedback actions of circulating GCs (Sanchez, Young, Plotsky, &
Insel, 2000). Moreover, animal research has shown that chronic
corticosterone administration and chronic behavioral stress can
lead to dendritic reorganization in the medial PFC (Radley et al.,
2004; Wellman, 1993), and that exposure to stress-induced cortisol
elevation can impair prefrontal behavioral control (Lyons, Lopez,
Yang, & Schatzberg, 2000).
Recent findings from studies in animals suggest that the influ-
ence of GCs on memory functioning depends on the level of
training-induced emotional arousal (Okuda, Roozendaal, & Mc-
Gaugh, 2004) and/or adrenergic activity (Roozendaal, 2000) dur-
ing memory performance. Okuda and colleagues (2004) showed
that only rats that were not habituated to an experimental context
(i.e., increased novelty-induced emotional arousal) showed an
inverted-U-shaped relationship between dose of corticosterone and
object recognition, whereas corticosterone did not affect object
recognition of rats that were habituated to the experimental context
(i.e., decreased novelty-induced emotional arousal). The activation
of noradrenergic mechanisms in the basolateral complex of the
amygdala (BLA) may be a critical component in mediating the
effects of emotional arousal on the effects of GCs on memory
functions in rats, by interacting with several other brain regions,
including the hippocampus and the PFC (McGaugh & Roozendaal,
2002; Roozendaal, 2000). Consistent with the suggested role of the
BLA in mediating the effects of GCs on prefrontal cognitive
functions, GC treatments induced impairments in the performance
of a spatial memory task, a task that depends on the medial PFC,
whereas BLA lesions blocked the impairments induced by GC
administration (Roozendaal, McReynolds, & McGaugh, 2004).
?-adrenoceptor antagonist propranolol prevented corticosterone-
induced memory impairment (Roozendaal et al., 2004), suggesting
that the modulation of working memory by cortisol elevation
requires noradrenergic activation. To our knowledge, the role of
Bernet M. Elzinga and Karin Roelofs, Section of Clinical and Health
Psychology, University of Leiden, Leiden, the Netherlands.
This study was supported by Veni Grant 451-02-116 from the Nether-
lands Organization for Scientific Research awarded to Bernet M. Elzinga.
We thank B. Poelmans, J. Meulman, R. Siekman, and F. Vulker for their
assistance during data collection, and R. de Kloet, M. Oitzl, and P.
Spinhoven for their valuable theoretical comments.
Correspondence concerning this article should be addressed to Bernet
M. Elzinga, University of Leiden, Section of Clinical and Health Psychol-
ogy, P. O. Box 9555, Leiden 2300 RB, the Netherlands. E-mail: elzinga@
2005, Vol. 119, No. 1, 98–103
Copyright 2005 by the American Psychological Association
0735-7044/05/$12.00 DOI: 10.1037/0735-7044.119.1.98
noradrenergic activation in mediating the effects of GCs on work-
ing memory has not been investigated in humans.
To investigate whether the effects of stress-induced cortisol
elevations on working memory depend on the level of adrenergic
activity (as measured by sympathetic autonomic activation) during
working memory performance, we randomly assigned participants
to either a psychosocial stress (Trier Social Stress Task [TSST]) or
a reading condition, after which working memory (Digit Span
forward and backward; Wechsler, 1987) was assessed at baseline
(i.e., when both cortisol levels and adrenergic activity are low),
during acute psychosocial stress (when both cortisol levels and
adrenergic activity are high), and after psychosocial stress (i.e.,
when cortisol levels are still elevated, but adrenergic activity is
low). We hypothesized that working memory would specifically
be impaired under conditions of both elevated cortisol levels and
increased adrenergic activation.
Forty-four college students, who participated for financial reward or
course credit, were randomly assigned to either the stress condition (n ?
22; 12 men and 10 women; mean age ? 21.33 years, SD ? 4.21) or control
reading condition (n ? 22; 11 men and 11 women; mean age ? 21.10
years, SD ? 1.84). Participants were recruited from the University of
Leiden through announcements. Exclusion criteria were as follows: any
Axis I psychiatric disorder, including substance abuse (DSM–IV); any
clinically significant medical disease; use of medication (including oral
contraceptives); and being less than 18 or more than 37 years of age. All
women were tested in the late luteal phase (Days 21–25) of their menstrual
cycle according to self-report. The luteal phase was chosen because there
are indications that during this phase, stress-induced cortisol levels do not
differ between men and women (see Kirschbaum, Kudielka, Gaab, Schom-
mer, & Hellhammer, 1999). Participants were asked to minimize physical
exercise during the hour preceding the experiment and not to take large
meals, coffee, drinks with low pH, or cigarettes, because these variables
can have an influence on cortisol levels. All participants signed informed
speech and mental arithmetic task of 15 min duration, has repeatedly been
found to induce significant endocrine and cardiovascular responses
(Kirschbaum, Pirke, & Hellhammer, 1993). Participants received instruc-
tions in which they were told that they would be taking on the role of a job
applicant for the position of research assistant at the University of Leiden.
Participants were given 5 min to prepare a 5-min free speech to an audience
of three psychologists, one of whom acted as chairman. They were told that
the speech would be videotaped, that the psychologists were trained to
monitor nonverbal behavior, that a voice-frequency analysis of nonverbal
behavior would be performed, and that the speech would be critiqued on
content and presentation style. Following preparation time, the audience
entered the room and obviously switched on the camera and microphone.
Participants were instructed to stand in front of a table with the audience
sitting at the other side, and the chairman asked the participant to describe
his or her qualifications for the job. Participants were expected to use the
entire 5 min for the speech, as described by Kirschbaum et al. (1993). For
the mental arithmetic task, participants were instructed to serially subtract
13, starting with 1587 (i.e., 1587 ? 13, 1574 ? 13, etc.). The audience
responded to any mistakes by instructing participants to start again from
This psychosocial challenge test, which mainly consists of a free
Working memory task.
Span subtest of the Wechsler Adult Intelligence Scale—Revised (Wech-
sler, 1987). Five series of numbers of increasing length (from 4 to 8 in the
forward condition, and 3 to 7 in the backward condition) were read to each
participant at the rate of one digit per second. Participants had to repeat the
numbers in the same order (forward condition) or in reversed order (back-
ward condition). Each set length was tested twice. For each correctly
repeated digit set, the number of digits was added up and the scores for the
forward and backward conditions were combined, so that scores could
range from 0 to 120. Three parallel versions of the working memory task
were administered (see Figure 1): one at baseline (no audience present),
one immediately after the stress task in front of the audience, and one after
the stress task (no audience present). Administration of the three versions
Working memory was measured with the Digit
All physiological assessments were obtained over a 95-min period at
eight assessment points: ?45, ?30, ?15, ?5, ?10, ?20, ?35, and ?50
min with reference to the start of the stressor.
Cortisol samples were obtained with Salivette collection de-
vices (Sarstedt, Rommelsdorf, Germany). Saliva samples were stored at
?20 °C before assay. Biochemical analysis of free cortisol in saliva was
performed with a competitive electrochemiluminescence immunoassay
(Elecsys 2010; Roche Diagnostics, Laval, Quebec, Canada), as described
elsewhere (van Aken, Romijn, Miltenburg, & Lentjes, 2003).
Heart rate was recorded continuously by an ambulatory
monitoring system (Version 3.6; Vrije Universiteit Amsterdam), a small
battery-powered device for ambulatory recording. It was measured with
three Ag–AgCl disposable electrodes (ConMed, Utica, NY) placed just
above the sternum, at the left side of the chest and at the bottom right side
of the chest. For each participant, heart rate was averaged for 2 min,
starting from a marker given at each of the eight assessment points.
Systolic and diastolic blood pressure were measured
from the nondominant arm with a blood pressure monitor (Model 705CP;
Omron, Bannockburn, IL).
Subjective measures of anxiety, nervousness,
and feelings of insecurity were assessed on a visual analogue scale ranging
from 0 to 10.
Participants arrived at the laboratory at 9 a.m., after which
the ambulatory monitoring system device was connected and checked by
the experimenter. Participants were randomly assigned to the stress or
reading condition. At ?45 min with reference to the start of the stressor, a
first series of physiological and subjective measures were assessed. At ?35
min, the working memory task was administered for the first time (base-
line), followed by a few additional tasks, results of which will be reported
elsewhere (Roelofs, Elzinga, & Rotteveel, 2004), and three series of
physiological and subjective measures at ?30, ?15, and ?5 min. Subse-
quently, the TSST took place between 0 and ?15 min. Immediately after
the TSST, the chairman administered the second working memory task
while the participant was standing in front of the audience, so that the
social stress context remained present during the second administration of
the working memory task (stress period). At ?35, min the audience left the
room. At ?50 min, working memory performance was assessed for a third
time (recovery period), followed by the last physiological and subjective
measures. After finishing the experiment, the audience returned for a short
To compare the experimental condition and the control condition on
working memory performance and on the physiological and subjective
stress responses, we used analyses of variance (ANOVAs) for repeated
measures, with condition (stress vs. control) as a between-subjects factor
ACUTE STRESS, CORTISOL, AND WORKING MEMORY
and time as a within-subjects factor. Analyses were performed with SPSS
11.5 (SPSS, Chicago, IL). The criterion for statistical significance was p ?
A separate ANOVA with repeated measures for the physiolog-
ical and subjective stress measures showed significant increases on
all stress measures over time in the stress, but not in the control,
condition: significant interactions for Group ? Time for heart rate,
F(7, 287) ? 6.38, p ? .0001; systolic blood pressure, F(7, 294) ?
5.31, p ? .0001; diastolic blood pressure, F(7, 294) ? 8.26, p ?
.0001; anxiety, F(7, 294) ? 4.94, p ? .0001; nervousness, F(7,
294) ? 15.26, p ? .0001; and insecurity, F(7, 294) ? 8.07, p ?
.0001. Post hoc t tests in the stress group indicated that compared
to baseline, all measures were elevated during the stress phase at
?10 and ?20 min (all ps ? .001) and were back to baseline levels
at ?50 min (all ps ? .10).
An ANOVA with repeated measures for mean cortisol levels
showed significant increases over time in the stress group, whereas
cortisol levels decreased over time in the control group, F(7,
294) ? 5.76, p ? .0001. Post hoc t tests in the stress group
indicated that there was an overall significant difference between
cortisol levels right before the TSST (8.69 ? 1.05 nmol/L) and the
peak cortisol response after stress (13.08 ? 1.89 nmol/L) of
50.5%, t(21) ? 3.30, p ? .005, whereas in the control condition
there was a significant decrease between cortisol levels right
before reading (8.63 ? 0.81 nmol/L) and at the end of the reading
phase (7.33 ? 0.62 nmol/L) of 15.0%, t(22) ? 4.53, p ? .0001.
When sex was entered in the repeated measures, a trend-level
interaction was found between cortisol, sex, and condition, F(7,
287) ? 1.96, p ? .06. In a post hoc analysis, it appeared that
whereas men showed an overall increase of 66.0% between cor-
tisol levels right before the TSST (11.06 ? 1.51 nmol/L) and after
the TSST (18.36 ? 2.59 nmol/L), t(11) ? 3.48, p ? .005, and
women showed only a slight increase of 15.4% between cortisol
levels right before the TSST (5.85 ? 0.82 nmol/L) and after the
TSST (6.75 ? 0.53 nmol/L), t(9) ? 2.20, p ? .06, no differences
were found on any of the cortisol measures between men and
women in the control condition.
Because of the high variance of cortisol levels in the TSST
group, a post hoc median-split was conducted on the basis of the
absolute difference between peak cortisol levels after 20 min and
baseline cortisol levels right before the TSST, at ?5 min (median
split at 1.90 nmol/L), resulting in a group of cortisol responders (9
men and 2 women) and nonresponders (3 men and 8 women).
Responders showed elevated cortisol levels from right after the
TSST until the end of the experiment (from ?10 to ?50 min)
compared with nonresponders, whereas no differences were found
between the two groups prior to the TSST (from ?45 to ?5 min):
interaction between group and time, F(7, 140) ? 4.06, p ? .0001
(see Figure 1). In repeated measures ANOVA analyses with heart
rate, blood pressure, and subjective measures as dependent vari-
ables, responders did not differ from nonresponders with respect to
heart rate, F(7, 126) ? 0.55, ns; systolic blood pressure, F(7,
140) ? 0.62, ns; diastolic blood pressure, F(7, 140) ? 1.04, ns, or
subjective measures of anxiety, F(7, 140) ? 0.39, ns; nervousness,
F(7, 140) ? 2.00, ns (trend for responders to be less nervous); or
feelings of insecurity, F(7, 140) ? 0.16, ns.
Regardless of condition, working memory performance im-
proved with repeated administration: main effect of time, F(2,
84) ? 4.91, p ? .01. Overall, the stress group did not differ from
controls on working memory performance: no main effect of
group, F(1, 42) ? 0.88, ns. With sex entered as an additional
factor, there was no main effect of sex on working memory
performance, F(1, 40) ? 2.20, ns, regardless of condition. Simi-
larly, within the stress condition, there was no main effect of sex,
F(1, 20) ? 0.01, ns, or interaction between sex and the three
working memory assessments, F(2, 40) ? 0.42, ns.
In order to test our main hypothesis on the effects of cortisol-
adrenergic increases on working memory, we performed a re-
peated measures ANOVA with working memory as a within-
subjects factor and responders versus nonresponders as a between-
subjects factor. Consistent with our hypothesis, working memory
performance of cortisol responders specifically deteriorated during
the TSST compared with that of nonresponders, whereas working
memory performance of responders did not differ from that of
nonresponders at baseline or during the recovery phase: quadratic
interaction between group and time, F(1, 20) ? 4.51, p ? .05 (see
Figure 2). When Digit Span forward and backward were analyzed
separately, the deterioration of responders during the TSST was
significant in the forward condition, F(1, 20) ? 5.17, p ? .05, but
not in the backward condition, F(1, 20) ? 0.54, ns.
When analyzed separately, participants in the control condition
tended to show an overall improvement in working memory per-
formance with repeated administration, similar to the nonre-
sponders in the stress condition, F(2, 42) ? 2.97, p ? .06 (M ?
SD: first assessment, 47.86 ? 4.41; second assessment, 49.55 ?
4.07; third assessment, 53.59 ? 4.68).
Finally, to analyze which factor was the best predictor of work-
ing memory during the TSST, we performed a regression analysis
with working memory performance during the TSST as the de-
pendent variable, and sex, increase in cortisol, heart rate, blood
pressure, and nervousness as independent variables. With this
before, during, and after the Trier Social Stress Task in responders and
nonresponders. WMb? working memory at baseline; WMs? working
memory during stress; WMr? working memory during recovery. * sig-
nificant difference at p ? .05.
Mean (? SEM) free salivary cortisol (in nanomoles per liter)
ELZINGA AND ROELOFS
model, 43% of the variance of working memory performance was
explained (r ? .66). Consistent with the findings of the ANOVA
analysis, the best and only significant predictor of working mem-
ory performance was the increase in cortisol levels (? ? ?0.54),
t(xx) ? ?2.23, p ? .05. Sex did not significantly contribute to the
working memory performance (? ? ?0.28), t(19) ? ?1.17, ns.
The purpose of the present study was to assess whether the
effects of stress-induced cortisol elevations on working memory
depend on the level of adrenergic activity during memory perfor-
mance. Consistent with our hypothesis, we found that working
memory was significantly impaired among cortisol responders
when they had to perform the memory task in a stressful context
(i.e., in front of an audience), during which both cortisol levels and
sympathetic activity were enhanced. In contrast, individuals who
did not show any cortisol increase in response to the stress task
(nonresponders) did not show working memory impairments when
performing the task in front of the same audience. In contrast, their
performance even improved. Moreover, during the recovery phase,
when cortisol levels were still elevated in responders, but the
audience had left and sympathetic activity was back to baseline,
working memory was no longer significantly impaired in respond-
ers. The fact that responders were exclusively distinguishable from
nonresponders on the basis of their cortisol responses, and not on
the basis of any other stress response, either sympathetic or sub-
jective, suggests that the working memory impairments during
stress were primarily associated with stress-induced cortisol levels.
The finding that the level of cortisol increase was the best and only
significant predictor for working memory performance during
stress in a regression analysis supports this idea. Altogether, these
findings further substantiate the hypothesis that stress-induced
adrenergic activation plays an essential role in the impairing ef-
fects of stress-induced cortisol elevations on working memory.
The present findings are remarkably consistent with studies in
animals showing that the effects of GCs on working memory
depend on the level of stress-induced arousal (Okuda et al., 2004)
and adrenergic activity (Roozendaal et al., 2004). The latter study
showed that the noradrenergic system, and more specifically the
BLA, interacts with the PFC in regulating the effects of GC
hormones on working memory functions, and that lesions or phar-
macological inactivation of the BLA block the effects of GCs on
prefrontal cognitive functions (Roozendaal et al., 2004). The
present findings are also consistent with the neurobiological ef-
fects of stress exposure on the PFC, as the PFC is a target for both
circulating GCs and norepinephrine (NE), and decreased PFC
functioning has been reported in relation to both substances (see
Arnsten & Goldman-Rakic, 1998; Lupien & Lepage, 2001). In-
creases in NE after stress exposure or administration of anxiogenic
drugs have been shown to impair PFC functioning (including
working memory) in rats through NE’s actions at postsynaptic ?-1
adrenergic receptors (Arnsten, 1998; Arnsten & Goldman-Rakic,
1998), which could be reversed by administration of NE ?-2
agonists (clonidine and guanfacine (Birnbaum, Podell, & Arnsten,
2000). In the present study, neither sympathetic activation nor
cortisol elevations alone resulted in working memory impairments,
as not all individuals showed memory impairments during the
stress phase (even though all participants had increased sympa-
thetic activation), and no impairments were found in responders
during recovery (when cortisol was still elevated). This suggests
that in humans, stress-induced cortisol elevations require sympa-
thetic activation for working memory impairments to occur.
The present findings provide an important extension of earlier
pharmacological studies showing that (only) high doses of cortisol
administration reduce working memory performance in humans
(Lupien et al., 1999; Wolf et al., 2001; Young et al., 1999). Lupien
et al. (1999), for example, found that only high doses of cortisol
administration (600 ?g/kg/hr?1) led to impairments of working
memory, whereas moderate doses did not. In these studies, work-
ing memory was tested in a nonstressful environment, comparable
to our third administration of the working memory task. These
findings, taken together with the results from our study, could
indicate that when acting alone, cortisol may diminish working
memory only at extremely high levels, whereas within the physi-
ological range, cortisol may induce working memory only when
secreted under conditions of acute stress.
Besides the physiological effects of the stress induction, work-
ing memory impairments during acute stress in cortisol responders
may also be related to the fact that the task was performed in a
stressful context, in front of an audience of three people. Studies in
animals have shown that cognitive impairments during stress de-
pend on an interplay between GC responses and the context in
which they are elicited, particularly the relevance of the task in
relation to its context, the presence of other distracting factors, et
cetera (De Kloet et al., 1999). Being engaged in a working memory
task in front of an audience can be seen as a form of dual tasking
in which metathoughts related to social evaluation may interfere
with the memory performance. In the present context, individuals
with high cortisol levels may have had more difficulties in inhib-
iting task-irrelevant thoughts, thereby preempting processing re-
sources and some of the available capacity of working memory.
However, responders did not report more anxiety, nervousness, or
feelings of insecurity, as may be expected if cortisol activity is
related to more intrusive thoughts about the stressful context. In
contrast, they even reported being slightly less nervous than
The working memory impairments of cortisol responders during
the acute stress phase were most prominent in the Digit Span
forward condition, whereas no significant differences were found
between responders and nonresponders in the backward condition.
Verbal working memory has been hypothesized to consist of two
and during the stress and recovery periods of the Trier Social Stress Task
for cortisol responders and nonresponders. * significant difference at p ?
Mean (? SD) on working memory performance at baseline,
ACUTE STRESS, CORTISOL, AND WORKING MEMORY
subprocesses, that is, maintenance (storage, rehearsal, and match-
ing) and manipulation (reordering or updating) of information,
with the Digit Span forward condition primarily assessing main-
tenance and the backward condition mainly measuring manipula-
tion. Moreover, the maintenance and manipulation subdivisions
have been suggested to correspond to distinct roles of the ventro-
lateral PFC and the dorsolateral PFC in working memory (Fletcher
& Henson, 2001), although recent imaging studies directly com-
paring maintenance versus manipulation processes in working
memory did not always find different brain activation patterns (see
Veltman et al., 2003). Because the sample sizes of the cortisol
responder and nonresponder groups were relatively small, the
specific effects on the Digit Span forward condition should be
interpreted with caution. If the present findings are replicated in
larger samples, however, this would imply that the maintenance of
information is more sensitive to the effects of stress-induced
cortisol-adrenergic activation than the manipulation of informa-
tion. Future studies using functional imaging techniques will be
necessary to assess whether the possible differential effects of
cortisol–adrenergic activation on the two subprocesses are associ-
ated with specific activation patterns in the ventrolateral versus the
Previous studies have implied that sex and age play important
roles in mediating the effects of cortisol on memory, with older
women being more susceptible to the effects of GCs (see Seeman,
McEwen, Singer, Albert, & Rowe, 1997; Wolf, Kudielka, Hell-
hammer, Hellhammer, & Kirschbaum, 1999), whereas in young
populations, men may be more sensitive to the effects of cortisol
than women (Wolf et al., 2001). To minimize sex differences in
cortisol responses, we excluded from the present study women
who were taking oral contraceptives and tested all women in the
late luteal phase (Days 21–25) of their menstrual cycle according
to self-report, as there are indications that during this phase,
stress-induced cortisol levels do not differ between men and
women (see Kirschbaum et al., 1999). Nevertheless, mean stress-
induced cortisol increases tended to be somewhat larger in men
than in women, and as a consequence the responder group con-
tained more men (n ? 9) than women (n ? 2). It should first be
noted that the memory impairments in the responder group were
not due to a simple sex effect on working memory, as no differ-
ences were found on working memory performance between men
and women in the nonresponders or the control group. Because of
the small sample sizes, we could not directly assess whether the
working memory performance of female cortisol responders was
differently affected by the cortisol increases than those of male
responders. With the responders and nonresponders taken together,
however, men did not show larger working memory impairments
than women, and sex did not contribute as a factor to working
memory performance during stress in the regression analysis.
Thus, although no sex effects on working memory were found,
men were overrepresented in the responder group, suggesting that
memory impairments may occur more often in men because men
generally show more pronounced stress-induced cortisol increases
than women. Additional studies in larger samples are clearly
needed to further address whether men and women differ in their
susceptibility to stress-induced cortisol effects on working mem-
ory, and whether this is dependent on the course of the menstrual
To our knowledge, this is the first time that the role of acute
stress has been studied in relation to the effects of GCs on memory
performance in humans. Although studies with larger samples are
necessary to further elucidate all the physiological and cognitive
factors involved in these impairments, these findings could shed
new light on the effects of stress on memory functioning. One of
the interesting questions that arises is whether acute stress medi-
ates only cortisol-induced effects on working memory, or whether
it also plays a role in mediating the effects of cortisol on declar-
ative memory tasks that depend on hippocampal functions, as
would be consistent with findings from studies in animals (Kim,
Lee, Han, & Packard, 2001; Roozendaal, Griffith, Buranday, de
Quervain, & McGaugh, 2003; Roozendaal & McGaugh, 1997;
Roozendaal, Nguyen, Power, & McGaugh, 1999). So far, two
studies in humans are consistent with this idea, reporting selective
enhanced delayed recall of emotionally arousing pictures com-
pared with neutral pictures after prelearning cortisol administration
(Buchanan & Lovallo, 2001), and after postlearning cortisol ele-
vations induced by cold presser stress (Cahill, Gorski, & Le,
2003). A third study did not replicate these findings, however
(Rimmele, Domes, Mathiak, & Hautzinger, 2003). Another inter-
esting implication that needs further investigation is related to the
fact that working memory of responders improved when sympa-
thetic activation was back to baseline, suggesting that working
memory impairments can be prevented by blocking sympathetic
activation. Studies in animals have shown that systemic injections
of the centrally acting ?-adrenoceptor antagonist propranolol pre-
vented corticosterone-induced memory impairment (Roozendaal et
al., 2004). Future studies investigating these questions will be very
relevant, as they may have important implications for the preven-
tion of blackouts and concentration problems during stressful
situations such as a job interview or examination, and may provide
helpful new venues for the treatment of psychiatric patients with
stress-induced prefrontal impairments such as posttraumatic stress
disorder or depression.
Arnsten, A. F. T. (1998). Catecholamine modulation of prefrontal cortical
cognitive function. Trends in Cognitive Sciences, 2, 436–447.
Arnsten, A. F. T., & Goldman-Rakic, P. S. (1998). Noise stress impairs
prefrontal cortical cognitive function in monkeys: Evidence for a hyper-
dopaminergic mechanism. Archives of General Psychiatry, 55, 362–369.
Baddeley, A. (1996). The fractionation of working memory. Proceedings
of the National Academy of Sciences, USA, 93, 13468–13472.
Birnbaum, S. G., Podell, D. M., & Arnsten, A. F. T. (2000). Noradrenergic
alpha-2 receptor agonists reverse working memory deficits induced by
the anxiogenic drug FG142, in rats. Pharmacology Biochemistry and
Behavior, 67, 397–403.
Braver, T. S., Cohen, J. D., Nystrom, L. E., Jonides, J., Smith, E. E., &
Noll, D. C. (1997). A parametric study of prefrontal cortex involvement
in human working memory. NeuroImage, 5, 49–62.
Buchanan, T. W., & Lovallo, W. R. (2001). Enhanced memory for emo-
tional material following stress-level cortisol treatment in humans. Psy-
choneuroendocrinology, 26, 307–317.
Cahill, L., Gorski, L., & Le, K. (2003). Enhanced human memory consol-
idation with post-learning stress: Interaction with the degree of arousal
at encoding. Learning & Memory, 10, 270–274.
De Kloet, E. R., Oitzl, M. S., & Joe ¨ls, M. (1999). Stress and cognition: Are
corticosteroids good or bad guys? Trends in Neuroscience, 22, 422–426.
De Quervain, D. J. F., Roozendaal, B., Nitsch, R. M., McGaugh, J. L., &
ELZINGA AND ROELOFS
Hock, C. (2000). Acute cortisone administration impairs retrieval of
long-term declarative memory in humans. Nature Neuroscience, 3, 313–
Fletcher, P. C., & Henson, R. N. (2001). Frontal lobes and human memory:
Insights from functional neuroimaging. Brain, 124, 849–881.
Fuster, J. M. (1997). The prefrontal cortex: Anatomy, physiology, and
neuropsychology of the frontal lobe. Philadelphia: Lippincott-Raven
Jansma, J. M., Ramsey, N. F., Coppola, R., & Kahn, R. S. (2000). Specific
versus nonspecific brain activity in a parametric N-back task. NeuroIm-
age, 12, 688–697.
Kim, J. J., Lee, H. J., Han, J. S., & Packard, M. G. (2001). Amygdala is
critical for stress-induced modulation of hippocampal long-term poten-
tiation and learning. Journal of Neuroscience, 21, 5222–5228.
Kirschbaum, C., Kudielka, B. M., Gaab, J., Schommer, N. C., & Hellham-
mer, D. H. (1999). Impact of gender, menstrual cycle phase, and oral
contraceptives on the activity of the hypothalamic-pituitary-adrenal axis.
Psychosomatic Medicine, 61, 154–162.
Kirschbaum, C., Pirke, K. M., & Hellhammer, D. H. (1993). The Trier
Social Stress Test: A tool for investigating psychobiological responses in
a laboratory setting. Neuropsychobiology, 28, 76–81.
Kirschbaum, C., Wolf, O. T., May, M., Wippich, W., & Hellhammer, D. H.
(1996). Stress- and treatment-induced elevations of cortisol levels asso-
ciated with impaired declarative memory in healthy adults. Life Sci-
ences, 58, 1475–1483.
Lupien, S. J., Gillin, C. J., & Hauger, R. L. (1999). Working memory is
more sensitive than declarative memory to the acute effects of cortico-
steroids: A dose-response study in humans. Behavioral Neuroscience,
Lupien, S. J., & Lepage, M. (2001). Stress, memory, and the hippocampus:
Can’t live with it, can’t live without it. Behavioural Brain Research,
Lyons, D. M., Lopez, J. M., Yang, C., & Schatzberg, A. F. (2000).
Stress-level cortisol treatment impairs inhibitory control of behavior in
monkeys. Journal of Neuroscience, 20, 7816–7821.
McGaugh, J. L., & Roozendaal, B. (2002). Role of adrenal stress hormones
in forming lasting memories in the brain. Current Opinion in Neurobi-
ology, 12, 205–210.
Newcomer, J. W., Craft, S., Hershey, T., Askins, K., & Bardgett, M. E.
(1994). Glucocorticoid-induced impairment in declarative memory per-
formance in adult humans. Journal of Neuroscience, 14, 2047–2053.
Newcomer, J. W., Selke, G., Melson, A. K., Hershey, T., Craft, S.,
Richards, K., & Alderson, A. L. (1999). Decreased memory performance
in healthy humans induced by stress-level cortisol treatment. Archives of
General Psychiatry, 56, 527–533.
Okuda, S., Roozendaal, B., & McGaugh, J. L. (2004). Glucocorticoid
effects on object recognition memory require training-associated emo-
tional arousal. Proceedings of the National Academy of Sciences, USA,
Radley, J. J., Sisti, H. M., Hao, J., Rocher, A. B., McCall, T., Hof, P. R.,
et al. (2004). Chronic behavioral stress induces apical dendritic reorga-
nizations in pyramidal neurons of the medial prefrontal cortex. Neuro-
science, 125, 1–6.
Rimmele, U., Domes, G., Mathiak, K., & Hautzinger, M. (2003). Cortisol
has different effects on human memory for emotional and neutral stim-
uli. NeuroReport, 14, 2485–2488.
Roelofs, K., Elzinga, B. M., & Rotteveel, M. (2004). The effects of
stress-induced cortisol responses and trait avoidance on approach-
avoidance behavior. Manuscript submitted for publication.
Roozendaal, B. (2000). Glucocorticoids and the regulation of memory
consolidation. Psychoneuroendocrinology, 25, 213–238.
Roozendaal, B., Griffith, Q. K., Buranday, J., de Quervain, D. J. F., &
McGaugh, J. L. (2003). The hippocampus mediates glucocorticoid-
induced impairment of spatial memory retrieval: Dependence on the
basolateral amygdala. Proceedings of the National Academy of Sciences,
USA, 100, 1328–1333.
Roozendaal, B., & McGaugh, J. L. (1997). Basolateral amygdala lesions
block the memory-enhancing effect of glucocorticoid administration in
the dorsal hippocampus of rats. European Journal of Neuroscience, 9,
Roozendaal, B., McReynolds, J. R., & McGaugh, J. L. (2004). The baso-
lateral amygdala interacts with the medial prefrontal cortex in regulating
glucocorticoid effects on working memory impairment. Journal of Neu-
roscience, 24, 1385–1392.
Roozendaal, B., Nguyen, B. T., Power, A. E., & McGaugh, J. L. (1999).
Basolateral amygdala noradrenergic influence enables enhancement of
memory consolidation induced by hippocampal glucocorticoid receptor
activation. Proceedings of the National Academy of Sciences, USA, 96,
Rypma, B., Prabhakaran, V., Desmond, J. E., Glover, G. H., & Gabrieli,
J. D. (1999). Load-dependent roles of frontal brain regions in the
maintenance of working memory. NeuroImage, 9, 216–226.
Sanchez, M. M., Young, L. J., Plotsky, P. M., & Insel, T. R. (2000).
Distribution of corticosteroid receptors in the rhesus brain: Relative
absence of glucocorticoid receptors in the hippocampal formation. Jour-
nal of Neuroscience, 20, 4657–4668.
Seeman, T. E., McEwen, B. S., Singer, B. H., Albert, M. S., & Rowe, J. W.
(1997). Increase in urinary cortisol excretion and memory declines:
MacArthur Studies of Successful Aging. Journal of Clinical Endocri-
nology and Metabolism, 82, 2458–2465.
van Aken, M. O., Romijn, J. A., Miltenburg, J. A., & Lentjes, E. G. W. M.
(2003). Automated measurement of cortisol. Clinical Chemistry, 49,
Veltman, D. J., Rombouts, S. A., & Dolan, R. J. (2003). Maintenance
versus manipulation in verbal working memory revisited: An fMRI
study. NeuroImage, 18, 247–256.
Wechsler, D. (1987). Wechsler Memory Scale—Revised. New York: Har-
court Brace Jovanovich.
Wellman, C. L. (1993). Dendritic reorganization in pyramidal neurons in
medial prefrontal cortex after chronic corticosterone administration.
Journal of Neurobiology, 49, 245–253.
Wolf, O. T., Convit, A., McHugh, P. F., Kandil, E., Thorn, E. L., Santi,
S. D., et al. (2001). Cortisol differentially affects memory in young and
elderly men. Behavioral Neuroscience, 115, 1002–1011.
Wolf, O. T., Kudielka, B. M., Hellhammer, D. H., Hellhammer, J., &
Kirschbaum, C. (1999). Opposing effects of DHEA replacement in
elderly subjects on declarative memory and attention after exposure to a
laboratory stressor. Psychoneuroendocrinology, 23, 617–629.
Young, A. H., Sahakian, B. J., Robbins, T. W., & Cowen, P. J. (1999). The
effects of chronic administration of hydrocortisone on cognitive function
in normal male volunteers. Psychopharmacology, 145, 260–266.
Received August 5, 2004
Revision received September 16, 2004
Accepted September 24, 2004 ?
ACUTE STRESS, CORTISOL, AND WORKING MEMORY