Hypervigilance–avoidance pattern in
Tobias Pflugshaupta,1, Urs P. Mosimanna,
Roman von Wartburga, Wolfgang Schmittb,2,
Thomas Nyffelera, Rene ´ M. Mu ¨ria,*
aPerception and Eye Movement Laboratory, Departments of Neurology and Clinical Research,
University of Berne, Inselspital, Freiburgstrasse 10, 3010 Berne, Switzerland
bDepartment of Psychiatry, University of Berne, Inselspital,
Freiburgstrasse 10, 3010 Berne, Switzerland
Received 4 August 2003; received in revised form 24 November 2003; accepted 3 December 2003
Cognitive-motivational theories of phobias propose that patients’ behavior is char-
acterized by a hypervigilance–avoidance pattern. This implies that phobics initially direct
their attention towards fear-relevant stimuli, followed by avoidance that is thought to
prevent objective evaluation and habituation. However, previous experiments with highly
anxious individuals confirmed initial hypervigilance and yet failed to show subsequent
avoidance. In the present study, we administered a visual task in spider phobics and
controls, requiring participants to search for spiders. Analyzing eye movements during
visual exploration allowed the examination of spatial as well as temporal aspects of phobic
behavior. Confirming the hypervigilance–avoidance hypothesis as a whole, our results
showed that, relative to controls, phobics detected spiders faster, fixated closer to spiders
during the initial search phase and fixated further from spiders subsequently.
# 2003 Elsevier Inc. All rights reserved.
Keywords: Human; Attention; Eye movements; Phobic disorders
19 (2005) 105–116
*Corresponding author. Tel.: þ41-31-632-3081; fax: þ41-31-632-9679.
E-mail address: firstname.lastname@example.org (R.M. Mu ¨ri).
1Tel.: þ41-31-632-3368; fax: þ41-31-632-9679.
2Tel.: þ41-31-632-8811; fax: þ41-31-632-8950.
0887-6185/$ – see front matter # 2003 Elsevier Inc. All rights reserved.
Fear is an aversive emotional state that serves a protective purpose. It signals
danger and prepares us to deal with it. However, for people with specific phobias,
fear is pathological and thus interferes with a person’s ability to cope with
everyday situations. In this case, fear is related to the presence or anticipation of a
particular object or situation, and patients are usually aware of how unreasonable
their fear is (American Psychiatric Association, 2000). Phobic behavior typically
includes avoidance, which is supposed to have two main functions. On the one
hand, avoidance enables patients to reduce anxious mood states by keeping
encounters with fear-relevant stimuli to a minimum (Thorpe & Salkovskis, 1999).
On the other hand, the same behavior is regarded as a major factor in maintaining
phobia, since avoidance prevents habituation to, or objective evaluation of, fear-
related stimuli (Mogg, Bradley, De Bono, & Painter, 1997).
Numerous studies have shown that there is an attentional bias in processing
fear-related information and that this bias is particularly pronounced in phobics
(see Mogg & Bradley, 1998 for a review). For example, Ohman, Flykt, and
Lundqvist (2000) provided evidence that healthy subjects detect fear-relevant
stimuli such as snakes or spiders faster than fear-irrelevant ones. This ‘snake-in-
the-grass effect’ is enhanced in phobics, depending on the nature of their disorder
(e.g., spider phobics were faster in detecting spiders as opposed to snakes). The
authors attributed their findings to preattentive processes that are triggered by
biologically prepared aversive stimuli. Once a threat stimulus is detected,
attention automatically shifts to its location. In phobics, the aversive emotion
is thought to accelerate detection. A recent study reported that, compared to
control subjects, spider phobics more often perceived their responses to spiders as
automatic, i.e., not under intentional control (Mayer, Merckelbach, & Muris,
2000). This finding highlights the preattentive and automatic nature of the
mechanisms that direct the attention of phobics towards fear-relevant stimuli.
Fast detection abilities of phobics have commonly been attributed to the more
general concept of hypervigilance, which describes the tendency to constantly
scan the environment for any signs of threat (Thorpe & Salkovskis, 1999). The
prefix hyper indicates that, although most people are vigilant for potential threat,
this vigilance is strongly enhanced in phobics. As fast identification of threat
allows early activation of defenses (Ohman et al., 2000), hypervigilance may
serve a protective purpose. However, it is regarded as an important factor in the
acquisition of pathological fear as well (Williams, Watts, MacLeod, & Mathews,
1997). Individuals who are permanently scanning the environment for threat may
be more likely to perceive the world as a dangerous place, which in turn enhances
their anxious mood and might increase the probability that they develop patho-
logical anxiety (Mogg & Bradley, 1998).
So far, two aspects of phobic behavior with regardto fear-relevant stimuli have
been described: avoidance and hypervigilance. It has been assumed that everyday
routines of phobics are influenced by both tendencies (e.g., Lang, Bradley, &
106T. Pflugshaupt et al./Anxiety Disorders 19 (2005) 105–116
One might ask how phobics manage to avoid something that permanently attracts
their attention. Cognitive-motivational theories (e.g., Mathews, 1990; Mogg,
Mathews, & Weinman, 1987) postulate that hypervigilance and avoidance co-
such stimuli in an attempt to reduce their anxious mood state (Mogg et al., 1987).
This hypervigilance–avoidance hypothesis was investigated in non-clinical
anxiety (Mogg et al., 1997) and in subjects with general anxiety disorder (Bradley
et al., 1999). Both studies were based on modified versions of the dot probe task.
During a typical dot probe task, pairs of stimuli (e.g., a threat word and a neutral
word) are presented on a screen, followed by a small dot probe that appears in the
location of one of the stimuli. Participants are required to respond manually to the
probe as quickly as possible. The idea behind this task is that response latencies
will be accelerated when probes occur in an attended, rather than unattended part
of the display (Posner, Snyder, & Davidson, 1980). In the two studies mentioned
above, the duration of stimulus exposure was manipulated and the focus of
analysis was on two predictions. First, due to initial hypervigilance, anxious
individuals should respond faster to probes replacing briefly presented threat
stimuli than healthy subjects. Second, as a consequence of subsequent avoidance,
it was expected that anxious individuals respond more slowly to these probes than
controls when longer stimulus durations were applied. Both studies found ample
evidence for hypervigilance in anxious individuals. However, the same subjects
did not show avoidance after longer stimulus durations.
This absence of evidence for avoidance can be interpreted in several ways.
Mogg et al. (1997) and Bradley et al. (1999) provided two explanations. First, it
was assumed that the hypervigilance–avoidance hypothesis might be incorrect.
Results indicated that both initial shifting and subsequent maintenance of
attention were consistently biased towards threat information in anxious indivi-
duals. Thus, a simple hypervigilance hypothesis would explain the results
sufficiently. Second, the stimuli may not have been anxiety-provoking enough
to prompt avoidance strategies. In our opinion, there are two further explanations.
First, the applied duration of stimulus exposure (i.e., up to 1500 ms) was possibly
not long enough to allow for elaborative processing that prompts avoidance
behavior. Second, the dot probe task may not be the method of choice to
investigate hypervigilance–avoidance patterns. Researchers cannot observe
hypervigilance and avoidance behavior per se in this task, since both tendencies
have to be inferred from manual reaction times. In order to investigate them in a
more direct manner, we used eye movement measurements to analyze phobic
behavior. Eye movement tracking allows the registration of temporal as well as
spatial aspects of fixations and saccades, which in turn are closely related to
attentional mechanisms (e.g., Kowler, Anderson, Dosher, & Blaser, 1995). Thus,
eye movement analysis is particularly suitable for investigating attentional biases
such as hypervigilance or avoidance.
T. Pflugshaupt et al./Anxiety Disorders 19 (2005) 105–116107
The main objective of the present study was to empirically validate the
hypervigilance–avoidance hypothesis (Mogg et al., 1987) in phobics during a
visual task that required participants to search for fear-relevant stimuli. Due to a
relatively high prevalence rate (Szymanski & O’Donohue, 1995), we decided to
concentrate on spider phobia. Data analysis was focussed on variables that
represent hypervigilance or avoidance. As a consequence and based on the
hypervigilance–avoidance hypothesis, three main predictions were investigated:
(a) phobics detect spiders faster than controls; (b) relative to controls, phobics’
fixations are spatially closer to spiders during the initial search phase; and (c)
phobics subsequently fixate further from spiders than controls.
Twenty-six spider phobics (22 females, 4 males; age in years: mean ¼ 28:04,
S:D: ¼ 8:92;20 right-handers, 6left-handers)and26controlsubjects(22females,
4males;ageinyears:mean ¼ 28:73,S:D: ¼ 4:41;23right-handers,3left-handers)
volunteered for the study. Participants were assigned to groups according to self-
evaluation. Standardized phobia questionnaires (see the following) allowed the
validation of this self-evaluative assignment. Exclusion criteria were depressive
mood states (Beck Depression Inventory/BDI; Hautzinger, Bailer, Wollall, &
Keller, 1994), color blindness (Ishihara’s color plates; Ishihara, 1999), visual
disturbances (e.g., ocular motor paralysis, strabism), and limited visual acuity
(i.e., normal or corrected to normal). The local ethical committee approved the
study. All participants gave written informed consent prior to the examination.
2.2. Spider phobia questionnaires
Theextenttowhich participants suffered fromspider phobiawas assessed with
two questionnaires, the Watts and Sharrock’s Spider Fear Questionnaire (WSQ)
(Watts & Sharrock, 1984) and the Fear of Spiders Questionnaire (FSQ) (Szy-
response. For example, participants are asked whether they are always on the
lookout for spiders. With an internal consistency of .78 (Johnsen & Hugdahl,
1990) and a test–retest reliability of .94 (Muris & Merckelbach, 1996), the WSQ
has demonstrated adequate psychometric properties. The FSQ is an 18-item
questionnaire and each item is rated on a seven-point likert-type scale (i.e.,
?3 ¼ strongly disagree, þ3 ¼ strongly agree). For instance, participants have to
rate the following statement: If I came across a spider now, I would leave the
room. With regard to psychometric properties, the FSQ has demonstrated an
internal consistency of .92 (Szymanski & O’Donohue, 1995) and a test–retest
reliability of .91 (Muris & Merckelbach, 1996).
108T. Pflugshaupt et al./Anxiety Disorders 19 (2005) 105–116
Eye movements were recorded using a video-based tracking system (Eye-
LinkTM, SensoMotoric Instruments GmbH, Berlin, Germany). This system
collects eye movement data at a sampling rate of 250 Hz with a spatial resolution
of 0.018. It provides a gaze-position accuracy relative to stimulus coordinates of
0.5–1.08, depending largely on participants’ fixation accuracy during calibration.
A chin rest was used both to ensure constant distance and minimize head
movements. Furthermore, head motion compensation was calculated by the
tracking system. Participants were seated 70 cm in front of a 19-in. screen,
resulting in a visual angle of approximately 29?? 228. Eye movements of either
the right or the left eye were recorded. The system was calibrated prior to each
block of stimuli.
2.4. Search task
During the search task, participants were confronted with 16 everyday scenes
onto each of which 1, 2, or 3 identical black spiders were placed as targets. Blocks
of four scenes were formed to allow for recalibration in-between. Scenes covered
the entire screen (i.e., 29?? 228) and spider targets were 1:81?? 1:278 in size.
Each scene included two horizontal black bars, one at the top and the other at the
bottom, in order to maintain the original aspect ratio of the pictures. Six scenes
with one target, four scenes with two targets, and six scenes with three targets
were applied,resulting inatotalnumberof32targets.Inordertodistributetargets
in a balanced manner, scenes were symmetrically subdivided into 16 rectangles,
each of which was used twice as target location. Prior to each scene, a central
fixation point was shown. Every scene was presented for 5 s, immediately
followed by a forced-choice task, during which participants had to decide and
click whether they had seen 0, 1, 2, or 3 spiders. An example is given in Fig. 1.
2.5. Experimental procedure
The experimental procedure was held constant across groups. After partici-
pants had given informed consent, the two spider phobia questionnaires were
administered, followed by the search task that was introduced with verbal
instructions and two practice trials. Eye movements were recorded while parti-
cipants sat inadimly litroom.Theaverage duration ofthe entireexamination was
approximately 45 min.
2.6. Data analysis
The main focus of data analysis was on the three predictions inferred from the
hypervigilance–avoidance hypothesis. The duration from stimulus onset to the
first fixation-on-a-spider allowed investigating whether phobics detect spiders
T. Pflugshaupt et al./Anxiety Disorders 19 (2005) 105–116109
faster than controls. A fixation was considered to be on a spider if the distance
between fixation coordinates and the center ofthe spider was less than 38 ofvisual
angle. The second and third prediction could be examined by calculating the
distanceto the nearest spider foreach fixationand by analyzing thegroup-specific
time course of these distances.
A second focus of analysis was on search performance and basic oculomotor
parameters. On the one hand, we were interested in whether the number of
omissions indicated by the results from the forced-choice task differed signifi-
cantly betweengroups.In doing this, the interrelation of oculomotor behavior and
Fig. 1. An example stimulus of the visual search task. Everyday scene with two targets (top),
superimposed white arrows indicate target locations, left target magnified for enhanced visibility;
forced-choice task (bottom).
110 T. Pflugshaupt et al./Anxiety Disorders 19 (2005) 105–116
search performance could be examined. On the other hand, two basic oculomotor
parameters (i.e., fixation duration, saccade amplitude) were used to analyze
general aspects of oculomotor behavior.
Since relevant eye movement parameters did not deviate from normal dis-
tribution (Kolmogorov–Smirnov tests), parametric tests (i.e., t-tests, ANOVAs)
were performed for group comparisons. For the same reason, means and standard
deviations will be presented as central tendency and dispersion measures. A P-
value of less than .05 was considered statistically significant and all tests were
3.1. Spider phobia questionnaires
from controls. Therefore, the self-evaluative assignment of participants to groups
(i.e., phobics vs. controls) was validated by means of standardized, clinically
relevant questionnaires. Controls’ relatively low a-scores indicate that their
performance was rather randomly distributed in both questionnaires, whereas
phobics showed a more consistent response pattern. Overall, our results are
comparable with those Watts and Sharrock (1984) and Szymanski and O’Dono-
hue (1995) obtained in their studies.
3.2. Search performance and basic oculomotor parameters
Table 2 shows that the two groups did not differ significantly from each other
with regard to the number of omissions in the forced-choice task. Thus, whatever
Spider phobia questionnaires
Phobics (n ¼ 26)Controls (n ¼ 26)t-test (df ¼ 50)
Means, standard deviations and Cronbach’s a of the WSQ (Watts & Sharrock, 1984) and the FSQ
(Szymanski & O’Donohue, 1995).
***Significant at the .001 level.
T. Pflugshaupt et al./Anxiety Disorders 19 (2005) 105–116111
the following group differences in eye movement behavior may be, they did not
result in differential search performance. However, basic oculomotor behavior
was group-specific. Phobics made slightly shorter fixations and significantly
longer saccades than controls.
3.3. Hypervigilance–avoidance hypothesis
With regard to the first of our main predictions (i.e., phobics detect spiders
faster than controls), we compared the duration from stimulus onset to the first
fixation-on-a-spider between groups. Table 3 shows that this duration was
significantly shorter in phobics compared with controls. Thus, the first fixa-
tion-on-a-spider occurred earlier in phobics, indicating the expected hypervigi-
The second and third prediction could be examined by calculating the distance
to the nearest spider for each fixation and by analyzing the group-specific time
courseof these distances.In order to examinetimecoursesin detail,we decided to
after 2946 ms(phobics: mean ¼ 2902 ms,
mean ¼ 2991 ms, S:D: ¼ 491 ms). A two-way ANOVA with time (i.e., first to
vs. controls) as a between-subject factor revealed a significant main effect of time
(Fð9;450Þ ¼ 7:454, P < :001) and a significant interaction ‘time ? group’
(Fð9;450Þ ¼ 3:029,P < :001).Fig.2 illustrates that bothofourpredictionscould
S:D: ¼ 406 ms;controls:
Search performance and basic oculomotor parameters
Phobics (n ¼ 26)
Controls (n ¼ 26)
t (df ¼ 50)
Number of omissions
Fixation duration (ms)
Saccade amplitude (degrees)
*Significant at the .05 level.
Duration from stimulus onset to the first fixation-on-a-spider
Phobics (n ¼ 26)
Controls (n ¼ 26)
t (df ¼ 50)
**Significant at the .01 level.
112 T. Pflugshaupt et al./Anxiety Disorders 19 (2005) 105–116
temporal order of fixations
Fig. 2. Time course of distance from fixation to nearest spider.
Fig. 3. An example of group-specific fixation behavior; based on averaged coordinates of the first 10
fixations; phobics (red) versus controls (blue); filled circles mark the first fixation; the superimposed
white arrow points at the target location.
T. Pflugshaupt et al./Anxiety Disorders 19 (2005) 105–116 113
initial phase (i.e., from second to fifth fixation) and subsequently further from
spiders. The sixth fixation, which occurred after 1728 ms on average (phobics:
mean ¼ 1707 ms, S:D: ¼ 311 ms; controls: mean ¼ 1749 ms, S:D: ¼ 367 ms),
denotes the turning point in phobics’ attentional bias.
An example of this group-specific fixation behavior is given in Fig. 3, which is
based on fixation coordinates averaged within groups. During this particular
search task, phobics fixated closer to the spider target than controls from the
second up to the fourth fixation and further from it thereafter.
The hypervigilance–avoidance hypothesis (Mogg et al., 1987) states that,
when confronted with threat stimuli, phobics initially direct their attention
towards, and subsequently away from threat. Previous empirical validation
attempts(Bradleyetal., 1999;Mogg etal., 1997)confirmed initial hypervigilance
and yet failed to show subsequent avoidance. The results of the present visual
hand and in contrast to previous findings, avoidance occurred in phobic behavior
and it followed hypervigilance as proposed by the hypervigilance–avoidance
hypothesis. On the other hand, our results can be used to explain why previous
studies did not reveal avoidance.
The main focus of analysis was on three predictions inferred from the
were expected to fixate fear-relevant stimuli earlier than controls. This is in line
with previous findings about fast detection abilities in phobics (Ohman et al.,
2000). Our results verified the prediction and thereby constitute further evidence
that these abilities indeed exist. The second prediction was that phobics fixated
closer to fear-relevant stimuli than controls during the initial phase of search
behavior, again due to hypervigilance. Confirming findings from studies that
applied dot probe tasks (Bradley et al., 1999; Mogg et al., 1997), the present
results verified this second prediction as well. Third and due to avoidance, it was
assumed that phobics fixate further from spiders than controls during subsequent
stages of information processing. Contrasting with previous findings, our results
confirmed such avoidance in phobic behavior. To the best of our knowledge, the
present study is therefore the first empirical validation of thevigilance–avoidance
hypothesis as a whole.
1999; Mogg et al., 1997), one is tempted to ask why such behavior was shown by
phobics in our study. First, eye movement analysis may be a more direct method
for investigating avoidance than dot probe tasks. Eye movement tracking allows
the assessment of both hypervigilance and avoidance by analyzing the temporal
and spatial distribution of fixations, whereas dot probe tasks require inferences
114T. Pflugshaupt et al./Anxiety Disorders 19 (2005) 105–116
frommanualreactiontimes.Asecond explanationisrelatedtothetime frame that
was under investigation. While previous studies applied time frames of up to
1500 ms, our results showed that avoidance became evident after 1700 ms of
stimulus exposure at the earliest. Controlled avoidance strategies might thus need
that much time to fully develop and replace initial hypervigilance.
hypothesis, a further focus of analysis was on search performance. In this regard,
the two groups did not differ significantly from each other, as indicated by the
number of omissions in the forced-choice task. Hence, the hypervigilance–
avoidance pattern observed in phobics’ fixation behavior had no effect on search
performance. This may, however, depend on the time frame of 5 s in our
experiment. Given the findings that, relative to controls, phobics fixated earlier
on a spider and closer to spiders up to the fifth fixation, it seems plausible that a
time frame of 2 s could have allowed phobics to surpass controls with regard to
search performance. Since the main focus of the present study was on the
hypervigilance–avoidance hypothesis rather than on search performance, we
decided to apply a relatively long time frame in order to include fast detection
as well as subsequent avoidance.
A third focus of analysis was on fixation duration and saccade amplitude.
These two basic oculomotor parameters were used to evaluate general group
differences in visual search behavior. Results showed that phobics made slightly
shorter fixations and significantly longer saccades than controls during the search
task. This can be interpreted in terms of an accelerated and more far-ranging
search behavior in phobics, possibly due to the arousing nature of the stimuli and/
or the task per se. Although search performance was not group-specific, these
results nevertheless indicate that the stimuli of our experiment were not exces-
sively frightening for phobics. Otherwise we might probably have observed a
near-to-refusal avoidance with enhanced fixation durations, smaller saccadic
amplitudes, and a poorer search performance in phobics.
In sum, the main objective of the present study was to empirically validate the
hypervigilance–avoidance hypothesis (Mogg et al., 1987) in spider phobics
during a visual task that required participants to search for spiders. Our results
clearly confirmed this hypothesis. In contrast to previous findings that exclusively
demonstrated initial hypervigilance (Bradley et al., 1999; Mogg et al., 1997), the
present study revealed evidence for subsequent avoidance as well, possibly due to
tasks, extended time frame). We thus consider our findings as the first empirical
validation of the hypervigilance–avoidance hypothesis as a whole. According to
Mogg et al. (1997), this behavioral pattern may be a key factor in maintaining
phobia. As a consequence of initial hypervigilance, phobics are more likely to
detect potentially threatening events and thus perceive the world as a dangerous
place, while subsequent cognitive avoidance prevents objective evaluation and
habituation to such events. In this respect, the present study contributed to the
empirical validation of a clinically relevant concept in phobia research.
T. Pflugshaupt et al./Anxiety Disorders 19 (2005) 105–116115
American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders,
(4th ed., rev.). Washington, DC: Author.
Bradley, B. P., Mogg, K., White, J., Groom, C., & de Bono, J. (1999). Attentional bias for emotional
faces in generalised anxiety disorder. British Journal for Clinical Psychology, 38, 267–278.
Hautzinger, M., Bailer, M., Wollall, H., & Keller, F. (1994). Beck-Depressions-Inventar (BDI):
Testhandbuch. Bern: Hans Huber.
Ishihara, S. (1999). Ishihara’s test for color deficiency. Tokyo: Kanehara & Co.
Johnsen, B. H., & Hugdahl, K. (1990). Fear questionnaires for simple phobias: psychometric
evaluations for a Norwegian sample. Scandinavian Journal of Psychology, 31, 42–48.
Kowler, E., Anderson, E., Dosher, B., & Blaser, E. (1995). The role of attention in the programming
of saccades. Vision Research, 35, 1897–1916.
Lang, P. J., Bradley, M. M., & Cuthbert, B. N. (1990). Emotion, attention and the startle reflex.
Psychological Review, 97, 377–398.
Mathews, A. (1990). Why worry? The cognitive function of anxiety. Behaviour Research and
Therapy, 28, 455–468.
Mayer, B., Merckelbach, H., & Muris, P. (2000). Self-reported automaticity and irrationality in spider
phobia. Psychological Reports, 87, 395–405.
Mogg, K., & Bradley, B. P. (1998). A cognitive-motivational analysis of anxiety. Behaviour Research
and Therapy, 36, 809–848.
Mogg, K., Bradley, B. P., de Bono, J., & Painter, M. (1997). Time course of attentional bias for threat
information in non-clinical anxiety. Behaviour Research and Therapy, 35, 297–303.
Mogg, K., Mathews, A., & Weinman, J. (1987). Memory bias in clinical anxiety. Journal of
Abnormal Psychology, 96, 94–98.
Muris, P., & Merckelbach, H. (1996). A comparison of two spider fear questionnaires. Journal of
Behavioural Therapy and Experimental Psychiatry, 27, 241–244.
Ohman, A., Flykt, A., & Lundqvist, D. (2000). Unconscious emotion: evolutionary perspectives,
psychophysiological data, and neuropsychological mechanisms. In: R. D. Lane, & L. Nadel
(Eds.), Cognitive neuroscience of emotion (pp. 296–327). New York: University Press.
Posner, M. I., Snyder, C. R. R., & Davidson, B. J. (1980). Attention and the detection of signals.
Journal of Experimental Psychology: General, 109, 160–174.
Szymanski, J., & O’Donohue, W. (1995). Fear of spiders questionnaire. Journal of Behavioural
Therapy and Experimental Psychiatry, 26, 31–34.
Thorpe, S. J., & Salkovskis, P. M. (1999). Animal phobias. In: D. C. L. Davey (Ed.), Phobias: a
handbook of theory, research and treatment (chap. 4, pp. 81–105). Chichester: Wiley.
Watts, F. N., & Sharrock, R. (1984). Questionnaire dimensions of spider phobia. Behaviour Research
and Therapy, 22, 575–580.
Williams, J. M. G., Watts, F. N., MacLeod, C., & Mathews, A. (1997). Cognitive psychology and
emotional disorders. Chichester: Wiley.
116 T. Pflugshaupt et al./Anxiety Disorders 19 (2005) 105–116