ترس مربوط به ارتباط های انتخابی و اضطراب اجتماعی: فقدان یک بایاس مثبت

An illusory correlation paradigm was used to compare high and low socially anxious individuals’ initial, on-line and a posteriori covariation estimates between emotional faces and aversive, pleasant and neutral outcomes. Overall, participants demonstrated an initial expectancy bias for aversive outcomes following angry faces, and pleasant outcomes following happy faces. On-line expectancy biases indicated that initial biases were extinguished during the task, with the exception of low socially anxious individuals who continued to over-associate positive social cues with pleasant outcomes. In addition to lacking this protective positive on-line bias, the high social anxiety group reported retrospectively more negative social cues than the low socially anxious group. Findings are discussed in relation to similar evidence from recent interpretive and memory paradigms.

According to recent cognitive theories, biases in information processing play an important role in the etiology and maintenance of emotional disorders (e.g. Beck, 1976; Eysenck, 1997; Williams, Watts, MacLeod, & Mathews, 1997). Specific cognitive models of social anxiety propose that socially anxious individuals make anticipatory and retrospective judgments about social situations that appear consistent with a negative interpretative bias, which serves to maintain and increase social anxiety (Wells & Clark, 1997). Consistent with these predictions, recent evidence suggests that individuals with social phobia exhibit interpretational biases that favor socially threatening information (Amir, Foa, & Coles (1998a) and Amir, Foa, & Coles (1998b); Stopa & Clark, 2000; Wallace & Alden, 1997; see review by Heinrichs & Hofmann, 2001). However, there has been considerably less research into biases in judgments, which may underlie socially anxious individuals’ negative appraisal and association of social stimuli with threat. The illusory correlation/covariation bias paradigm was developed by Tomarken, Mineka, and Cook (1989) to assess individual differences in the association of fear-relevant stimuli with aversive outcomes. In a series of studies, Tomarken et al. (1989) showed high and low spider-fearful participants slides of fear-relevant stimuli (i.e. spiders) and fear-irrelevant stimuli (i.e. flowers and mushrooms) that were followed by either an aversive outcome (electric shock), a tone, or nothing at all (neutral outcomes). Across 72 slide-outcome trials, the three categories of slide were randomly related to the type of outcome that followed the slide. At the end of the sequence, participants were asked to judge the degree of covariation between the various slide categories and outcomes. Results indicated that spider-fearful participants substantially overestimated the association between slides of spiders and the aversive shock outcome relative to both the estimations of low fear participants and the veridical contingency (p=.33p=.33; Tomarken et al., 1989). Evidence supporting Tomarken et al.'s (1989) findings has been reported in studies using similar paradigms (see also Tomarken, Sutton, & Mineka, 1995). For example, Amin and Lovibond (1997) assessed initial and a posteriori (post-task) covariation estimates between biological phylogenetic (e.g. snake/spiders), technological ontogenetic (e.g. gun/knife) fear-relevant stimuli and electric shocks. Results indicate that at the outset of the experiment, all participants, regardless of prior fear, had elevated expectancies for shock following both phylogenetic and ontogenetic fear stimuli. However, at the end of the experiment, covariation biases were only present between phylogenetic stimuli and shocks in the high fear group (see also Kennedy, Rapee, & Mazurski, 1997). It has also been proposed that selective associations may play an important role in the maintenance of phobic disorders. For example, de Jong, Merckelbach, and Arntz (1995) reported significant correlations between residual covariation bias after treatment and return of phobic fear two years later (see also de Jong, Merckelbach, Arntz, & Nijman, 1992). These findings are consistent with the view that a tendency to associate threat outcomes with stimuli central to the concerns of anxious individuals is likely to confirm and maintain perceptions of danger. However, fear-relevant selective associations found in specific animal phobia and also in panic disorder (e.g. Pauli, Montoya, & Markz, 1996) have not been observed in all anxiety disorders. Pury and Mineka (1997) found elevated selective covariation estimates between mutilation slides and aversive shocks, irrespective of participants’ prior blood-injury fear. Sutton, Luten, Pury, and Mineka (1990, cited in Öhman & Mineka, 2001) found evidence of covariation bias for fear-relevant social stimuli (such as slides of angry or disgusted facial expressions) and aversive outcomes in four separate experiments, but no evidence of covariation bias for fear-irrelevant stimuli (i.e. happy and neutral faces). Contrary to findings in specific animal phobias, the magnitude of covariation bias was not affected by participants’ levels of anxiety. In a similar study, de Jong, Merckelbach, Bogels, and Kindt (1998) assessed high and low socially anxious individuals’ initial and a posteriori covariation estimates for angry, happy and neutral faces, and aversive shock outcomes. In addition, participants’ outcome expectancies were also monitored on a trial by trial basis. Results indicated that, irrespective of prior social anxiety levels, participants demonstrated a priori and a posteriori covariation bias between angry faces and shocks. These findings, together with the pattern of participants’ on-line outcome expectancies suggest that a posteriori covariation biases arose from initial expectancies that survived extinction (see also de Jong et al., 1995). The failure to find enhanced negative covariation biases in social anxiety (Sutton et al., 1990; de Jong et al., 1998) contrasts with studies showing that socially anxious individuals tend to make negative judgments regarding ambiguous social stimuli (Amir, Foa, & Coles (1998a) and Amir, Foa, & Coles (1998b); Stopa & Clark, 2000; Wallace & Alden, 1997). For example, Amir et al. (1998a) found that individuals with generalized social phobia were more likely to judge social scenarios as negative (as opposed to positive or neutral) compared to individuals with obsessive-compulsive disorder and low anxious controls (see also Stopa & Clark, 2000). The latter studies suggest that individuals with social anxiety demonstrate negative judgmental biases when required to evaluate previously described social interactions off-line. However, it remains unclear whether socially anxious individuals jump to negative conclusions as an event unfolds (on-line), or whether they primarily make distorted judgments after the interaction (Clark & McManus, 2002). In a recent assessment of on-line processing biases, Hirsch & Mathews (1997) and Hirsch & Mathews (2000) used lexical decision tasks to measure participants’ speed to assess the grammatical compatibility of (threatening or non-threatening) target words with the endings of socially relevant sentences. Results suggested that non-patient control participants demonstrated a positive on-line inferential bias, whilst patients with social phobia failed to demonstrate either positive or negative on-line inferences. Thus, high socially anxious individuals appeared to lack the on-line positive bias characteristic of low anxious controls. Moreover, Hirsch and Mathews (2000) propose that this positive inferential bias may serve an important protective role in maintaining high levels of self-esteem and low levels of social anxiety. To summarize, evidence of judgmental biases in social anxiety is mixed. Few studies of covariation bias have been carried out to date (de Jong et al., 1998; unpublished studies by Sutton et al., 1990), and these have failed so far to find differences between high and low socially anxious individuals. On the other hand, some studies, which have examined interpretative biases, suggest that socially anxious individuals have a reduced positive bias when evaluating ambiguous social information on-line, whereas others suggest they have an enhanced negative bias in retrospective (off-line) judgments regarding hypothetical social scenarios. A notable limitation of previous covariation bias studies is that they only used negative and neutral outcomes (e.g. cutaneous shocks, lights and tones). Thus, it is unclear whether the lack of positive inferential bias in social anxiety (found so far only during on-line resolution of ambiguous information) also extends to those processes involved in learning associations between social cues and outcomes. Thus, further research is required, using both positive and negative cues and outcomes, in order to investigate whether social anxiety is associated with impaired processing of positive information. Moreover, if social anxiety is found to be primarily associated with an impaired positive bias in on-line judgmental processes, this would have implications for cognitive models of anxiety which have tended to emphasise the selective processing of socially threatening information, rather than deficits in processing positive material (e.g. Clark & Wells, 1995; Rapee & Heimberg, 1997). Thus, it is not only theoretically important to clarify whether, in social anxiety, biases in specific cognitive operations are restricted to certain types of emotional stimuli (e.g. positive, rather than negative, information), but also clinically relevant, as a reduced tendency to associate social cues, particularly those signaling positive appraisal (e.g. smiling faces) with positive outcomes, would hinder correction of socially anxious individuals’ pre-existing fears of negative evaluation. The main aim of the present study was to investigate further both on-line and off-line judgmental biases in high and low socially anxious individuals using a modified version of the covariation detection paradigm used by de Jong et al. (1998). As discussed earlier, it is possible that social anxiety might be characterized not only by an enhanced negative bias in retrospective evaluation of social information, but more specifically, by a reduced positive bias in associating pleasant outcomes with social stimuli. Hence, the present covariation bias task was modified to assess both positive and negative outcome expectancies. Given that other peoples’ faces are considered to be relevant to the concerns of individuals with social anxiety (Trower & Gilbert, 1989), the present study presented high and low socially anxious individuals with a series of angry, happy and neutral faces. Following each face, participants were presented with one of four outcomes which were either aversive, pleasant or neutral emotional images, selected from the International Affective Picture Set (IAPS: Lang, Bradley, & Cuthbert, 1995), or nothing at all. The present study assessed three measures of covariation bias: initial expectancies, on-line outcome expectancies on a trial by trial basis, and a posteriori covariation estimates (cf. de Jong et al., 1998). Evidence of negative judgmental biases for social information in social anxiety is mixed. However, given that overestimated probabilities and exaggerated costs of threat may play an important role in the maintenance of anxiety disorders (Foa & Kozak, 1986), it was predicted that high socially anxious individuals would demonstrate a greater judgmental bias for negative social cues and unpleasant outcomes, throughout, and after the task, compared to low anxious individuals. Furthermore, following Hirsch & Mathews (1997) and Hirsch & Mathews (2000), it was predicted that high socially anxious individuals would show a reduced bias in positive outcome expectancies on-line, and following the task, compared with low socially anxious individuals.