Journal of Anxiety Disorders 22 (2008) 1120–1127

The role of trait anxiety in the recognition of emotional

facial expressions
Robbie M. Cooper *, Angela C. Rowe, Ian S. Penton-Voak
University of Bristol, Bristol, UK
Received 25 July 2007; received in revised form 22 November 2007; accepted 26 November 2007

Abstract
Previous work has suggested that elevated levels of trait anxiety are associated with an increased ability to accurately recognize
the facial expression of fear. However, to date, recognition has only been assessed after viewing periods of 10 s, despite the fact that
the process of emotion recognition from faces typically takes a fraction of this time. The current study required participants with
either high or low levels of non-clinical trait anxiety to make speeded emotional classification judgments to a series of facial
expressions drawn from seven emotional categories. Following previous work it was predicted that recognition of fearful facial
expressions would be more accurate in the high-trait anxious group compared with the low-trait anxious group. However, contrary
to this prediction, no anxiety-related differences in emotion perception were observed across all seven emotions. This suggests that
anxiety does not influence the perception of fear as has been previously proposed.
# 2007 Elsevier Ltd. All rights reserved.

Keywords: Anxiety; Emotion; Facial expressions; Recognition; Perception

1. Introduction production and recognition of emotional facial expres-

The ability to interpret correctly the emotional facial
expressions of others is a skill fundamental to
successful human interaction. It has been suggested
that emotional facial expressions evolved as an external
representation of internal emotional states as a means
for swift communication of these states between
individuals (Blair, 2003). Darwin (1872) believed that
facial expressions are an innate and universal compo-
nent of non-verbal communication. In support of this
view, Ekman (1972) has shown comparable patterns of
* Corresponding author at: Department of Experimental Psychol-
ogy, 12a Priory Road, University of Bristol, Bristol, BS8 1TU, UK.
Tel.: +44 117 928 8547; fax: +44 117 928 8588.
E-mail address: robbie.cooper@bris.ac.uk (R.M. Cooper).
sions across different cultures.
Atypical processing of emotional facial expressions is
thought to be a feature of some neuropsychological and
clinical disorders. For example, recognition of disgust is
impaired in Huntington’s disease (Gray, Young, Barker,
Curtis, & Gibson, 1997). Furthermore, recognition of
anger is thought to be impaired in patients with clinical
depression (Mendlewicz, Linkowski, Bazelmans, &
Philippot, 2005; although see Persad and Polivy (1993)
who present evidence for a general processing deficit of
emotional facial expressions in depression). In contrast to
these processing deficits associated with some condi-
tions, recent evidence has emerged suggesting that
heightened (but non-clinical) levels of trait anxiety are
associated with an ability to more accurately recognize
fearful facial expressions (Surcinelli, Codispoti,
Montebarocci, Rossi, & Baldaro, 2006) relative to lower
levels of trait anxiety. The current paper explores the

0887-6185/$ – see front matter # 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.janxdis.2007.11.010
 R.M. Cooper et al. / Journal of Anxiety Disorders 22 (2008) 1120–1127
relationship between recognition of emotional facial
expressions and individual differences in trait anxiety.
Early cognitive theories of emotional disorders (Beck,
1976; Bower, 1981) predict that anxiety should be
associated with biases favoring the processing of emotio-
nal stimuli across all domains of information processing
(e.g., attention, memory, etc.). However, despite attempts
to find such an information processing bias in memory
(e.g., Mogg, Mathews, & Weinman, 1987) there
currently exists no consensus on the relationship between
memory for emotional stimuli and anxiety (Richards
et al., 2002). Some studies have shown that high levels of
trait anxiety are associated with improved retrieval of
threat-related information compared with neutral infor-
mation (e.g., Claeys, 1989; Eysenck & Byrne, 1994),
while others have found no effects (e.g., Foa, McNally, &
Murdock, 1989; Lang & Craske, 1997). Contrary to this,
there is a large body of evidence suggesting anxiety is
associated with a bias in attentional processing; a bias
that favors the processing of threat-related information
(e.g., Eysenck, 1992; Mathews & Mackintosh, 1998; see
Bar-Haim, Lamy, Pergamin, Bakermans-Kranenburg, &
van Ijzendoorn, 2007 for a recent meta-analysis).
Furthermore, it has been argued that this bias in
attentional processing towards threat stimuli is the most
important cognitive factor in the etiology and main-
tenance of anxiety (Mogg & Bradley, 1998).
While such attentional bias is the most documented
of information processing biases in anxiety (Mathews,
Mackintosh, & Fulcher, 1997) it is not the only reported.
Anxiety is also thought to influence the interpretation of
ambiguous stimuli. For example, when presented with
threat/neutral homophones and asked to make a
classification judgment, high-anxious individuals are
more likely than low-anxious individuals to interpret the
word as threat-related rather than neutral (Mathews,
Richards, & Eysenck, 1989).
Evidence for such anxiety-related biases in attention
and interpretation are also displayed in relation to the
processing of emotional facial expressions. In the dot-
probe task for example,1 individuals with high levels of
1 discrete categories) might be influenced by anxiety.
In its typical form, the dot-probe task presents pairs of stimuli
(e.g., an angry face and a neutral face) for a brief period of time
(typically 500 ms). The allocation of spatial attention between the
stimuli is determined by response times to either identify or localize
the ‘probe’ which appears in the location previously occupied by one
of the stimuli. An attentional bias towards a stimulus category is
assumed to be present if response times to probes occurring in the
location previously occupied by that stimulus type (e.g., angry faces)
are faster than when the probes appear in the location of the other
stimulus type (e.g., neutral faces).
1121
anxiety demonstrate attentional biases towards the
location of both angry (e.g., Bradley, Mogg, Falla, &
Hamilton, 1998) and fearful (Fox, 2002) expressions
relative to expressions that are happy or neutral.2
Furthermore, in an interpretation task, Richards et al.
(2002) presented ambiguous emotional facial expres-
sions that had been constructed by morphing together
two different emotional expressions (e.g., fear and
surprise). When these ambiguous expressions were
presented, high-anxious participants were more likely
than low-anxious participants to identify them as
fearful. Taken together, these studies demonstrate that
at least some aspects of the processing of threat-related
emotional facial expressions (i.e., allocation of atten-
tion; interpretation of ambiguous stimuli) are suscep-
tible to individual differences in anxiety.
However, even in the study of attention there have
been mixed findings in terms of how anxiety influences
the way attentional resources are allocated to emotional
facial expressions. Rossignol, Phillippot, Douilliez,
Crommelinck, and Campanella (2005) gave partici-
pants a visual oddball task where they had to detect a
discrepant fearful or happy face amongst a series of
neutral faces. Participants who rated highly on self-
report measures of anxiety were quicker to identify
discrepant faces than non-anxious participants. How-
ever, fearful faces were detected just as quickly as
happy faces. Similarly, Fox et al. (2000) gave
participants a visual search task in which they had to
search for either a positive or negative schematic face
amongst sets of distracter faces. They reported that
negative faces were found more efficiently than positive
faces but that this effect was not modulated by
participants’ level of self-reported anxiety. Thus, some
tasks demonstrate anxiety-related differences in the way
threat-related emotional facial expressions are pro-
cessed while others do not, suggesting anxiety may only
be influencing certain aspects of the way attention is
allocated (Mathews & Mackintosh, 1998).
More recently, and importantly for purposes of the
present study, research has examined how emotion
recognition (i.e., the ability to classify different
emotional facial expressions as belonging to different
Surcinelli et al. (2006) gave a non-clinical sample of
2
Biases towards threat-related facial expressions have also been
demonstrated in individuals not selected on the basis of anxiety level
(e.g., Cooper & Langton, 2006) but it is the hyper-vigilance to these
stimuli that is thought to characterize anxiety (Mogg & Bradley,
1998).
1122 R.M. Cooper et al. / Journal of Anxiety Disorders 22 (2008) 1120–1127
high- and low-trait anxious participants an emotion
recognition task. Participants were presented with a
series of images of emotional facial expressions, each
for 10 s. Once each image disappeared from the screen,
participants were required to select one of seven
emotional labels (anger, sadness, happiness, fear,
surprise, disgust, and neutral) they thought best
described the face. Surcinelli et al. reported that
recognition accuracy was equivalent across all emo-
tional expressions with the exception of fear. Fearful
faces were more accurately recognized by high-trait
anxious compared with low-trait anxious participants.
There are, however, two potential limitations to
Surcinelli et al.’s (2006) study that question the specifics
of their conclusion that trait anxiety is associated with an
improvement in the ability to recognize fearful emotional
expressions. Firstly, although Surcinelli et al. assume
they are measuring recognition of the emotional facial
expressions in their experiment, the details of their
methodology are such that this assumption may not be
appropriate. The faces in their experiment were presented
for 10 s before a response was allowed. Recognition rates
(i.e., the accuracy with which the emotional facial
expressions are correctly identified) similar to those
reported by Surcinelli et al. can be achieved with
presentation times of only 100 ms (Prkachin, 2003).
Thus, if the process of recognition takes only a fraction of
the time allowed to Surcinelli et al.’s participants, it may
not be recognition per se that is being measured in their
study. For example, deployment of visual attention to
emotional facial expressions is thought to differ between
high- and low-anxious individuals with high-anxious
individuals being less able to disengage their attention
from threat-related faces (Fox, Russo, Bowles, & Dutton,
2001). Across the 10-s presentation time in Surcinelli
et al.’s study, it could be that the high-anxious participants
spend more time attending to the fearful faces than the
low-anxious participants and this difference in attention
could account for the reported difference in recognition.
Furthermore, this relatively long presentation time of
10 s is a concern for reasons of ecological validity. In
evolutionary terms, a 10-s latency for recognition of the
emotional content of a conspecific’s facial expression
would render the system redundant in terms of signaling
imminent danger. Thus, testing recognition rates
following a shorter exposure duration would provide a
more robust test of whether or not trait anxiety influences
the recognition of emotional facial expressions and not
some other cognitive process that might be influenced by
anxiety in the 10 s that elapse before a response is made.
Secondly, and presumably because of the 10-s
exposure duration, only accuracy, and not reaction time,
was recorded in Surcinelli et al.’s (2006) study. Accuracy,
as a measurement, gives only an absolute value for
whether or not the emotions on the faces were recognized
correctly. It does not, for example, provide any
information as to whether elevated levels of anxiety
are associated with faster, more efficient, recognition of
emotional expressions such as fear. There are many
studies of information processing in anxiety that show
differences in choice reaction time as a function of
anxiety level, but no such differences in task accuracy (for
a summary see Eysenck, Derakshan, Santos, & Calvo,
2007). Thus, reaction time to recognize the displayed
emotion might be a more sensitive measure to assess the
influence of anxiety on the processing of emotional facial
expressions compared with accuracy alone.
The Surcinelli et al. (2006) study is, to our knowledge,
the only one to date that has examined recognition of
emotional facial expressions as a function of non-clinical
trait anxiety.3 However, similar studies using word and
picture stimuli have provided equivocal findings. Across
three experiments, and against their predictions, Math-
ews and Milroy (1994) found no evidence that anxious
participants were any better at making a speeded valence
judgment concerning negative words (e.g., funeral,
stupid) than positive words (e.g., healthy, clever) when
compared with non-anxious participants. In contrast,
Bradley and Lang (1999) required participants to perform
a very similar task to Mathews and Milroy, this time in
response to the display of pleasant (e.g., flowers) and
unpleasant (e.g., a snarling dog) images. They found
anxious participants were faster at identifying the
unpleasant images compared with non-anxious partici-
pants but there was no difference in participants’ ability
to identify the pleasant images.
Given the equivocal nature of findings in this area with
non-face stimuli, we look again at the relationship
between anxiety and the recognition of emotional facial
expressions. The current study attempts to extend the
work conducted by Surcinelli et al. (2006). A non-clinical
sample of high- and low-trait anxious participants were
asked to perform a speeded emotional facial expression
recognition task in which both response time and
accuracy were recorded. Following findings of Surcinelli
et al. it was predicted that participants with high self-
reported levels of trait anxiety should be more accurate in
3
Two other studies have shown that anxiety modulates the ERP
signal that arises from seeing emotional facial expressions (Rossignol,
Anselme, Vermeulen, Philippot, & Campanella, 2007; Rossignol
et al., 2005) but recognition of emotion was not directly measured
in these studies.
 R.M. Cooper et al. / Journal of Anxiety Disorders 22 (2008) 1120–1127
the recognition of fear compared to those with low levels
of trait anxiety. In addition, following the work of
Bradley and Lang (1999), response times to identify
fearful emotional expressions should be faster in
individuals with elevated levels of trait anxiety compared
with individuals with low levels of trait anxiety.
2. Method
2.1. Participants
A total of 109 undergraduate students (58 female)
completed both the state and trait sections of the
Spielberger State-Trait Anxiety Inventory (STAI;
Spielberger, Gorsuch, Lushene, Vagg, & Jacobs,
1983). Only participants who scored above the 75th
percentile or below the 25th percentile on the trait
component of the STAI were selected, giving 54
participants, 27 in each of the high-trait (M = 53.1,
S.D. = 5.0, 18 females) and low-trait (M = 29.3,
S.D. = 3.0, 9 females) anxiety groups.
2.2. Materials
Seventy images from the Ekman and Friesen (1976)
pictures of facial affect series were selected. These
comprised seven images each from 10 individuals (5
female), each individual posing seven expressions
(anger, disgust, fear, happiness, neutral, sadness, and
surprise). The images were cropped to show only the
face (measuring 12.5 cm  16 cm) and were viewed
from approximately 60 cm. Image presentation was
controlled by E-prime software running on a Dell
Optiplex desktop PC and displayed on a 15-in. CRT
monitor set to 85 Hz.
2.3. Procedure
Images were presented individually in a randomized
order with each image displayed centrally on the screen.
By means of a key-press, participants’ task was to quickly
and accurately identify which one of seven emotions was
being represented in the facial image.4 Each emotional
4
The 70 images from the Ekman and Friesen (1976) set were
interspersed amongst 105 images from a new set of emotional facial
expressions (Benton, Clark, Cooper, Penton-Voak, & Nikolov, 2007)
with the goal of comparing recognition accuracy across the two sets.
However, given that the Benton et al. set had not previously been
validated, and to maintain similarity with Surcinelli et al. (2006), the
data pertaining to these faces is not presented here. We return to this
issue in the discussion.
1123
term was assigned to a numeric key on the keyboard.
Half of the participants used one key assignment
(1 = angry, 2 = disgust, 3 = fear, 4 = happy, 5 = neutral,
6 = sad, 7 = surprise) while the other half used another
(1 = sad, 2 = surprise, 3 = neutral, 4 = fear, 5 = disgust,
6 = angry, 7 = happy). These emotional terms and their
associated key assignments were displayed on screen
beneath the image throughout the experiment. Each
image remained on the screen until a response was
detected or 4 s elapsed (whichever was sooner).
Experimental trials were preceded by six practice trials.
3. Results
Two participants were excluded due to accuracy
being at, or below, chance level. Another participant
was excluded after boxplots identified their response
times as outliers. This left 51 participants (26 high-
anxious). The two groups differed in trait-anxiety score,
t(49) = 21.0, p < .001, as well as state anxiety score,5
t(46) = 5.4, p < .001.
3.1. Accuracy6
Mean recognition accuracy scores are displayed in
Fig. 1. A 2 (trait anxiety group: high vs. low)  7
(emotional facial expression) repeated measures analysis
of variance (ANOVA) revealed a main effect of emotion,
F(6,294) = 35.6, p < .001. Post-hot tests with Sidak
corrections revealed that happiness was recognized more
accurately than all other emotional expressions (all
ps < .01), with the exception of surprise. Similarly,
surprise was recognized more accurately than all other
expressions with the exception of happiness and neutral
(all ps < .01). In turn, neutral expressions were
recognized more accurately than all other expressions
with the exception of happiness and surprise (all
ps < .01). Recognition accuracy for anger, disgust, fear,
and sadness did not differ significantly ( ps > .05).
Importantly, the predicted trait anxiety level  emotional
facial expression interaction was not significant ( p > .6).
5
Three participants failed to complete the state component of the
STAI.
6
Recognition tasks such as this one can be subject to signal
detection analysis (e.g., Prkachin, 2003). This produces a new depen-
dent variable (d0 or A0 ) which takes into account both accurate (hits)
and inaccurate (false alarms) responses, therefore giving a measure of
sensitivity. To allow easier comparison with Surcinelli et al. (2006)
this analysis is not presented here. However, consistent with the
analysis of the accuracy scores alone, A0 analysis did not reveal
any anxiety-related modulation in the recognition of the various
emotional facial expressions, including fear.
1124 R.M. Cooper et al. / Journal of Anxiety Disorders 22 (2008) 1120–1127
Fig. 1. Mean accuracy (%) to identify each emotional facial expression as a function of trait anxiety level (error bars represent 1 standard error of the
mean).
These results are in contrast with those of Surcinelli
et al. (2006) who found that participants who rated high
in trait anxiety were more accurate at recognizing
fearful facial expressions compared with low-anxious
participants. One possible explanation for the discre-
pancy in the results of the present study and those of
Surcinelli et al. is a difference in anxiety scores across
the two populations tested. In support of this
interpretation, the high-anxious sample tested by
Surcinelli et al. were, at least numerically, more
anxious than the current sample (mean = 57.1 and
53.1 respectively). In order to provide a further test of
the possibility that an increased ability to accurately
recognize fearful facial expressions is only present in
those participants who rate most highly in trait anxiety a
further analysis was conducted on 20 participants; those
with the 10 highest (M = 58.3, S.D. = 4.2) and lowest
(M = 26.0, S.D. = 1.6) trait anxiety scores. Recognition
accuracy for fear faces was compared directly in these
two groups. This analysis revealed no evidence that
high-trait anxious individuals recognize fearful facial
expressions more accurately than low-trait anxious
individuals, t(18) = .65, p = .26 (one-tailed).7
7
We also carried out the same analysis replacing trait anxiety scores
with state anxiety scores. Again, no anxiety-related differences were
observed and so these data are not reported here.
3.2. Response time
The inter-participant means of median correct
response times are displayed in Fig. 2. A 2 (trait
anxiety group: high vs. low)  7 (emotional facial
expression) mixed ANOVA revealed a main effect of
emotion, F(6,294) = 32.9, p < .001. Post-hoc tests with
Sidak corrections revealed that recognition of happy
and neutral faces was faster than all other expressions
(all ps < .007) but response time did not differ between
the two expressions ( p = .11). Latency to identify
disgust, sadness, and surprise did not differ (all
ps < .15) and all three were recognized more quickly
than both fear and anger (all ps < .003), the recognition
time for which did not differ ( p > .11). Importantly, and
in keeping with the accuracy data, the predicted anxiety
level  emotional facial expressions interaction for
response time was not significant ( p > .5).
4. Discussion
The main goal of the present study was to investigate
how self-reported levels of trait anxiety influence the
ability to recognize emotional facial expressions,
particularly fear. It was predicted that participants with
high levels of trait anxiety would demonstrate greater
accuracy in recognizing the facial expression of fear
compared with participants with low levels of trait
anxiety. However, the main finding of the present study
 R.M. Cooper et al. / Journal of Anxiety Disorders 22 (2008) 1120–1127
Fig. 2. Mean response time (ms) to identify each emotional facial expression as a function of trait anxiety level (error bars represent 1 standard error
of the mean).
is that participants’ trait anxiety level did not affect the
accuracy or speed with which fear, or any of the other
presented facial expressions (happy, sad, angry,
disgusted, surprised, and neutral), were recognized.
This is despite overall recognition rates showing a very
similar pattern to those observed by Surcinelli et al.
(2006).
The failure to observe any modulation of recognition
rates for emotional facial expressions as a function of
anxiety level contrasts sharply with the findings of
Surcinelli et al. (2006) who reported that recognition
accuracy of fearful facial expressions improves at
elevated levels of trait anxiety. There are, however, a
number of differences between the current study and
that of Surcinelli et al. that could account for this
discrepancy: We now discuss each in turn:
(1) The trait anxiety level of the participants in
Surcinelli et al.’s (2006) sample was marginally
higher than that in the current study. This holds
for both low-anxious (29.3 vs. 31.2) and high-
anxious (53.1 vs. 57.1) participants. Given that
Surcinelli et al.’s results suggest that higher levels
of trait anxiety are associated with greater
recognition accuracy of fearful facial expressions
it is possible that an increased level of trait
anxiety in the Surcinelli et al. sample might go
some way to account for the discrepant findings.
However, even when the current study compared
1125
the 10 participants who score most highly on trait
anxiety with the lowest scoring 10 participants on
a one-tailed test, there was no evidence to suggest
recognition of fearful facial expressions is
modulated by anxiety.
(2) The current study also differs from Surcinelli
et al. (2006) in terms of the number of
participants tested and the number of stimuli
presented to participants. However, the current
study included greater numbers of each. There
were more participants (54 vs. 39) and more
exemplars of each emotional face stimulus (10
vs. 7) than the Surcinelli et al. study. Increased
numbers of participants and stimuli should lead
to increased power, and thus reduce the like-
lihood of a Type I error. Despite this, there was no
evidence to suggest recognition of fearful facial
expressions is modulated by anxiety.
(3) One potential criticism of the current study is that
interspersed among the test images were 105
other images of emotional facial expressions;
part of a set that were being validated. It could be
argued that the inclusion of these extra stimuli
may have contributed to the lack of any anxiety-
related differences in the current data through,
perhaps, fatigue effects. However, while the
current study did present more stimuli than those
seen in Surcinelli et al.’s (2006) study, they were
each displayed for approximately 2 s as opposed
1126 R.M. Cooper et al. / Journal of Anxiety Disorders 22 (2008) 1120–1127
to 10 s in the Surcinelli et al. study meaning the
current experiment would have actually been
shorter than that of Surcinelli et al. Furthermore,
experiments looking at the allocation of attention
in anxiety tend to have procedures that contain
hundreds of trials and these are sensitive to
differences in anxiety (e.g., Bradley et al., 1998).
Accordingly, while not being able to rule out this
potential criticism, it does seem unlikely that it
can account for the difference in findings
between the two studies.
(4) The biggest difference between the current study
and that of Surcinelli et al. (2006) (and the one we
believe best accounts for the difference in the
findings of the two experiments) is the way in
which recognition was measured. Images in
Surcinelli et al.’s experiment were presented for
10 s with participants being asked to respond as
accurately as possible under no time pressure. In
contrast the images in the current study were
presented until the participant responded (an
average of 2059 ms across expressions) follow-
ing instructions to identify the faces as quickly
and accurately as possible. Thus, while both the
current study and that of Surcinelli et al. claim to
be measuring the recognition of emotional facial
expressions, this may not be the case. The relative
improvement in the ability of the high-anxious
group to accurately respond to the presence of
fearful facial expressions in Surcinelli et al.’s
study may be due to the contribution of another
cognitive process that is influencing responses
later in the time course of stimulus presentation.
For example, in Surcinelli et al.’s study, it is
possible that the high-trait anxious group spent
longer attending to the fearful faces compared to
the low-anxious group (c.f., Fox et al., 2001), and
it is this increased attention that might account
for better performance on the subsequent
recognition task. A future experiment could
directly test this idea by measuring eye move-
ments when performing Surcinelli et al.’s version
of the task.
The crucial point here is that the apparent
improvement in recognition performance demon-
strated by the high-anxious group in Surcinelli
et al.’s (2006) experiment is not necessarily due to
anxiety directly influencing the recognition stage of
processing. Thus, while there is good evidence
suggesting the allocation of attention to threat-related
facial expressions is modulated by anxiety (e.g., Bar-
Haim et al., 2007) the process of recognizing those
expressions may not be similarly modulated by
anxiety. This finding goes against early cognitive
theories of emotional disorders (Beck, 1976; Bower,
1981) which predict that anxiety should be associated
with biases favoring the processing of emotional
stimuli across all domains of information processing.
However, it is consistent with other evidence
suggesting processing domains such as memory are
relatively unaffected by participants’ level of anxiety
(e.g., Lang and Craske, 1997). The data are also
consistent with those obtained by Mathews and
Milroy (1994) who found no anxiety-related differ-
ences in the classification of emotional words. We
echo Mathews and Milroy’s conclusion that these data
suggest that the processes required to make a
conscious assessment of a stimuli’s valence are
distinct from those that lead to biases in attention
(e.g., Fox et al., 2001) and interpretation (e.g.,
Richards et al., 2002), and the former does not appear
to be not influenced by anxiety, at least not to the
same extent.
However, the current results are somewhat incon-
sistent with the results of Bradley and Lang (1999) who
found that non-clinical high-anxious participants were
faster at making a valence decision to negative images
than were low-anxious participants. Unfortunately,
differences in methodology between the current study
and that of Bradley and Lang make it difficult to directly
compare these experiments.8
What is clear is that this area would benefit from a
more detailed look at the influence of trait anxiety on the
recognition of emotional stimuli in general. The current
experiment presents a clear challenge to the notion that
the recognition of emotional facial expressions is
modulated by elevated levels of trait anxiety. On the
basis of the extant data overall, the evidence for a
general anxiety-related processing bias at the level of
stimulus recognition (be it word, picture, or face) is far
from compelling.
Acknowledgement
This work was funded by Grant RES 062-23-0052
from the ESRC.
8
Bradley and Lang (1999) employed a two-alternative forced
choice task (i.e., decide whether the picture is positive or negative)
while the current study used a seven-alternative forced choice task
(i.e., choose one of seven possible emotions). It is possible that the
former task is more sensitive to picking up anxiety-related differences
in emotional stimuli at shorter presentation times.
 R.M. Cooper et al. / Journal of Anxiety Disorders 22 (2008) 1120–1127
References
Bar-Haim, Y., Lamy, D., Pergamin, L., Bakermans-Kranenburg, M., &
van Ijzendoorn, M. (2007). Threat-related attentional bias in
anxious and nonanxious individuals: a meta-analytic study. Psy-
chological Bulletin, 133(1), 1–24.
Beck, A. T. (1976). Cognitive therapy and the emotional disorders.
New York: International Universities Press.
Benton, C., Clark, A., Cooper, R. M., Penton-Voak, I., & Nikolov, S.
(2007). Different views of facial expressions: an image sequence
dataset. Poster presented at the Vision Sciences Society 7th
Annual Meeting, Sarasota, US.
Blair, R. J. R. (2003). Facial expressions, their communicatory func-
tions and neuro-cognitive substrates. Philosophical Transactions
of the Royal Society of London Series B – Biological Sciences,
358(1431), 561–572.
Bower, G. H. (1981). Mood and memory. American Psychologist, 36,
129–148.
Bradley, M. M., & Lang, P. J. (1999). Fearfulness and affective picture
evaluations of pictures. Motivation and Emotion, 23(1), 1–13.
Bradley, B. P., Mogg, K., Falla, S. J., & Hamilton, L. R. (1998).
Attentional bias for threatening facial expressions in anxiety:
manipulation of stimulus duration. Cognition and Emotion,
12(6), 737–753.
Claeys, W. (1989). Social anxiety, evaluative threat, and incidental
recall of trait words. Anxiety Research, 2, 27–43.
Cooper, R. M., & Langton, S. R. H. (2006). Attentional bias to angry
faces using the dot-probe task? It depends when you look for it.
Behaviour Research and Therapy, 44, 1321–1329.
Darwin, C. (1872). The expression of the emotions in man and the
animals. London: John Murray.
Ekman, P. (1972). Universals and cultural differences in facial expres-
sions of emotion. In: J. Cole (Ed.), Nebraska symposium on
motivation, 1971 (pp. 207–282). Lincoln: University of Nebraska
Press.
Ekman, P., & Friesen, W. V. (1976). Pictures of facial affect. Palo Alto,
CA: Consulting Psychologists Press.
Eysenck, M. W. (1992). Anxiety: the cognitive perspective. Hove,
England: Erlbaum.
Eysenck, M. W., & Byrne, A. (1994). Implicit memory bias, explicit
memory bias, and anxiety. Cognition and Emotion, 8, 415–431.
Eysenck, M. W., Derakshan, N., Santos, R., & Calvo, M. G. (2007).
Anxiety and cognitive performance: attentional control theory.
Emotion, 7(2), 336–353.
Foa, E. B., McNally, R. J., & Murdock, T. B. (1989). Anxious mood
and memory. Behaviour Research and Therapy, 27, 141–147.
Fox, E. (2002). Processing emotional facial expressions: the role of
anxiety and awareness. Cognitive, Affective, & Behavioral Neu-
roscience, 2(1), 52–63.
Fox, E., Lester, V., Russo, R., Bowles, R. J., Pichler, A., & Dutton, K.
(2000). Facial expressions of emotion. Are angry faces detected
more efficiently? Cognition and Emotion, 14, 61–92.
1127
Fox, E., Russo, R., Bowles, R., & Dutton, K. (2001). Do threatening
stimuli draw or hold visual attention in subclinical anxiety?
Journal of Experimental Psychology: General, 130(4), 681–700.
Gray, J. M., Young, A. W., Barker, W. A., Curtis, A., & Gibson, D.
(1997). Impaired recognition of disgust in Huntington’s disease
gene carriers. Brain, 120, 2029–2038.
Lang, A. J., & Craske, M. G. (1997). Information processing in
anxiety and depression. Behaviour Research and Therapy, 35,
451–455.
Mathews, A., & Mackintosh, B. (1998). A cognitive model of selective
processing in anxiety. Cognitive Therapy and Research, 22(6),
539–560.
Mathews, A., Mackintosh, B., & Fulcher, E. P. (1997). Cognitive
biases in anxiety and attention to threat. Trends in Cognitive
Sciences, 1(9), 340–345.
Mathews, A., & Milroy, R. (1994). Processing of emotional meaning
in anxiety. Cognition and Emotion, 8, 535–553.
Mathews, A., Richards, A., & Eysenck, M. (1989). Interpretation of
homophones related to threat in anxiety states. Journal of Abnor-
mal Psychology, 98(1), 31–34.
Mendlewicz, L., Linkowski, P., Bazelmans, C., & Philippot, P. (2005).
Decoding emotional facial expressions in depressed and anorexic
patients. Journal of Affective Disorders, 89, 195–199.
Mogg, K., & Bradley, B. P. (1998). A cognitive-motivational
analysis of anxiety. Behaviour Research and Therapy, 36(9),
809–848.
Mogg, K., Weinman, J., & Mathews, A. (1987). Memory bias in
clinical anxiety. Journal of Abnormal Psychology, 96(2), 94–98.
Persad, S. M., & Polivy, J. (1993). Differences between depressed and
nondepressed individuals in the recognition of and response to
facial emotional cues. Journal of Abnormal Psychology, 102, 358–
368.
Prkachin, G. C. (2003). The effects of orientation on detection and
identification of facial expressions of emotion. British Journal of
Psychology, 94(1), 45–62.
Richards, A., French, C. C., Calder, A. J., Webb, B., Fox, R., & Young,
A. (2002). Anxiety-related bias in the classification of emotionally
ambiguous facial expressions. Emotion, 2(3), 272–287.
Rossignol, M., Anselme, C., Vermeulen, N., Philippot, P., & Campa-
nella, S. (2007). Categorical perception of anger and disgust facial
expression is affected by non-clinical social anxiety: an ERP
study. Brain Research, 1132, 116–176.
Rossignol, M., Phillippot, P., Douilliez, C., Crommelinck, M., &
Campanella, S. (2005). The perception of fearful and happy facial
expression is modulated by anxiety: an event-related potential
study. Neuroscience Letters, 377(2), 115–220.
Spielberger, C. D., Gorsuch, R. L., Lushene, R., Vagg, P. R., & Jacobs,
G. A. (1983). Manual for the State-Trait Anxiety Inventory (Form
Y). Palo Alto, CA: Consulting Psychologists Press.
Surcinelli, P., Codispoti, M., Montebarocci, O., Rossi, N., & Baldaro,
B. (2006). Facial emotion recognition in trait anxiety. Journal of
Anxiety Disorders, 20, 110–117.