In line with the hypotheses, participants who were led to believe that their arousal had increased during a conversation with a stranger felt more anxious, felt that they were coming across badly, and reported more bodily sensations than participants who did not receive feedback on their arousal levels or those who were led to believe that their arousal had decreased. This finding is in line with cognitive models that state that attention to internal cues of perceived arousal in social situations influences anxiety, and actual and perceived performance.
However, contrary to expectation, we did not find that this effect was more pronounced in participants in high compared to low social anxiety. False feedback of increased arousal affected participants with low social anxiety to the same extent as those with high anxiety, although their overall level of anxiety was lower and they rated their social performance during the conversation more favourably. The finding of similar increments in anxiety in high and low anxiety groups is consistent with Mansell and Clark (1999)
who found that both groups experienced the same increments and level of anxiety following a social threat induction. It is possible that information regarding increased arousal acts in the same way as perceived social threat possibly drawing attention to the importance of performance and immediate attention on the self. However, if this were solely the case, then one would expect no differences in anxiety and performance between high and low socially anxious individuals in the false feedback study of Papageourgiou and Wells (2002)
. They found that only high anxious individuals receiving information about increased heart rate reported significantly greater anxiety and negative performance. The effect was not seen in low anxious individuals. However, they provided information about heart rate only, rather than physiological information which could correlate with a wide range of anxiety symptoms and social performance concerns, such as blushing, shaking, and sweating. Further, false feedback was provided prior to participants’ conversations. It was therefore not possible to determine the online impact of such information. Low anxious individuals may not attend to body state information throughout a conversation when this information is provided beforehand. However, when it is provided online, they may be more likely to shift their attention and attend to it with a subsequent impact on anxiety and performance.
The most likely reason for the pattern of results then is that the experimental instructions focused the attention of all participants’ on their “arousal”, regardless of social anxiety status. There was no difference between the high and low social anxiety groups in the number of vibrations they detected or in the extent to which they believed the feedback reflected their bodily state. Thus, low socially anxious participants in the Increase condition were experimentally made to behave like patients with social phobia, in that they were made to shift their attention away from the conversation to the monitoring of their increasing arousal. This would explain why they showed greater anxiety and thought they came across less favourably than low socially anxious people in the Decrease and Control conditions. Other research suggests that when people are able to allocate their attention freely, high socially anxious people are more likely than those with low social anxiety to shift their attention away from facial expressions to internal cues in conditions of social-evaluative threat (Mansell, Clark, & Ehlers, 2003
), and would therefore be expected to be more likely to experience the adverse effects of monitoring their increasing arousal.
This fits with the Clark and Wells (1995)
cognitive model of social phobia in which self-focused attention is a key maintaining factor. Self-focused attention facilitates self-monitoring and hence, attention to physical cues of arousal while at the same time preventing individuals with the disorder from processing how they are actually coming across. Wells and Papageorgiou (2001)
found that information about increased arousal was associated with increased self-focused attention in patients with social phobia. They suggest that attention may be the mechanism by which perception of arousal affects anxiety. Self-focused attention has been associated with anxious appearance (e.g., Woody, 1996
) and poorer performance (e.g., Daly, Vangelisti, & Lawrence, 1989
In this study, high socially anxious participants overrated how negatively they came across to a greater extent than low socially anxious participants, compared to both the confederate and the independent assessor. This is in line with Mellings and Alden (2000)
who found that socially phobic and non-anxious individuals overestimated their anxiety-related behaviour relative to observers, but socially phobic individuals did so to a greater extent. Furthermore, as expected, the degree of overestimation varied with the experimental condition. Participants who thought that their arousal had increased showed greater overestimation than participants in the Control condition.
Did participants actually use their perceived physiological state to infer how they came across to others? The effects of the experimental manipulation on participants’ ratings of their social performance, and the pattern of correlations support this relationship. In the high social anxiety group, perceived bodily sensations during the conversation correlated highly with the perceived success of the conversation and more negative ratings of social performance. In addition, perceived bodily sensations correlated with the degree of underestimation of performance and overestimation of anxiety visibility, in line with the findings of Mansell and Clark (1999)
and McEwan and Devins (1983)
The experimental manipulation not only affected the participants’ impression of how they came across, but also the actual impression they made on the confederate and an independent assessor. In line with previous studies of social phobia, the high social anxiety group made a less favourable impression overall (Alden & Wallace, 1995
; Jones & Carpenter, 1986
; Norton & Hope, 2001
; Stopa & Clark, 1993
). As expected and consistent with Wells and Papageorgiou (2001)
, participants in the Increase condition made a less favourable impression than participants in the Decrease condition. Further, participants in the Decrease condition were also rated more positively than participants in the Control condition. However, there was no significant difference between the Increase and Control conditions, in contrast to the self-report measures.
The most likely explanation for the overall pattern of findings is as follows: Believing one's arousal is increasing during a social interaction has a large impact on the individual's perception of how well the interaction is going and how well they come across, as these are influenced by anxiety level and perceived bodily sensations. However, this internal state is not necessarily obvious to the social interaction partner and other observers, as indicated by the absence of condition effects on the item “How anxious did the person look?”, and on negative items of the Behaviours Checklist for the independent assessor. Other people appear more likely to pick up differences in positive behaviours, and accordingly, the effects of the experimental manipulation on the assessor ratings were restricted to positive behaviours. The positive behaviours rated by the observers favoured the Decrease group, which had been reassured by the feedback that their arousal was decreasing and may thus have come across in a more relaxed and accessible way. Thus, the pattern of results suggests, in line with models of social phobia (Clark & Wells, 1995
; Rapee & Heimberg, 1997
), that relying on internal anxiety cues to infer how one comes across does not give a good estimate of what others perceive during a social interaction.
The present results support several procedures used in cognitive therapy (Clark et al., 2003
) to treat social phobia. Discussion and behavioural experiments (including videofeedback) are used to help patients to realise that focusing their attention on their anxiety symptoms gives a misleading impression of how they come across and makes them feel more anxious. They are also taught to shift their attention from monitoring their internal state to the interaction with others, which seems to be helpful in both correcting their impression that they are performing badly and in creating a more favourable impression on others.
The present study had some limitations. First, the sample size was relatively small and the random allocation led to a difference in trait anxiety and trait APQ scores between high social anxiety participants allocated to the Increase versus Decrease and Control conditions. However, the pattern of this difference does not explain the pattern of results observed for the dependent measures (main effects of condition). Furthermore, ANCOVAs established that the differences between the experimental conditions remained significant when STAI-T and APQ scores were controlled statistically. Second, the experiment depended on false feedback and thus relied on the extent to which participants believed the manipulation. A proportion of participants who did not believe the manipulation had to be excluded. The majority of them were nursing students who may have had more knowledge of physiology than students of other disciplines. Although we did not observe systematic differences between people who did and did not believe the feedback, it is conceivable that the results do not generalise to all people. Third, the participants were student volunteers selected on the basis of their scores on a measure of social anxiety, and it remains to be tested whether they generalise to patients with social phobia. However, the results of previous studies using other paradigms suggest that such a generalisation is likely (Chen, Ehlers, Clark, & Mansell, 2002
; Mansell, Clark, Ehlers, & Chen, 1999
; Wells & Papageorgiou, 2001, 2002
). Finally, we assessed bodily sensations during the conversation by self-report and did not measure actual changes in physiology. Thus, we are unable to determine whether the instructions led to any group differences in the participants’ arousal. It is unlikely given that Papageourgiou and Wells (2002)
found no differences in actual physiological arousal between high and low socially anxious individuals when they were given false feedback of increased arousal before a conversation with a confederate. Other studies have also found no difference in actual heart-rate during a social task between socially anxious and non-socially anxious individuals (e.g., Grossman, Wilhelm, Kawachi, & Sparrow, 2001
; Puigcerver, Martinez-Selva, Garcia-Sanchez, & Gomez-Amor, 1989
This study looked at online false feedback of arousal and found that participants who were given false feedback of increased arousal during a conversation reported greater anxiety, poorer perceived performance, more physical cues of anxiety, and greater underestimation of their performance and overestimation of the visibility of their anxiety. Although the effects of false feedback were similar across high and low social anxiety groups, the feedback may have further influenced those with high social anxiety in that their baseline anxiety is greater. Inducing attention to increased internal sensations demonstrates how socially anxious individuals may be at a further disadvantage in social situations. Shifting attention away from the negative meaning of internal cues of arousal (as in the Decrease condition) improved performance and anxiety. This supports techniques used in cognitive therapy of social phobia and underscores the need to help sufferers shift attention away from internal cues of arousal.