You are on page 1of 5

Journal of Abnormal Psychology Copyright 2005 by the American Psychological Association

2005, Vol. 114, No. 2, 314 –318 0021-843X/05/$12.00 DOI: 10.1037/0021-843X.114.2.314

Attentional Biases to Internal and External Sources of Potential Threat in


Social Anxiety
Suzanne L. Pineles and Susan Mineka
Northwestern University

D. M. Clark and A. Wells (1995) proposed that a shift of attention inward toward interoceptive
information is a central feature of social phobia. However, few studies have examined attentional biases
toward internal physiological cues in social phobia. The current experiment assessed whether socially
anxious individuals exhibit an attentional bias (a) toward cues for an internal source of potential threat
(heart-rate information), (b) toward cues for an external source of potential threat (threatening faces) or
(c) both. Ninety-one participants who were selected to form extreme groups based on a social anxiety
screening measure performed a dot-probe task to assess location of attention. Results showed that socially
anxious participants exhibited an attentional bias toward cues of internal, but not external, sources of
potential threat.

Keywords: attentional bias, threat, social anxiety, social phobia

Clark and Wells (1995) proposed that a shift of attention inward some studies have found an attentional bias toward threat faces in
toward interoceptive information is a central feature of social persons with high social anxiety, whereas others have not.
phobia. They hypothesized that socially anxious individuals shift There has been very little research on the role of internal
their attention inward and use their interoceptive information and somatic cues in social anxiety. Most such evidence derives indi-
feared behavioral symptoms to construct impressions of them- rectly from the literature on self-focused attention showing that
selves that they believe reflect how others see them. Further, such socially anxious individuals report more self-focused attention
individuals’ preoccupation with their somatic responses and neg- than do nonanxious individuals (e.g., Mellings & Alden, 2000).
ative social-evaluative thoughts interferes with their ability to However, recently, Mansell, Clark, and Ehlers (2003) used a new
process external social cues. To date, however, research supporting experimental paradigm to compare high versus low speech-
this theory is very sparse because most of it has either not distin- anxious individuals on their relative attention to a tactile cue for
guished between internal and external sources of threat (e.g., threat internal physiological changes versus to an external visual cue.
words can represent either internal or external threat, but no such They compared the balance of attention to an external visual probe
comparisons have been made) or has focused exclusively on on some trials (superimposed on pictures of happy, angry, or
external sources of threat (e.g., threatening faces). The results of neutral faces or household objects) versus to a tactile probe to the
studies assessing attentional bias for threatening (vs. neutral) faces fingers on other picture presentation trials (ostensibly signaling
in social phobia are very mixed (e.g., Mogg, Philippot, & Bradley, changes in heart rate and sweating). As compared with low speech-
2004, for vigilance results; R. Rapee & A. Öhman, personal anxious participants, high speech-anxious participants asked to
communication, October 2003, for null results; Chen, Ehlers, give a speech responded more quickly when the tactile probe
Clark, & Mansell, 2002, for avoidance results). It is unclear why signaling internal physiological changes was presented than when
the external probe was presented. This suggests that high speech-
anxious participants exhibited a relative attentional bias for inter-
Suzanne L. Pineles and Susan Mineka, Department of Psychology, nal cues relative to external cues. However, the only significant
Northwestern University.
effect was in relative bias score; high speech-anxious participants
This research was partially supported by a Dissertation Year Fellowship
did not respond faster than low speech-anxious participants to
and a Dissertation Year Grant by the University Research Grants Commit-
tee at Northwestern University granted to Suzanne L. Pineles. This article probes for potential internal (or external) threat. Furthermore, and
is based on research conducted by Suzanne L. Pineles in partial fulfillment unexpectedly, this bias was not specific to angry faces because the
of the requirements for her doctoral dissertation. high speech-anxious group exhibited a similar bias toward tactile
We thank Rick Zinbarg, Marcia Grabowecky, Galen Bodenhausen, and probes delivered with happy faces. Obviously, further research is
Iftah Yovel for their comments and contributions to this research. We are needed to better understand the role of attention to physiological
also grateful to Thomas Powers, Amanda Moates, Kim Baglien, Anita cues in social anxiety.
Kalhan, and Julie Landa for their help in collecting data for this experi- The current study was designed to assess whether an attentional
ment.
bias for cues of either external or internal sources of potential
Correspondence concerning this article should be addressed to Suzanne
L. Pineles who is now at the National Center for PTSD, Women’s Health threat could be demonstrated in high social anxiety. A dot-probe
Sciences Division (116B-3), Veteran’s Affairs Boston Healthcare System, paradigm with stimuli that represented both external and internal
150 South Huntington Avenue, Boston, MA 02130. E-mail: Suzanne sources of potential social threat was used. External sources of
.Pineles@med.va.gov threat were manipulated via exposure to threatening, happy, and
314
BRIEF REPORTS 315

neutral faces. Internal sources of threat were manipulated using segments of a sound wave (from a horse neighing and galloping). There
exposure to visual depictions of what was ostensibly the partici- were 32 slightly different, static, photo-like EKG images (half were re-
pant’s own heart rate and segments of a perceptually similar sound peated for each participant), and the sound wave was divided into 48
wave. Attention was assessed through comparing reaction times segments. The images were designed to be perceptually similar in both
shape and saturation of gray color of the wave.1 Participants have been
(RTs) for pairings of different combinations of the three types of
reported to believe (and have shown increased anxiety to false feedback
faces (for external sources of potential threat) and for pairings of
for) heart-rate information in both attentional studies of panic disorder and
the heart-rate information with sound-wave segments (for internal studies inducing anxiety with false heart-rate feedback (e.g., Ehlers, Mar-
sources of potential threat). The two between-groups variables graf, Roth, Taylor, & Birbaumer, 1988; Kroeze & van den Hout, 2000).
were the social anxiety group (high social anxiety vs. low social Modified dot-probe task. This task was programmed using the
anxiety) and speech condition (speech vs. no speech). PsyScope program (Version 1.1; Cohen, MacWhinney, Flatt, & Provost,
On the basis of Clark and Wells’s (1995) model, we hypothe- 1993). Participants used a chin rest and were seated with their eyes 114 cm
sized that high social anxiety (but not low social anxiety) partic- from the monitor and level with the center of the screen. Before each of the
ipants would exhibit an attentional bias toward heart rate (relative 192 trials, participants focused on a black central fixation cross for 500 ms.
to the sound wave) information. Second, because of mixed results The pairs of stimuli (face vs. face; heart rate vs. sound wave) were
presented vertically on the screen for 500 ms with the centers of the images
in the literature, we tentatively hypothesized that participants in the
separated by a distance of 11 cm, creating a visual angle of 5.5° between
high social anxiety group would be more likely than those in the
the centers of the pictures. Immediately following the offset, the probe (a
low social anxiety group to exhibit a bias toward threat faces small x) was presented in the former location of one of the stimuli, and
relative to both happy and neutral faces. Finally, because increas- participants indicated the location of the probe by pressing one of two
ing state anxiety is sometimes necessary to detect attentional response buttons on a button box. RTs measured the time interval between
biases in studies with analog participants (e.g., Mansell et al., the onset of the probe and the response. The position of the pairs of stimuli
2003; Williams, Watts, Macleod, & Mathews, 1997), we believed and of the probe were counterbalanced so that they appeared in either
that these attentional effects would be potentiated for high social location with equal frequency. There were two blocks of trials (10 practice
anxiety participants in the speech condition, relative to the no- and 96 experimental trials each) separated by a 30-s break; the intertrial
speech condition. interval between a response and a new trial was 500 ms. Within each block,
there were 24 of each of the four types of stimulus pairs. Each participant
saw the 192 trial pairs in a different random order.
Method
Participants Procedure
From an initial pool of 480 students who completed the Social Phobia Participants in the speech condition were given 2 min to plan a 5-min
Scale (SPS; Mattick & Clarke, 1998), 133 students who scored in the top speech, describing their experiences in college to be given at the end of the
(SPS ⱖ 26) or bottom (SPS ⱕ 6) 15th percentile on the SPS were chosen session, videotaped, and rated by experts. Participants in the no-speech
to participate in the study. All participants were randomly assigned to condition were simply asked to wait a few minutes until the experimenter
condition (speech vs. no speech) and were told that they were going to see was ready to begin. When the experimenter returned, participants were told
images of their own heart rate throughout the experiment. Nine participants that they were going to complete a computer task in which they would see
who did not believe the request to give a speech (if applicable) or heart-rate three types of images: faces, pictures of their heart rate, and segments of a
information to be real and 33 participants whose SPS scores were no longer visual depiction of a sound wave. In explaining the heart-rate images, the
in the same range at the time of the study were excluded from the analyses, experimenter told participants that their heart rates would be measured and
leaving a total of 91 participants in the final sample. The number of asked participants to put their index finger into a finger plethysmograph
participants in each group were as follows: for the high social anxiety/ that was connected to a heart-rate monitor in the adjoining room, which
speech condition, n ⫽ 24; for the high social anxiety/no-speech condition, was connected to the computer used in the task. The experimenter briefly
n ⫽ 25; for the low social anxiety/speech condition, n ⫽ 19; and for the left the room to allegedly check to make sure the heart-rate monitor was
low social anxiety/no-speech condition, n ⫽ 23. The SPS scores at the time recording. Participants were then shown what was ostensibly their heart
of the experiment in our high social anxiety group (M ⫽ 35.32, SD ⫽ 7.87) rate at the current moment on the computer screen and were told that they
are quite close to those reported by the diagnosed socially phobic individ- would see snapshots depicting their current heart rates throughout the
uals (M ⫽ 40.00, SD ⫽ 16.00) from the SPS validation study (Mattick & experiment. To explain the sound-wave images, the experimenter ensured
Clarke, 1998). The mean score in the low social anxiety group was 3.81 that participants heard a sound clip (of a horse neighing and galloping) and
(SD ⫽ 2.38). told them that they would be seeing segments of this sound in visual format
throughout the experiment; participants were then shown the first sound-
wave image. Participants were asked to identify differences between the
Materials
sound-wave image and the heart-rate image to ensure that they understood
Stimuli. Two types of stimulus pairs were used in this study: (a) pairs and were paying attention to the perceptual features of the stimuli. Partic-
of faces and (b) pairs of visual depictions of heart rates versus sound ipants were then told that their task was to identify the location (top or
waves. The three possible face stimulus pairs consisted of two photographs bottom) of an “x” on the computer screen and that a pair of the images
of the same person exhibiting different facial expressions: (a) threat versus previously described would precede the presentation of the “x.” Partici-
neutral (48 pairs), (b) happy versus neutral (48 pairs), and (c) threat versus
happy (48 pairs). The faces were selected from two sets that have been used
1
in previous studies of facial emotion (e.g., Mogg et al., 2004; Lundquist, In two pilot experiments, unselected participants attended to the sound-
Flykt, & Öhman, 1998). wave and heart-rate information equally both when participants were given
For measuring attention to cues of internal sources of potential threat, we information as to what the pictures represented (n ⫽ 11) and when only
randomly presented 48 trials, with each trial pairing pictures ostensibly shown the physical stimuli (without information as to what the stimuli
representing static segments of the participant’s heart rate with different represented; n ⫽ 8).
316 BRIEF REPORTS

pants then completed the dot-probe task with the experimenter present in a 2 ⫻ 2 ⫻ 4 ⫻ 2 ⫻ 2 ANOVA with two between-groups variables
the room. After the task, they completed a series of self-report measures of anxiety group and speech condition and three within-participant
including the SPS (Mattick & Clarke, 1998) and the State–Trait Anxiety variables of stimulus type (heart rate vs. sound wave, threat vs.
Inventory—State version (STAI–S; Spielberger, Gorusch, & Lushene, neutral faces, happy vs. neutral faces, and threat vs. happy faces),
1970). Finally, as part of the debriefing, participants were told that they
stimulus location (up vs. down), and probe location (up vs. down).
would not have to give a speech and were asked whether they had believed
that the request for a speech (only for relevant participants) and the The main effect of stimulus type, F(3, 85) ⫽ 4.74, ES ⫽ .14, p ⬍
heart-rate information were real. .05, was significant, with participants responding more quickly on
the heart-rate trials than on all face trials ( ps ⬍ .05; see Table 1).
Results This main effect was modified by a two-way interaction of Stim-
ulus Type ⫻ Stimulus Location, F(3, 85) ⫽ 3.09, ES ⫽ .10, p ⬍
State Anxiety .05; a three-way interaction of Stimulus Type ⫻ Stimulus Loca-
tion ⫻ Probe Location, F(3, 85) ⫽ 7.97, ES ⫽ .22, p ⬍ .05; and
Although firm conclusions cannot be drawn due to the lack of
a four-way interaction of Stimulus Type ⫻ Stimulus Location ⫻
pre- and postmanipulation measures, we wanted to assess whether
Probe Location ⫻ Anxiety Group, F(3, 85) ⫽ 3.38, ES ⫽ .11, p ⬍
the speech manipulation affected state anxiety levels. As expected,
.05. However, neither the main effect of speech condition nor the
a 2 (Anxiety Group) ⫻ 2 (Speech Condition) analysis of variance
interaction of Anxiety Group ⫻ Speech Condition was significant
(ANOVA) on the STAI–S revealed a main effect of anxiety group:
(Fs ⬍ 1, ES ⬍ .01). To test our a priori hypotheses about perfor-
for the high social anxiety group (M ⫽ 54.12, SD ⫽ 9.94); for the
mance within stimulus sets, we conducted separate ANOVAs for
low social anxiety group (M ⫽ 32.56, SD ⫽ 6.93), F(1, 87) ⫽
attention to heart rate versus sound waves and for attention to
142.45, effect size (ES) ⫽ .62, p ⬍ .05. However, neither the main
different combinations of faces.
effect of condition nor the interaction of Anxiety Group ⫻ Speech
Attention to heart rate. A 2 (anxiety group) ⫻ 2 (speech
Condition was significant, Fs ⬍ 3.35, ES ⬍ .04, ns. Thus, the
speech manipulation either did not affect state anxiety levels for condition) ⫻ 2 (stimulus location) ⫻ 2 (probe location) mixed-
socially anxious participants or the STAI–S may not have been design ANOVA revealed a main effect of stimulus location, F(1,
sensitive enough to detect potential group differences in anxiety 87) ⫽ 7.98, ES ⫽ .09, p ⬍ .01; a two-way interaction of Stimulus
aroused by the speech manipulation. Alternatively, to assess the Location ⫻ Probe Location, F(1, 87) ⫽ 18.84, ES ⫽ .18, p ⬍ .001;
possibility that we had inadequate power to detect a 2 ⫻ 2 and the predicted three-way interaction of Anxiety Group ⫻ Stim-
interaction— given our sample size, a medium effect size, and an ulus Location ⫻ Probe Location, F(1, 87) ⫽ 8.98, ES ⫽ .09, p ⬍
␣ ⫽ .05—we found the power was .65. Thus, we may have had .01. Again, neither the main effect of speech condition nor the
insufficient power to detect a medium-sized effect of speech. Anxiety Group ⫻ Speech Condition was significant (Fs ⬍1, ES ⬍
.02). To break down these interactions, we computed a bias score
for each participant that compares RTs when the probe is in the
Attentional Task
location of the heart rate versus RTs when the probe is in the
The mean percentages of data lost due to errors and outliers location of the sound wave (cf. MacLeod & Mathews, 1988).
were 0.7% and 1.4%, respectively. The RT data were subjected to Positive values reflect an attentional bias toward heart rate, and

Table 1
Mean Response Latencies (and Standard Deviations) to Probes (in ms)

Location Low SA High SA

No No
Stimulus pair Stimulus Probe Speech speech Speech speech

Threat/neutral faces Tu/Nd Up 432 (58) 433 (115) 430 (78) 412 (76)
Tu/Nd Down 440 (77) 433 (98) 441 (66) 416 (62)
Nu/Td Up 430 (68) 437 (110) 435 (88) 415 (77)
Nu/Td Down 448 (76) 428 (89) 435 (64) 418 (61)
Happy/neutral faces Fu/Nd Up 433 (69) 434 (102) 438 (79) 410 (73)
Fu/Nd Down 437 (66) 438 (96) 435 (67) 412 (62)
Nu/Fd Up 432 (64) 437 (109) 429 (81) 412 (79)
Nu/Fd Down 436 (81) 436 (110) 441 (55) 415 (55)
Threat/happy faces Tu/Fd Up 434 (70) 438 (109) 438 (80) 415 (73)
Tu/Fd Down 437 (77) 436 (114) 445 (69) 416 (67)
Fu/Td Up 432 (70) 433 (98) 445 (65) 422 (66)
Fu/Td Down 441 (74) 446 (105) 431 (78) 421 (72)
Heart rate/sound wave Hu/Sd Up 430 (68) 431 (103) 426 (80) 399 (69)
Hu/Sd Down 431 (60) 444 (118) 443 (79) 418 (61)
Su/Hd Up 418 (62) 428 (105) 441 (96) 415 (76)
Su/Hd Down 419 (71) 429 (103) 435 (77) 399 (58)

Note. Standard deviations are in parentheses. In the stimuli location column, T ⫽ threat faces; N ⫽ neutral
faces; F ⫽ happy faces; H ⫽ heart-rate information; S ⫽ sound-wave information; u ⫽ up; d ⫽ down; SA ⫽
social anxiety.
BRIEF REPORTS 317

Figure 1. Bias scores for the four pairs of stimuli by group. Positive values represent bias toward the first
stimulus of the pair, and negative values represent bias toward the second stimulus. SA ⫽ social anxiety.

negative values reflect an attentional bias away from heart rate. If socially anxious individuals shift their attention to monitor these
The two groups were compared on the heart-rate bias score. High internal cues, they may also assume that others can both see these
social anxiety participants (M ⫽ 17.47, SD ⫽ 21.26) showed signs of anxiety and judge them harshly, further reinforcing their
greater attentional bias scores toward heart rate than did low social perceived dangerousness of the social situation.
anxiety participants (M ⫽ 4.23, SD ⫽ 23.87), t(89) ⫽ 2.80, p ⬍ Some other evidence also suggests that socially anxious indi-
.01. In addition, separate t tests compared each group’s bias score viduals exhibit an internal shift of attention. Indirect support for
with zero (0 ⫽ no bias) and found that the high social anxiety
group’s bias score differed from zero, t(48) ⫽ 5.75, p ⬍ .01,
whereas the low social anxiety group’s bias score did not signif- 2
Despite the fact that the mixed-design ANOVA did not yield signifi-
icantly differ from zero, t(41) ⫽ 1.14, ns (see Figure 1).
cant main effects or interactions, we conducted a few specific planned
Attention to faces. A 2 (anxiety group) ⫻ 2 (speech condi- comparisons to determine whether there was a trend for an attentional bias
tion) ⫻ 3 (type of face pair) ⫻ 2 (stimulus location) ⫻ 2 (probe to threat faces given that some prior studies have found this effect. We
location) mixed-design ANOVA revealed no significant main ef- conducted two planned comparisons using only participants who were in
fects or interactions (all ps ⬎ .10).2 the speech condition because we believed that the speech manipulation
might have had a slight impact on speech or social anxiety that the
Discussion STAI–State was not sensitive enough to detect. Bias scores were computed
for the two types of face pairs that included threat faces using the same
The present study investigated whether socially anxious indi- formula that was used to form the bias score for the heart-rate information.
viduals preferentially attended to signals representing internal Thus, one bias score was for attention toward threat versus neutral faces
physiological cues as might be expected according to Clark and and one was for attention toward threat versus happy faces. T tests
Wells’s (1995) cognitive model of social phobia. The results comparing the high and low social anxiety groups in the speech condition
indicated that high (but not low) socially anxious individuals revealed no group differences on the threat/happy bias scores, t(41) ⫽ .05,
ns, but there was a trend for a group effect on the threat/neutral bias scores,
clearly showed preferential attention to information about their
t(41) ⫽ 1.86, p ⫽ .07. The high social anxiety group (M ⫽ 5.82, SD ⫽
heart rate (relative to a sound wave), thus supporting Clark and 18.2) attended slightly more to the threat faces when paired with the neutral
Wells’s hypothesis that socially phobic individuals turn their at- faces as compared with the low social anxiety group, which, if anything,
tention inward to monitor internal physiological cues. This shift of showed a bias away from threat faces (M ⫽ ⫺4.42, SD ⫽ 17.62). However,
attention to cues for possible internal sources of threat is thought these results should be interpreted cautiously because any effects of the
to be central to the “vicious cycle” of anxiety in social situations. speech manipulation on anxiety were not apparent using the STAI–S.
318 BRIEF REPORTS

this bias has been found in studies assessing reported self- focused internal sources of threat in research on social anxiety parallels an
attention (e.g., Mellings & Alden, 2000). Somewhat more direct earlier development that occurred 20 –25 years ago with agorapho-
evidence is derived from Mansell et al.’s (2003) study in which bia and panic disorder when focus on the fear of external situations
socially anxious individuals asked to give a speech responded in agoraphobia began to change toward an additional focus on the
more quickly to a tactile probe ostensibly signaling internal phys- importance of fear of internal bodily sensations in panic disorder.
iological changes relative to an external visual probe. The current Although very few studies using tightly controlled information-
study adds to this body of literature by using a different paradigm processing methodology have yet assessed attentional biases to-
that looked at socially anxious participants’ attentional bias to a ward cues of internal threat in social anxiety, we hope this study
cue ostensibly representing heart-rate changes versus a visually provides a useful start for future work in this area.
similar sound-wave cue. Thus, socially anxious individuals have
now been shown not only to report an internal shift of attention but References
also to exhibit an attentional bias toward cues ostensibly signaling Bradley, B. P., Mogg, K., Millar, N., Bonham-Carter, C., Fergusson, E.
internal physiological changes. Jenkins, J., & Parr, M. (1997). Attentional biases for emotional faces.
Not unlike some other studies using faces as stimuli with dot- Cognition and Emotion, 11, 25– 42.
probe paradigms that assessed attentional biases in high socially Chen, Y. P., Ehlers, A., Clark, D. M., & Mansell, W. (2002). Patients with
anxious individuals (e.g., Bradley et al., 1997; R. Rapee and A. generalized social phobia direct their attention away from faces. Behav-
Öhman, personal communication, October 2003), we failed to find iour Research & Therapy, 40, 677– 687.
significant group differences in location of attention to threat Clark, D. M., & Wells, A. (1995). A cognitive model of social phobia. In
versus neutral faces (although there was a trend for this effect in R. G. Heimberg, M. R. Liebowitz, D. A. Hope, & F. R. Schneier (Eds.),
Social phobia: Diagnosis, assessment, and treatment (pp. 69 –93). New
the speech condition; see Footnote 2). The exact parameters gov-
York: Guilford Press.
erning the occurrence of attentional bias toward (or away from) Cohen, J. D., MacWhinney, B., Flatt, M., & Provost, J. (1993). PsyScope:
threatening faces obviously awaits further research. An interactive graphic system for designing and controlling experiments
There are several limitations of this study. One is that the speech in the psychology laboratory using Macintosh computers. Behavior
manipulation was not successful in inducing anxiety, at least as Research Methods, Instruments, & Computers, 25, 257–271.
measured by the STAI, possibly because it was not sensitive Ehlers, A., Margraf, J., Roth, W. T., Taylor, C. B., & Birbaumer, N. (1988).
enough to detect any speech or social anxiety that participants may Anxiety induced by false heart rate feedback in patients with panic
have experienced. Alternatively, the high social anxiety group may disorder. Behaviour Research & Therapy, 26, 1–11.
not have been very afraid of giving speeches given that they were Kroeze, S., & van den Hout, M. A. (2000). Selective attention for cardiac
information in panic patients. Behaviour Research & Therapy, 38,
selected on a general measure of social (rather than speech) anx-
63–72.
iety (see Mansell et al., 2003). Finally, all the high social anxiety Lundquist, D., Flykt, A., & Öhman, A. (1998). Karolinska directed emo-
participants might have been socially stressed because an experi- tional faces [CD]. Stockholm, Sweden: Section of Psychology, Karolin-
menter was in the room at the time of the experiment to ensure that ska Hospital.
the participants focused on the task. A second limitation of this MacLeod, C., & Mathews, A. (1988). Anxiety and the allocation of
study is that it was conducted with an analog sample, and so the attention to threat. Quarterly Journal of Experimental Psychology: Hu-
results await replication with a clinical sample. However, the man Experimental Psychology, 38, 659 – 670.
average levels of social anxiety in the high social anxiety group Mansell, W., Clark, D. M., & Ehlers, A. (2003). Internal versus external
(M ⫽ 35.32, SD ⫽ 7.87) were fairly close to those reported by attention in social anxiety: An investigation using a novel paradigm.
Behaviour Research & Therapy, 41, 555–572.
Mattick and Clarke’s (1998) diagnosed socially phobic individuals
Mattick, R. P., & Clarke, J. C. (1998). Development and validation of
(M ⫽ 40.00, SD ⫽ 16.00). Further, because the groups were measures of social phobia scrutiny fear and social interaction anxiety.
selected on a social anxiety measure, we do not know whether the Behaviour Research & Therapy, 36, 455– 470.
bias toward heart-rate information is specific to social anxiety. Mellings, T. B., & Alden, L. E. (2000). Cognitive processes in social
The final limitations of the study involve the stimuli involved. It anxiety: The effects of self-focus, rumination and anticipatory process-
was not feasible to measure an attentional bias toward actual ing. Behaviour Research & Therapy, 38, 243–257.
interoceptive information, and so internal cues were represented by Mogg, K., Philippot, P., & Bradley, B. P. (2004). Selective attention to
pictures of what participants believed to be their own heart rate. angry faces in clinical social phobia. Journal of Abnormal Psychology,
However, two studies have now found evidence for an attentional 113, 160 –165.
Spielberger, C. D., Gorusch, R. L., & Lushene, R. E. (1970). Manual for
bias for interoceptive information signaled through either visual or
the State-Trait Anxiety Inventory. Palo Alto, CA: Consulting Psycholo-
tactile modalities. Moreover, this seems to make this a particularly
gists Press.
conservative test of Clark and Wells’s (1995) model because the Williams, J. M., Watts, F. N., MacLeod, C., & Mathews, A. (1997).
attentional bias detected in socially anxious individuals was to Cognitive psychology and emotional disorders (2nd ed.). Chichester,
information about internal events and not to internal signals per se. England: Wiley.
Overall, findings of preferential attention toward cues for inter-
nal sources of threat lend support to Clark and Wells’s (1995) Received January 21, 2004
contention that socially anxious individuals may shift attention Revision received July 16, 2004
inward to monitor their physiological cues. Addressing the role of Accepted July 19, 2004 䡲

You might also like