Information processing in social phobia: a critical review

Information processing in social phobia: a critical review

Clinical Psychology Review, Vol. 21, No. 5, pp. 751±770, 2001 Copyright D 2001 Elsevier Science Ltd. Printed in the USA. All rights reserved 0272-7358...

171KB Sizes 0 Downloads 53 Views

Clinical Psychology Review, Vol. 21, No. 5, pp. 751±770, 2001 Copyright D 2001 Elsevier Science Ltd. Printed in the USA. All rights reserved 0272-7358/01/$ ± see front matter

PII S0272-7358(00)00067-2

INFORMATION PROCESSING IN SOCIAL PHOBIA: A CRITICAL REVIEW Nina Heinrichs and Stefan G. Hofmann Boston University

ABSTRACT. This review critically discusses the empirical evidence for information-processing biases in social phobia. Distortions in attention, interpretation, and memory processes are analyzed as they apply to individuals with social phobia. The literature provides evidence for a specific attentional bias towards socially threatening stimuli and a specific interpretational/ judgment bias towards self-relevant social information. However, there is little evidence to suggest that social phobia is associated with a memory bias for socially threatening stimuli. Furthermore, the relationship between the empirical evidence from information processing studies and the cognitive model of social phobia by Clark and Wells (1995) will be discussed. D 2001 Elsevier Science Ltd. All rights reserved. KEY WORDS. Information processes, Bias, Social phobia, Attention, Interpretation, Judgment, Perception, Memory. COGNITIVE THEORIES ASSUME that information processing contributes significantly to the maintenance of mood and anxiety disorders (e.g., Beck, Emery, & Greenberg, 1985; Mathews & MacLeod, 1994). Attention, memory, interpretation, and judgment processes have been most widely studied in anxiety disorders (Mathews & MacLeod, 1994). Overall, anxious individuals seem to selectively process threatening information. However, it remains unclear if anxious individuals process information selectively only if the content of the information completely matches the core of their anxiety or if they show a general bias towards emotionally loaded words (e.g., Asmundson & Stein, 1994; Martin, Williams, & Clark, 1991). Furthermore, it is unclear if selective processing occurs throughout all stages of information processing or if anxious individuals are prone to be biased in specific stages of the information processing flow (e.g., Cloitre, Cancienne, Heimberg, Holt, & Liebowitz, 1995; Lundh È st, 1996b). Moreover, not all anxiety disorders may be affected by these biases in &O

Correspondence should be addressed to Nina Heinrichs, Dipl.-Psych., Psychologisches Institut, Abteilung, Klinische Psychologie, Psychotherapie, und Diagnostik, Spielmannstraûe 12a, D-38106 Braunschweig, Germany. E-mail: [email protected] 751

752

N. Heinrichs and S. G. Hofmann

the same way. Cloitre et al. (1995), for example, published an article with the characteristic title ``Memory bias does not generalize across anxiety disorders.'' Indeed, there is some evidence suggesting that anxiety disorders might differ from each other in how threatening information is processed (Amir, McNally, & Wiegartz, 1996; McNally, Foa, & Donnell, 1989; Mogg, Mathews, & Weinman, 1987; Rapee, McCallum, Melville, Ravenscroft, & Rodney, 1994; Wilhelm, McNally, Baer, & Florin, 1996; Vrana, Roodman, & Beckham, 1995). Several information-processing models have been discussed (e.g., Beck & Clark, 1997; Beck et al., 1985; Bower, 1981; Foa & Kozak, 1986; Williams, Watts, MacLeod, & Mathews, 1988). Initially, it was hypothesized that anxious individuals are hypersensitive towards threatening information, which facilitates the processing of danger (Beck et al., 1985). It was therefore assumed that anxious individuals would show a bias towards threatening information (hypervigilance hypothesis). Foa and Kozak (1986) later argued that anxious individuals tend to inhibit or even completely avoid deep processing of threatening information, leading to ``cognitive avoidance'' of threatening stimuli (avoidance hypothesis, see also Mogg et al., 1987). Recently, these contradictory assumptions were integrated in a two-stage model of information processing (Mogg, Bradley, Bono, & Painter, 1997). This model suggests that anxious individuals who are hypervigilant to threatening information in the initial stage of its processing avoid such information in a later stage (hypervigilance ±avoidance hypothesis, e.g., Amir, Foa, & Coles, 1998b). Furthermore, it has been suggested that the bias only occurs if the presented information represents the core of the anxiety disorder (specificity hypothesis). It has also been stated that anxious individuals show overall impaired performance when processing information (Cloitre, Heimberg, Holt, & Liebowitz, 1992; Maidenberg, Chen, Craske, Bohn, & Bystritzky, 1996; Rapee, 1995). The goal of the present review is to summarize and discuss the empirical evidence for information processing biases coming from cognitive ± experimental approaches for social phobia. Furthermore, we will discuss this empirical evidence within the context of Clark and Wells' (1995) cognitive model of social phobia.

ATTENTION As of to date, at least 10 studies have been published using cognitive ±experimental tasks to explore attentional processes in social phobia (Amir, McNally, Riemann, et al., 1996; Asmundson & Stein, 1994; Cloitre et al., 1992; Hope, Rapee, Heimberg, & È st, 1996a; Maidenberg et al., Dombeck, 1990; Horenstein & Segui, 1997; Lundh & O 1996; Mattia, Heimberg, & Hope, 1993; McNeil et al., 1995; Niekerk, Moeller, & Nortje, 1999). Most of these studies used either the dot-probe paradigm or the Stroop task (e.g., Stroop, 1938), two common cognitive± experimental approaches to measure attentional processes. The dot-probe paradigm measures the distribution of visual attention. As part of a typical dot-probe experiment, participants are asked to press one of two buttons to identify the location of a dot that follows one of two words presented on a computer screen. These words typically vary in their emotional valence. The dot detection latencies determine whether visual attention has shifted toward or away from the threatening stimulus. A typical modified Stroop test asks participants to name the

Information Processing in Social Phobia

753

color of words with different emotional significance while ignoring the words' content (e.g., ``humiliation'' written in red). Dot-probe Paradigm Two studies used the dot-probe paradigm with social phobic individuals (Asmundson & Stein, 1994; Horenstein & Segui, 1997). Asmundson and Stein (1994) compared individuals with a generalized subtype of social phobia (GSP) with a normal control group. The authors replicated previous findings (MacLeod, Mathews, & Tata, 1986): Individuals with GSP responded faster to probes that followed social threat cues than probes that followed either neutral or physical threat cues. However, this result only occurred when cues appeared in the upper area of the monitor to which subjects are initially directing their attention. These results suggest that individuals who actively read threat stimuli show an attentional bias towards threatening information. In contrast, the attention did not shift towards the threatening cue when it appeared in the lower area. This suggests that selective attention only occurs if a threat cue is actively perceived. Furthermore, the speeded response toward probes in general indicates that patients with social phobia may exhibit generally heightened environmental awareness and show selective processing of social threat cues. Interestingly, the treatment outcome literature suggests that effective psychotherapy primarily leads to a decrease in negative self-focused attention, whereas the external focus of attention remains unchanged (Hofmann, 2000a; Woody, Chambless, & Glass, 1997). Future studies will need to integrate these different lines of attention research. Horenstein and Segui (1997) compared social phobics to individuals with panic disorder and a control group. The authors did not find an attentional bias towards threatening information among social phobics. However, panic disordered patients exhibited a tendency to react faster to physically threatening words when those words were presented in the upper area. A number of factors might have been related to these results, including sample parameters and stimuli material. Sample parameters are difficult to estimate in their influence since sample characteristics, including comorbid disorders, are often not sufficiently described. For example, Asmundson and Stein (1994) provided only limited information about comorbid disorders, although other anxiety disorders are also associated with cognitive biases (e.g., McNally et al., 1989; Wilhelm et al., 1996). Therefore, it might be possible that two or more disorders have no effect, an additive effect or an interactive effect on cognitive biases of socially threatening information. Furthermore, the social phobia sample in Asmundson and Stein's study yielded a BDI score of 16.4 indicating a mild to moderate mood disturbance, while the social phobia sample in Horenstein and Segui's study yielded a score of 9.6, indicating a normal mood. Thus, the attentional bias found by Asmundson and Stein might have been moderated by depression. Finally, the social phobia subtype issue has often not been adequately addressed (Horenstein & Segui, 1997). Similarly, it is not clear what role trait and state anxiety might play in information processing biases (e.g., MacLeod & Rutherford, 1992; Mathews & Mackintosh, 1998; Rusting, 1998). Stimulus material is another relevant variable. The word pools used generally differ from study to study, except for several Stroop task studies (e.g., Hope et al., 1990; Mattia et al., 1993; Niekerk et al., 1999). Horenstein and Segui used more physical threatening words like ``fracture'' and ``mutilated'' and more general threatening words like ``violence,'' ``danger,'' and ``victim,'' whereas Asmundson and Stein used

754

N. Heinrichs and S. G. Hofmann

more panic-related words like ``attack,'' ``collapse,'' and ``fatal.'' The chosen words should represent the ``core'' of the anxiety disorder. For example, panic disorder patients should ideally respond best to core anxious thoughts and sensations like ``suffocating,'' ``nausea,'' and ``collapse,'' whereas social phobics should respond best to words like ``incompetent'' and ``criticized.'' In contrast, both anxiety groups may respond to words such as ``humiliation,'' ``embarrassed,'' or ``losing control.'' Thus, a word pool is needed which most closely represents the ``anxiety core,'' and is ``cleaned'' from those words both clinical groups might react to. In addition, few studies included emotionally positive words. It might therefore be possible that participants reacted to threatening words only because they are emotionally loaded (Martin et al., 1991). Furthermore, future studies will need to include other types of stimulus material than just words. A study by Yuen (1994) suggests that high socially anxious individuals were slower at locating the dot if it appeared in the location where a negative face had been than if it appeared in the location where a neutral face had been. Interestingly, an opposite pattern of the results was found when using words as stimulus material, which suggests that semantic and facial stimuli might be processed differently in social phobia. It has been hypothesized that face-processing studies measure attention to actual social cues, whereas word-processing studies are more closely related to mental preoccupation (Clark & Wells, 1995). Therefore, it might be possible that the bias for faces is related to external cues of a social interaction, whereas the bias for words is more closely related to thoughts about self-perception. However, more studies are needed before any further conclusions can be drawn. Modified Stroop Task At least seven information-processing studies have used this paradigm to examine attentional processes in social phobia (Amir, McNally, Riemann, et al., 1996; Hope et È st, 1996a; Maidenberg et al., 1996; Mattia et al., 1993; McNeil et al., 1990; Lundh & O al., 1995; Niekerk et al., 1999). One of the earlier studies found a distinct pattern comparing social phobia and panic disorder (Hope et al., 1990). Individuals with social phobia showed longer color-naming latencies for words with a socially threatening connotation than for words with a neutral connotation. Similarly, individuals with panic disorder showed longer color-naming latencies for words with a physically threatening connotation than for words with a neutral connotation. This result is consistent with the specificity hypothesis. However, physically and socially threatening words were not directly matched to each other, but to two (different) pools of neutral words. Therefore, no direct comparisons can be made between the response latencies of socially and physically threatening words. Furthermore, emotionally positive words were not included. Thus, it remains unclear how anxious individuals would react to positive information as compared to neutral and negative information. It might be possible that attention in general is captured by information that is personally relevant rather than by information that is negative or neutral. Riemann and McNally (1995) found supporting evidence for this assumption in a student sample. It has been suggested that schemata have to be activated to unfold their anxietyprovoking effects (Beck et al., 1985). Hope et al. (1990) also mentioned that one had to expect smaller differences between the latencies if the schemata were not activated. However, anxiety ratings prior to engaging in the paradigm typically reveal

Information Processing in Social Phobia

755

only mild anxiety. Other studies using the State version of the State Trait Anxiety Inventory (STAI, Spielberger, Gorsuch, Lushene, Vagg, & Jacobs, 1983) report a mean score that is often smaller than the mean STAI-Trait score (e.g., Cloitre et al., 1992). Because most studies did not administer the state version repeatedly (e.g., preand post instruction), a general increase in state anxiety due to the general experimental setting cannot be separated from the specific anxiety-provoking potential of the experiment. We would also expect that activation of an anxious schema is accompanied by heightened state anxiety relative to how the participants are feeling in general. Otherwise, the information processing biases should occur independently from induced anxiety. A related problem is the unclear definition of an activated schema. Furthermore, the nature of the activating stimulus and other variables involved in a schema activation remain largely unexplored. Amir, McNally, Riemann, et al. (1996) found an inhibition effect of Stroop interference under high anxiety conditions. Although both social phobics and controls showed a reduction in response latencies across all word types, social phobics were better able to suppress the interference for socially threatening words. The authors suggested that, while both social phobics and controls increased their effort to override an attentional bias, social phobics were more adept at overriding the bias because it is specific to their anxiety. Niekerk et al. (1999) did not find an association between STAI-S scores and the extent of interference on the Stroop task. However, two-thirds of the participants were receiving psychopharmacotherapy at the time of the study and for some participants Afrikaan words were used, whereas others received English words. Moreover, the mean STAI-S score reported in this study was similar to the mean STAI-S score reported by Amir and colleagues. Therefore, another possible explanation for the unexpected outcome might be that participants were anxious enough to compensate for the interference. Williams, Mathews, and MacLeod (1996) reported in their review of the modified Stroop task that high-trait anxious participants who do not demonstrate only interference, respond in general more quickly to the Stroop stimuli (including neutral words). The authors suggested that the shorter response latency may signify that participants are adopting a conscious strategy to override the effect of the salient threatening stimuli. They further argued that this bias may represent an early warning system that facilitates escape from harm in nonclinical participants. Williams et al. hypothesized that a ``mental breakdown'' occurs when an individual can no longer expand the extra effort required to override the tendency for concern-related stimuli that capture their attention. As a result, patients would then be unable to ``exit'' the vicious cycle of emotional disorders causing an attentional bias and leading to increased salience of negative material. However, Amir and colleagues reported that patients with social phobia were actually better able to override the bias than were control participants. This finding is inconsistent with the Williams et al. (1996) hypothesis. Future studies should focus more closely on mental suppression and cognitive avoidance in social phobia. As already pointed out by Asmundson and Stein (1994), the modified Stroop task does not necessarily provide sufficient information about an attentional bias because the results are confounded with response biases. Future experiments can avoid the response bias by using the dot-probe paradigm, which requires a bias-free response. In sum, the literature suggests that social phobia is associated with an attentional bias towards socially threatening information if the information is presented in words. For facial cues, the attentional bias seems to be directed away from the source

756

N. Heinrichs and S. G. Hofmann

of information. Dependent on the task requirement, the performance can either be inhibited or facilitated by this attentional bias. However, it remains unclear if the specific attentional bias in social phobia reflects a bias that is associated with generally impaired (Cloitre et al., 1992; Mattia et al., 1993) or generally unimpaired (Maidenberg et al., 1996) processing capacity. Further research about the specific role of trait and state anxiety is needed in order to further explore the nature of the cognitive schema activation. Amir et al.'s study indicates that under high anxiety condition, the attentional bias might be overridden based on more effort, but this result remains to be replicated. In summary, studies utilizing the modified Stroop test suggest that patients with social phobia selectively attend to socially threatening information. However, they seem to suppress the processing of socially threatening information under high anxiety conditions. Such suppression may subsequently lead to insufficient encoding of social information and an impairment of the self-presentation as a social object.

INTERPRETATION AND JUDGMENT Several studies analyzed information processes related to interpreting and judging information as socially threatening (Amir, Foa, & Coles, 1998a; Amir et al., 1998b; de Jong, Merckelbach, BoÈgels, & Kindt, 1998; Foa, Franklin, Perry, & Herbert, 1996; Lucock & Salkovskis, 1988; Mellings & Alden, 2000; Rapee & Lim, 1992; Stopa & Clark, 2000; Wallace & Alden, 1997; Wells, Clark, & Ahmad, 1998). Interpretations about the meaning of a social situation as well as its cost influence an individual's readiness to enter and re-enter the situation. Furthermore, the meaning assigned to a social situation may influence attentional and memory processes. Foa and Kozak (1986) have suggested that overestimated probabilities and exaggerated costs of threatening activities or situations play an important role in the maintenance of anxiety disorders in general. They also suggested that, while social phobia is characterized by exaggerated cost estimations of negative outcome or consequence of a social situation (e.g., ``I will blush,'' ``Everybody will notice that I am anxious''), panic disorder may be better characterized by incorrect probability estimations of a negative event (e.g., ``If I have a severe panic attack, I will die''; Foa & Kozak, 1985). It has further been suggested that a situation needs to be interpreted as a potential threat before attentional processes can be adjusted accordingly (Mathews & Mackintosh, 1998). Furthermore, social situations usually entail uncertainty and ambiguity. Therefore, misinterpretations of these situations are more likely to occur. Thus, judgmental biases seem to be implicated in the etiology and maintenance of social phobia. Amir and Foa (2001) hypothesized that alternative meanings of an ambiguous event are typically activated non-consciously, but then compete with each other via mutual inhibition for entry to awareness. The authors further assumed that meanings which have been repeatedly accessed in the past tend to dominate in this competitive process, so that they are accepted without the need for conscious selection. In fact, individuals with social phobia seem to have a tendency to misinterpret social situations and draw more negative inferences from the available social stimuli (Amir et al., 1998a, 1998b; Foa et al., 1996; Hirsch & Mathews, 1997; Stopa & Clark, 2000; Wallace & Alden, 1997). All studies reviewed here found supporting evidence for judgmental biases. Paradigms that have been used to investigate an interpretation

Information Processing in Social Phobia

757

bias include interpretations of homographs (i.e., words with almost identical spelling but different meaning, Amir et al., 1998b) and interpretations of ambiguous situations (``vignettes,'' Amir et al., 1998a). Other procedures are based on subjective probability and cost ratings and are more descriptive in nature (Lucock & Salkovskis, 1988; Foa et al., 1996). In a recent study, Amir et al. (1998a) presented participants with ambiguous social scenarios. Participants were 32 individuals with generalized social phobia, 13 individuals with obsessive ± compulsive disorder and 15 low socially anxious control participants. Each scenario was followed by three possible interpretations. One explanation was positive, one negative, and one neutral. Participants were asked to rank the three interpretations according to how likely they thought the interpretations would come to their mind (self-relevant condition). The same ranking was conducted according to how likely they thought the interpretations would come to a typical person's mind (other-relevant condition). The authors found that individuals with generalized social phobia were more likely to interpret social scenarios as negative when the situations were rated as self-referent. The other groups did not exhibit such bias. Similar results were found in a recent study by Stopa and Clark (2000). Another study conducted by Amir et al. (1998b) presented subjects with sentences ending either in homographs or ending in non-homographs followed by a cue word. Half of the homographs implied a socially threatening meaning. The authors hypothesized that individuals with generalized social phobia would be slower to reject cue words following sentences ending in homographs with a possible socially relevant meaning than controls. Consistent with this hypothesis, results showed that participants with generalized social phobia showed greater response latencies. It was therefore concluded that individuals with generalized social phobia show an activation of the inappropriate meanings of socially relevant homographs followed by a later inhibition of these meanings, which provides evidence for the vigilance ± avoidance model. Another study by Lucock and Salkovskis (1988) presented 12 positive and 12 negative events to 12 social phobics and 40 controls. Participants were asked to rate the probability of the occurrence of various positive and negative social and nonsocial events. Results showed that social phobics exhibited higher probability ratings of negative social events but not negative nonsocial events, which supports the specificity hypothesis in the judgmental stage of information processing. However, because the control group and the social phobia group differed in the number of assessments they underwent, the results could have been due to a methodological artifact (Foa et al., 1996). Thus, Foa et al. (1996) attempted to replicate the study while investigating the role of subjective cost associated with negative social and nonsocial events. They presented 40 negative hypothetical events to 15 generalized social phobics and 15 low socially anxious individuals. Results supported the idea that generalized social phobics exhibit a content specific judgmental bias for probability and cost of negative social events but not for probability and cost of negative nonsocial events. Furthermore, changes in cost estimates of social events accounted for most of the variance in improvement of social anxiety after treatment. The authors concluded that it was the change in cost estimates rather than the change in probability estimates that mediated treatment efficacy. Foa et al. (1996) further hypothesized that cognitive errors in social phobia are resistant to change because they pertain not only to individual experiences but also to others' thoughts and feelings, which are largely inaccessible for disconfirmation.

758

N. Heinrichs and S. G. Hofmann

Lucock and Salkovskis's study differs from Foa et al.'s (1996) study in that the former intentionally excluded items related to a participants' own social performance in order to minimize the influence of anxiety-related deficits in behavior. Unfortunately, events with positive valence were not used in Foa and colleagues' study. Thus, the interesting result that socially anxious patients in Lucock and Salkovskis' study rated the probability of both socially and nonsocially positive events as lower than controls has not yet been replicated. The processing of positive social information has also been investigated by Wallace and Alden (1997). This study asked 32 individuals with generalized social phobia and 32 controls to participate in either a successful or unsuccessful social interaction. Participants were asked to rate their ability, perceptions of others' standards, social goals and emotional responses before and after the social interaction. Individuals with social phobia judged their social abilities much lower than they judged others' standard for a successful social interaction. The control group did not show this discrepancy. The difference between the groups was based on the social phobics' low judgment of their own ability. The groups did not differ in their judgments of the standards that others use to evaluate social interactions. Surprisingly, even successful social interactions elicited negative affect in patients with social phobia. Although participants recognized they had done well in a social interaction, they reported the same negative feelings as patients who perceived the social interaction as unsuccessful. Furthermore, the successful interaction did not improve the evaluation of their own social ability. The authors hypothesized that social phobics may believe that others would require more from them following a successful interaction and, therefore, that they would fall short of others' expectations in future interactions. As a result, social success does not increase social self-efficacy beliefs after a successful social interaction. Wallace and Alden further hypothesized that social phobics engage in specific behaviors to be potentially less offensive to their partners. The careful behavior selection accounts for the successful interactions preventing the alternative explanation that success is due to being a likeable and socially adept individual. Unfortunately, this pathway of processing positive social information is self-defeating, as individuals with social phobia find their own catastrophic beliefs confirmed regardless of the nature of the interaction. The actual outcome of a social interaction (positive or negative) therefore does not seem to provide a beneficial learning experience. In a related study, Rapee and Lim (1992) investigated how social phobics judged their behavior after an impromptu speech to a small audience. Speakers and members of the audience rated each speaker on a scale containing 12 specific items to evaluate the public speaking performance (e.g., voice shook, kept eye contact with audience) and five global evaluative items (e.g., kept an audience interested). All participants rated their own performance worse than the audience, but for individuals with social phobia, this discrepancy was greater than for controls. However, the discrepancy only emerged if global features of the speech were rated (e.g., the overall impression made on the audience) but not specific behaviors. The fear of negative evaluation was the main predictor of self-other discrepancy for global items when evaluating social performance. The authors concluded that the results may reflect a basic difference in the perception or processing of performance information between social phobics and controls which only occurs for self-performance but not for other performance. Similar results have been found with high socially anxious individuals in a recent study by Mellings and Alden (2000). These studies suggest that individuals

Information Processing in Social Phobia

759

with social phobia have a tendency to overestimate the costs and probabilities of negative social interactions. Furthermore, they tend to globally overestimate their performance in a social situation as negative and insufficient. Even when the social interaction is perceived as positive, they seem to be unable to process this information appropriately. Interestingly, however, negative outcomes are not overestimated if the social stimuli consist of faces (de Jong et al., 1998). In summary, a review of the literature suggests that individuals with social phobia exhibit an interpretational/judgmental bias towards socially threatening information. The studies reviewed here may be differentiated according to how they approached the judgment process. Some experiments explored this process by using questionnaires (Amir et al., 1998a; Foa et al., 1996; Lucock & Salkovskis, 1988; Wallace & Alden, 1997) while others used cognitive± experimental tasks (Amir et al., 1998b; de Jong et al., 1998). Regardless of the approach, evidence for an interpretational bias was found in clinical samples. However, the quasi-experimental design of these studies limits interpretation of the results. Furthermore, particularly in studies which rely on subjective reports, participants might have been able to guess the experimenter's hypothesis, which could have biased the results. Future studies on information processing in social phobia should use different methods (e.g., experimental paradigms, self-report, neuropsychological correlates/neuroimaging) and a variety of stimuli (including faces). Using different cognitive ± experimental paradigms in combination with other methods to investigate attentional processes, such as neuroimaging, would provide researchers with valuable information. Neuroimaging, for example, may elucidate neural mechanisms that are associated with social phobia. Furthermore, it may help researchers to explore how learning and memorizing socially threatening information alters brain functions and structure. A recent review demonstrates the utility of such a method for posttraumatic stress disorder (Tyron, 1999).

MEMORY It has been suggested that cognitive interpretations of threat reside in a partially activated state in memory (McNally, 1994). Therefore, individuals with social phobia should display memory biases favoring information about social threat (Clark & Wells, 1995). Several memory studies have been conducted with social phobics (Becker et al., È st, 1996b, 1997; Rapee et al., 1994; Stopa & 1999; Cloitre et al., 1995; Lundh & O Clark, 1993) and socially anxious individuals (Breck & Smith, 1983; Claeys, 1989; Foa, McNally, & Murdock, 1989; Mansell & Clark, 1999; Mellings & Alden, 2000; O'Banion & Arkowitz, 1977; Sanz, 1996; Smith, Ingram, & Brehm, 1983). These studies provide a different picture from the one yielded in attentional and interpretation studies. Memory studies typically fail to find any supporting evidence for the hypothesis that social phobics recall socially threatening information different from other information which provides no evidence for the specificity hypothesis. However, taking into account the participants and stimuli used, it seems striking that those studies which did yield positive results have been conducted with subclinical samples and positive or negative (i.e., socially desirable and undesirable) trait adjectives. The following sections will review these studies in further detail.

760

N. Heinrichs and S. G. Hofmann

Memory Processes in Socially Anxious Individuals Previous studies showed that high socially anxious participants tend to recall more negative words than low-socially anxious participants (Breck & Smith, 1983; Claeys, 1989; Mansell & Clark, 1999; O'Banion & Arkowitz, 1977; Smith et al., 1983). Although all of these studies used a self-referent encoding task (SRET; e.g., Breck & Smith, 1983; Claeys, 1989; Mansell & Clark, 1999; Smith et al., 1983), public-selfreferent or other-referent condition was not always included (e.g., Breck & Smith, 1983). Superior recall in the private self-referent condition was found in two of four studies. Mansell and Clark (1999) and Smith et al. (1983) found an enhanced recall in the public-self-referent condition, but not in the private self-referent condition. This enhanced recall, however, was only found when social evaluation was anticipated. With exception of Mansell and Clark's experiment, these studies did not control for depressive symptoms. However, even subclinical depressive symptoms have been found to cause a memory bias (Sanz, 1996). Two studies that used subclinical samples did not find an increased recall for socially threatening words (Foa et al., 1989; Sanz, 1996). However, Foa et al. (1989) used a subclinical sample of speech-anxious individuals, a very circumscribed sample compared to social phobics who are typically afraid of more situations than only giving speeches. Another study by Sanz (1996) used 12 socially threatening words which were divided into different encoding conditions. Thus, only a few words were used for content specific analyses and the recall for these words was averaged across private and public self-referent encoding conditions. A recent study conducted by Mellings and Alden (2000) analyzed memories for self-related and external information for a social interaction. They found that memories for social events are biased in favor of negative self-related information. The authors concluded that selective attentional processes rather than selective retrieval processes explain their results. In summary, the literature inconclusively suggests that non-clinical socially anxious individuals show a selective memory bias for socially threatening words. It remains to be seen whether such memory bias is caused by attentional processes, by encoding processes or by retrieval processes, and whether it will occur only in the context of an anticipated evaluation. Memory Processes in Social Phobia To date, at least six published studies have investigated memory biases in social È st, 1996b, phobia. While two studies found evidence for a memory bias (Lundh & O 1997), four studies did not find any empirical support for it (Becker et al., 1999; È st (1997) Cloitre et al., 1995; Rapee et al., 1994; Stopa & Clark, 1993). Lundh and O found a subtype specific memory bias: individuals with discrete social phobia completed more socially threatening and emotionally positive words than individuals with generalized social phobia and individuals without a mental disorder. The È st's study was a word-stem completion task, memory task used in Lundh and O which tests implicit memory. No explicit memory biases were found in any of the groups. The terms ``explicit'' and ``implicit'' memory refer to the effects of an episode that are expressed with or without awareness of remembering, respectively (Kelley & Lindsay, 1996; Schacter, 1992). Theoretical formulations in cognitive psychopathology focus on the semantic aspects of anxiety-provoking words. Therefore, word-stem completion may not be the best task to analyze memory processes

Information Processing in Social Phobia

761

in anxious individuals because word-stem completion is affected by the physical features of a stimulus rather than by its semantic properties (see also McNally, 1994, p. 132). Furthermore, this task measures both implicit and explicit memory processes because participants have previously seen the words and may be consciously relying on previous experience to complete the task (Nugent & Mineka, 1994). Other implicit memory tasks, for example a noise judgment paradigm (e.g., Jacoby, Allan, Collins, & Larwill, 1988), may be more informative because they seem to be less contaminated by explicit memory processes. The noise judge paradigm might capture automatic rather than strategic processes and may therefore be useful in determining the extent to which individuals with social phobia process automatic or strategic social information. In a noise judgment task, participants are to rate the volume of the background noise of familiar and unfamiliar sentences. If the noise accompanying familiar sentences is rated less loud than the noise accompanying unfamiliar sentences, implicit memory is used. To our knowledge, no study using this or other implicit memory paradigms with social phobia has yet been published. An unpublished study using a noise judgment paradigm (Amir, Foa, & Coles, 1998c) has found an implicit, but not an explicit memory bias in social phobia indicating that automatic processes may play a more significant role in the initial processing stage in social phobia than do strategic processes. Another È st (1996b) assessed memory for facial expressions, which may study by Lundh and O be more ecologically valid stimuli in social phobia than words because facial expressions are more directly related to social signals than semantic stimuli. This study compared social phobics and low socially anxious controls in their judgment of 20 faces as either critical or accepting. Afterwards, participants were faced with an unexpected recognition task, which included the 20 original faces and 60 distractor faces. The authors found that SP recognized a higher proportion of faces they originally judged as critical than those they judged as accepting. However, the study does not provide clear evidence for a memory bias because participants did not judge the distractor faces, which might have led to a response bias based on the familiarity of previously seen photos. More research is warranted using stimuli that differ from the semantic nature of words and that are more directly related to social approval or disapproval. The majority of studies to date have used words to study memory biases. No memory bias in social phobia has been found even though different memory types have been investigated. Rapee et al. (1994) conducted a series of four studies examining unexpected free recall, cued recall, recognition (all explicit memory) and word-stem completion (implicit memory). Furthermore, they investigated autobiographical memory and different encoding conditions (self-referent, other-referent) and examined different situational conditions by combining conventional encoding tasks with the instruction to imagine words in a specific context, including a social-evaluative context. None of these experimental procedures yielded evidence for a memory bias. However, depending on the paradigm and stimuli used, physically threatening words or neutral words were better remembered. Surprisingly, socially threatening words were not better recalled. Furthermore, Rapee and colleagues did not find an enhanced recall of hypothetical negative feedback or an enhanced recall of past failures. However, participants rated the imaginal task as not very threatening. Another study by Cloitre et al. (1995) compared performance in a perceptual memory task with performance in a semantic memory task. Both tasks failed to find a memory bias. Affectively valenced words (socially threatening and emotionally

762

N. Heinrichs and S. G. Hofmann

positive words) were better recalled by all participants. Becker et al. (1999) also failed to find a memory bias in social phobia using the same procedure that had successfully supported biased memory processes in panic disorder. In their study, emotionally positive words were better recalled than any other word category. Taken together, these studies provided no evidence for the specificity hypothesis, which would predict a group  word category interaction effect. Furthermore, there is no evidence for the general impairment hypothesis in social phobia because social phobics did remember the content of a conversation or parts of the surrounding (Stopa & Clark, 1993). A different recall study was recently conducted by Wells et al. (1998), who investigated how individuals view themselves if they activate images of past anxietyprovoking situations. Twelve individuals with social phobia and 12 controls were asked to recall and imagine a recent social situation and a recent nonsocial situation. Once participants activated an image, they were asked if they were viewing ``the situation as if looking out through'' their eyes (field perspective) or as though they were outside themselves, ``looking at themselves from an external point of view'' (observer perspective). Social phobics were more likely to take an observer perspective when they recalled social situations than non-patients. However, regarding nonsocial situations the two groups did not differ. All participants were more likely to take a field perspective. The observer perspective focuses on the social phobic individuals and thus, restricts access to information about others' behaviors, particularly reactions towards the social phobic person. This could be related to post event processing. Post event processing (Clark & Wells, 1995) refers to the process during which patients review a previous interaction in detail to evaluate their past and future social performance in social interactions. If social phobics take an observer perspective, they may ignore other available information (e.g., approving or neutral interaction behaviors towards the patient). Unfortunately, the authors did not formally assess the content of the memory and they did not report reliability and validity information on the newly invented scale for measuring the perspective. In summary, low and high socially anxious individuals probably process information differently with high socially anxious individuals favoring socially threatening information. However, in non-clinical socially anxious individuals this bias seems to show only if a socially anxious state is activated at the time of retrieval (e.g., by introducing a social-evaluative cue) and also appears to be limited to concerns about how one appears to others (Mansell & Clark, 1999; Smith et al., 1983). The previously found bias in processing negative information about oneself (e.g., private self-referent; Breck & Smith, 1983; O'Banion & Arkowitz, 1977) needs to be further evaluated. These studies did not measure the influence of depressive symptoms on the results and they did not include a public-self-referent condition. Mansell and Clark (1999) suggested that private self-referent encoding seems to be more relevant to depression since it reflects how one thinks about oneself. It remains unclear if this explanation applies to the former studies, which found a private self-referent memory bias, but failed to control for depressive symptoms. In clinical samples, no explicit memory bias has been found with semantic material. When facial expressions are used as È st, 1996b), individuals seem to show an explicit memory bias. material (Lundh & O However, no firm conclusions can be drawn due to the limited amount of research in this area. Furthermore, many of the studies that have demonstrated the presence of a selective memory bias have not been replicated. It is therefore possible that the selective processing of socially threatening information is confined to specific interactions of memory systems and material (e.g., implicit memory bias for semantic

Information Processing in Social Phobia

763

material, explicit memory bias for facial expressions), and that these interactions are moderated by different variables which may include the social evaluative context at the time of retrieval or the level of trait anxiety and depressive symptomatology. A major task for future researchers will be to identify possible moderators, such as trait anxiety, depressive symptoms or social phobic subtypes and avoidant personality disorder (Hofmann, 2000b). Future research needs to focus on developing a theoretical model of selective memory for socially threatening information that is consistent with the empirical evidence. For example, Beck and Emery's theory implies that the more anxiety people experience, the stronger the anxious schemata are activated, and the greater the degree of the information processing bias. However, there is little evidence for this assumption in social phobia. It has also been stated that threatening information might be processed differently in different stages of the information process (Amir et al., 1998b). It may be that social phobics direct their attention toward threatening social information at an early stage, but avoid this material at a later stage of information processing (Mogg et al., 1997). Thus, threat-relevant information can be readily identified and cognitively avoided. In summary, the literature reports little evidence to suggest that social phobia is associated with a memory bias. It might be possible that distorted retrieving processes are absent because memory processes require more strategic than automatic processing of information. As already pointed out by Amir et al. (1998b), abnormalities in automatic activation and inhibition may account for information processing biases in social phobia. When strategic processes are involved, patients with social phobia may be more likely to show an inhibition of threat-relevant meanings. In contrast, threatrelevant meanings may be more easily activated when automatic processes are involved. However, additional studies on memory processes in social phobia are needed before any firm conclusion can be drawn.

DISCUSSION AND FUTURE DIRECTIONS Our review of the current literature suggests that different information processing biases in social phobia exist. Specifically, individuals with social phobia seem to show an attentional and judgmental bias towards socially threatening information but there is little evidence for a memory bias in social phobia. Few studies have used other stimuli than words. Facial expressions of emotions have only recently been used. Preliminary data seem to indicate that these stimuli produce a different cognitive bias than words but further research will be needed to draw a firm conclusion. Future studies should also consider auditory stimuli and focus more on threshold and visuospatial processes. These studies may provide a new understanding of how socially threatening information is processed. For example, the dichotic listening paradigm may provide valuable information to investigate the facilitative effect of selective attention on performance because some studies have found lower auditory thresholds for threatening information related to the core anxiety concerns in other anxiety disorders (e.g., Foa & McNally, 1986). Fig. 1 summarizes the biases in attention, memory and interpretation of information in socially anxious subjects: Attentional processes select what will be fed onto the working memory, the more durable storage system (for a comprehensive description of working memory, see Baddeley, 1986, 1992). Thus, attention marks the first discrimination of incoming information. In social phobia, attention is selectively

764

N. Heinrichs and S. G. Hofmann

FIGURE 1. A Summary of the Information Processing Literature. Arrows represent the direction of information flow. The rectangular boxes refer to the output and types of information processing biases. The diamonds reflect decision points. directed towards socially threatening information. Therefore, in a flow of different stimuli the information which signifies social threat is more likely to be selected and transferred to working memory than other stimuli in the sensory trace. Different memory systems then use different processes to move contents from one information processing system to another, including attention, encoding, elaboration and retrieval. It appears that encoding and elaboration processes play an important role in social anxiety. For example, individuals can remember clearly what they said and how they thought about a stimulus at the time of encoding (encoding specificity principle, e.g., Craik & Tulving, 1975; Tulving, 1983). Furthermore, once stimuli pass the gate of (selective) attention, they become accessible to more extensively perceptual

Information Processing in Social Phobia

765

analysis. Stimuli which are being perceptually processed may activate various representations in semantic long-term memory. It might be possible that previously repeated activations of socially threatening stimuli yield a strong representation in memory. A given stimulus might therefore be easier identified as novel stimuli which could explain the faster detection of threat-relevant information in attentional studies. However, over time, social phobics may compensate for this automatic activation by using controlled strategies to inhibit the threat meaning. Because no memory bias has been found in social phobia, it does not seem likely that social phobics encode stimuli differently from controls. However, several studies with high socially anxious participants suggest that more negative and less positive words are recalled if these words have been encoded in a public-self-referent way and are retrieved during a socially threatening situation (e.g., Mansell & Clark, 1999). Encoding processes and elaboration processes are closely associated, and retrieval increases with stimulus elaborateness at time of encoding. The most elaborate set of knowledge we have is probably the knowledge about ourselves which has been referred to as the ``self-reference effect'' (e.g., Rogers, Kuiper, & Kirker, 1977). However, in social anxiety this self-referent effect seems to be stronger if the encoding is based on the public part of self-reference. It is unclear if social phobics also show a stronger public self-referent effect. None of the studies with a clinical sample has varied the different self-referent encoding conditions. Rapee et al. (1994) compared two encoding conditions while holding the retrieval test constant, but found no such bias. The same authors also studied autobiographical memory which contains both spatiotemporal and factual knowledge (Conway, 1996). This combination of different knowledge types could make it more difficult to find a memory bias. However, autobiographical memories are highly related to ones' self, which seems to mediate memory biases in social anxiety. Again, no bias for autobiographical memories was detected. Free recall, cued recall and recognition are most often used to measure memory. However, a single test of memory is an insufficient indicator of memory performance. Repeated testing, for example, will change the pattern of recall, usually resulting in improved recall (Roediger & Guynn, 1996). In social phobia, this effect may play an important role, especially in postevent processing. Recall may improve across tests because people use subjective retrieval cues to guide their recall (Roediger & Thorpe, 1978). Repeated processing of a previous social interaction (as it is typical for social phobia) provides a background on which those subjective retrieval cues might emerge. Unfortunately, changes of recall over time have not been sufficiently studied in cognitive psychology, let alone in social phobia (Roediger & Guynn, 1996). Becker et al. (1999) further recommended using incidental learning tasks and free recall tests because they improve the likelihood to find effects of anxiety on memory. The authors stated that explicit learning tasks and cued recall tests might involve too many cues, which could override effects of anxiety on memory. We suggest that experimenters simultaneously manipulate both encoding and retrieval conditions (encoding/retrieval paradigm; Tulving, 1983) because this design can investigate the interaction of encoding and retrieval processes. In sum, we conclude that the hypervigilance model and the avoidance model may both be valid representations of the information processing state in social phobia. The literature shows clear evidence for enhanced vigilance towards social stimuli in social phobic individuals. In contrast, there is little evidence to suggest that social phobic individuals show greater cognitive avoidance than controls aside from the

766

N. Heinrichs and S. G. Hofmann

absence of predicted biases. The studies by Amir, McNally, Riemann, et al. (1996) and Amir et al. (1998b) are notable exceptions. It is possible that hypervigilance and avoidance seem to apply to different stages in the process. Thus, a two-stage hypervigilance± avoidance model as suggested by Mogg et al. (1997) seems to provide the best fit with the data. In memory studies only subclincial samples showed a memory bias to social desirable and undesirable trait adjectives. It is unclear if encoding or retrieval processes or a combination of both are responsible for the reported biases. The evidence suggests that there may be separate encoding systems for different types of information (e.g., faces vs. words). This is also consistent with neuropsychological studies (Farah, Wilson, Drain, & Tanaka, 1998; Kanwisher, Tong, & Nakayama, 1998; O'Craven, Downing, & Kanwisher, 1999). Moreover, an encoding bias only becomes evident in a social-evaluative context at the time of retrieval. This suggests an encoding ± retrieval interaction. However, future studies examining information processes in social phobia need to clearly specify what part of the process is being examined, and if the focus is on storage/capacity questions or process questions. Furthermore, the use of the term ``memory bias'' is often poorly defined and lead to misinterpretations (e.g., encoding and retrieval biases can both refer to memory). Regarding interpretational processes, it seems likely that individuals with social phobia judge probabilities and costs of social activity outcomes erroneously. However, more cognitive ± experimental approaches need to be employed to investigate the involved processes more closely (e.g., by studying the reasoning process).

RELATIONSHIP TO CLARK AND WELLS' COGNITIVE MODEL How does the empirical evidence from information processing biases reviewed and summarized here relate to the cognitive model of social phobia as presented by Clark and Wells (1995)? This model partly accounts for the empirical evidence coming from information processing studies. Similar models have also been developed by others (e.g., Rapee & Heimberg, 1997). According to the Clark and Wells' model, social phobics misinterpret social situations because they hold dysfunctional beliefs about themselves and their behavior. Furthermore, Clark and Wells assume that individuals with social phobia engage in different, dysfunctional mental operations to reduce or hide their anxiety. The authors discuss at least four psychopathological processes that prevent social phobics from disconfirming their beliefs: When individuals with social phobia enter a social situation they shift their attention to detailed monitoring and observations of themselves. Such attentional shift produces an enhanced awareness of feared anxiety responses, interferes with processing the situation and other people's behavior, and produces interoceptive information which is used to construct an impression of themselves. Moreover, social phobics are thought to engage in a variety of safety behaviors to reduce the risk of rejection. These behaviors prevent them from critically evaluating their feared outcomes (e.g., shaking uncontrollably) and catastrophic beliefs. In addition, Clark and Wells assume that social phobics show an anxiety-induced performance deficit. They also overestimate how negatively other people evaluate their performance. Finally, the authors suggest that prior to and after a social event, social phobics think about the situation in detail primarily focusing on past failures, negative images of themselves in the situation, and other predictions of

Information Processing in Social Phobia

767

poor performance and rejection. Clark and Wells' cognitive model assumes that these anxious feelings and negative self-perception are strongly encoded in memory because they are processed in such detail. Their model is therefore consistent with many empirical studies suggesting that cost and probability estimations of social events are biased (Amir et al., 1998a, 1998b; Foa et al., 1996; Lucock & Salkovskis, 1988). Furthermore, Clark and Wells' model incorporates the dysfunctional beliefs found in Wallace and Alden's (1997) study. In sum, this model explains the crucial role of interpretational processes in the maintenance of social phobia. In addition, it predicts a self ± other discrepancy when social situations are judged. Empirical evidence supports this assumption (e.g., Mellings & Alden, 2000; Rapee & Lim, 1992). However, the model also suggests that this bias is manifest in specific physical and behavioral signs of anxiety. This has not consistently been found (Rapee & Lim, 1992). Finally, Clark and Wells explicitly assume the existence of a memory bias (encoding, elaboration, and retrieval). Specifically, the model assumes that social phobics show a memory bias towards socially threatening information (e.g., anxious feelings, negative self-perceptions) because they are predominant at the time of the actual social interaction and thus at the time of encoding. The authors also emphasize the importance of encoding material in a public-self-referent fashion, suggesting that encoding processes may be more crucial for maintaining social phobia than retrieval processes but they also assume that social phobics are more likely to recall socially threatening information. While this has been supported by memory studies with subclinical samples, no evidence has been found in clinical samples. Thus, there is only minimal support for Clark and Wells' cognitive model of social phobia regarding memory processes. Another assumption of the cognitive model that has not been confirmed relates to enhanced recall of past failures. Rapee et al. (1994) did not find that social phobics recall threat-related memories from their own lives better than controls. Clark and Wells' cognitive model further suggests that social success experiences can lose their positive connotation during postevent processing. However, Wallace and Alden's (1997) study indicates that, although social success can be recognized, it does not elicit positive affect. Instead, it is associated with negative affect because social phobics are afraid that their interaction partners also expects them to perform well in the future. In sum, we conclude that Clark and Wells' cognitive model receives supporting evidence from studies investigating attentional and judgmental biases. However, recent results suggest that the two processes of activation and inhibition in the dispensation of socially threatening information play an important role. Regarding memory processes the evidence remains inconclusive. The information processing flow chart (Fig. 1) is directly based on the empirical evidence and is partly consistent with the cognitive model as proposed by Clark and Wells. Future models of social phobia will need to integrate these findings in order to develop a cognitive theory of social phobia that would allow researchers to derive specific and testable predictions. AcknowledgmentsÐSupported in part by NIMH grant MH ±57326 to Dr. Hofmann. We thank Molly Choate for her valuable comments on an earlier version of this paper.

768

N. Heinrichs and S. G. Hofmann

REFERENCES Amir, N., & Foa, E. B. (2001). Cognitive biases in social phobia. In: S. G. Hofmann & P. M. DiBartolo (Eds.), From social anxiety to social phobia: multiple perspectives (pp. 254 ± 267). Needham Heights, MA: Allyn & Bacon. Amir, N., Foa, E. B., & Coles, M. E. (1998a). Negative interpretation bias in social phobia. Behaviour Research and Therapy, 36, 945 ± 957. Amir, N., Foa, E. B., & Coles, M. E. (1998b). Automatic activation and strategic avoidance of threat-relevant information in social phobia. Journal of Abnormal Psychology, 107, 285 ± 290. Amir, N., Foa, E. B., & Coles, M. (1998). Implicit and explicit memory bias in social phobia. Manuscript submitted for publication. Amir, N., McNally, R. J., & Wiegartz, P. S. (1996a). Implicit memory bias for threat in posttraumatic stress disorder. Cognitive Therapy and Research, 20, 625 ± 635. Amir, N., McNally, R., Riemann, B. C., Burns, J., Lorenz, M., & Mullen, J. T. (1996b). Suppression of the emotional stroop effect by increased anxiety in patients with social phobia. Behaviour Research and Therapy, 34, 945 ± 948. Asmundson, G. J. G., & Stein, M. B. (1994). Selective processing of social threat in patients with generalized social phobia: evaluation using a dot-probe paradigm. Journal of Anxiety Disorders, 8, 107 ± 117. Baddeley, A. D. (1986). Working memory. Oxford: Oxford Univ. Press. Baddeley, A. D. (1992). Working memory. Science, 255, 556 ± 559. Beck, A. T., & Clark, D. A. (1997). An information processing model of anxiety: Automatic and strategic processes. Behaviour Research and Therapy, 35, 49 ± 58. Beck, A. T., Emery, G., & Greenberg, R. (1985). Anxiety disorders and phobias: a cognitive perspective. New York: Basic Books. Becker, E. S., Roth, W. T., Andrich, M., & Margraf, J. (1999). Explicit memory in anxiety disorders. Journal of Abnormal Psychology, 108, 153 ± 163. Bower, G. H. (1981). Mood and memory. American Psychologist, 36, 129 ± 148. Breck, B. E., & Smith, S. H. (1983). Selective recall of self-descriptive traits by socially anxious and nonanxious females. Social Behavior and Personality, 11, 71 ± 76. Clark, D. M., & Wells, A. (1995). A cognitive model of social phobia. In: R. G. Heimberg, M. R. Liebowitz, D. A. Hope, & F. R. Schneier (Eds.), Social phobia: diagnosis, assessment, and treatment ( pp. 69 ± 93). New York, NY: Guilford Press. Claeys, W. (1989). Social anxiety, evaluative threat, and incidental recall of trait words. Anxiety Research, 2, 27 ± 43. Cloitre, M., Cancienne, J., Heimberg, R. G., Holt, C. S., & Liebowitz, M. R. (1995). Memory bias does not generalize across anxiety disorders. Behaviour Research and Therapy, 33, 305 ± 307. Cloitre, M., Heimberg, R. G., Holt, C. S., & Liebowitz, M. R. (1992). Reaction time to threat stimuli in panic disorder and social phobia. Behaviour Research and Therapy, 30, 609 ± 617. Conway, M. A. (1996). Autobiographical memory. In: E. L. Bjork, & R. A. Bjork (Eds.), Memory ( pp. 165 ± 194). San Diego, CA: Academic Press. Craik, F. I. M., & Tulving, E. (1975). Depth of processing and retention of words in episodic memory. Journal of Experimental Psychology, 104, 268 ± 294. de Jong, P. J., Merckelbach, H., Boegels, S., & Kindt, M. (1998). Illusory correlation and social anxiety. Behaviour Research and Therapy, 36, 1063 ± 1073. Farah, M. J., Wilson, K. D., Drain, M., & Tanaka, J. N. (1998). What is ``special'' about face perception? Psychological Review, 105, 482 ± 498. Foa, E. B., Franklin, M. E., Perry, K. J., & Herbert, J. D. (1996). Cognitive biases in generalized social phobia. Journal of Abnormal Psychology, 105, 433 ± 439. Foa, E. B., & Kozak, M. J. (1985). Treatment of anxiety disorders: Implications for psychopathology. In: A. H. Tuma, & J. D. Maser (Eds.), Anxiety and the anxiety disorders ( pp. 421 ± 452). Hillsdale, NJ: Erlbaum. Foa, E. B., & Kozak, M. J. (1986). Emotional processing of fear: Exposure to corrective information. Psychological Bulletin, 99, 20 ± 35. Foa, E. B., & McNally, R. J. (1986). Sensitivity to feared stimuli in obsessive ± compulsives: A dichotic listening analysis. Cognitive Therapy and Research, 10, 477 ± 485. Foa, E. B., McNally, R. J., & Murdock, T. (1989). Anxious mood and memory. Behaviour Research and Therapy, 27, 141 ± 147. Hirsch, C., & Mathews, A. (1997). Interpretative inferences when reading about emotional events. Behaviour Research and Therapy, 35, 1123 ± 1132.

Information Processing in Social Phobia

769

Hofmann, S. G. (2000a). Self-focused attention before and after treatment of social phobia. Behaviour Research and Therapy, 38, 717 ± 725. Hofmann, S. G. (2000b). Treatment of social phobia: Potential mediators and moderators. Clinical Psychology: Science and Practice, 7, 3 ± 16. Hope, D. A., Rapee, R. M., Heimberg, R. G., & Dombeck, M. J. (1990). Representations of the self in social phobia: Vulnerability to social threat. Cognitive Therapy and Research, 14, 177 ± 189. Horenstein, M., & Segui, J. (1997). Chronometrics of attentional processes in anxiety disorders. Psychopathology, 30, 25 ± 35. Jacoby, L. L., Allan, L. G., Collins, J. C., & Larwill, L. K. (1988). Memory influences subjective experience: Noise judgment. Journal of Experimental Psychology: Learning, Memory and Cognition, 14, 240 ± 247. Kanwisher, N., Tong, F., & Nakayama, K. (1998). The effect of face inversion on the human fusiform face area. Cognition, 68, 1 ± 11. Kelley, C. M., & Lindsay, D. S. (1996). Conscious and unconscious forms of memory. In: E. L. Bjork, & R. A. Bjork (Eds.), Memory ( pp. 33 ± 63). San Diego, CA: Academic Press. Lucock, M. P., & Salkovskis, P. M. (1988). Cognitive factors in social anxiety and its treatment. Behaviour Research and Therapy, 26, 297 ± 302. È st, L. G. (1996a). Stroop interference, self-focus and perfectionism in social phobics. Lundh, L.-G., & O Personality and Individual Differences, 20, 725 ± 731. È st, L.-G. (1996b). Recognition bias for critical faces in social phobics. Behaviour Research Lundh, L.-G., & O and Therapy, 34, 787 ± 794. È st, L.-G. (1997). Explicit and implicit memory bias in social phobia. The role of subLundh, L.-G., & O diagnostic type. Behaviour Research and Therapy, 35, 305 ± 317. MacLeod, C., Mathews, A., & Tata, P. (1986). Attentional bias in emotional disorders. Journal of Abnormal Psychology, 95, 15 ± 20. MacLeod, C., & Rutherford, E. M. (1992). Anxiety and the selective processing of emotional information: Mediating roles of awareness, trait and state variables, and personal relevance of stimulus materials. Behaviour Research and Therapy, 30, 479 ± 491. Maidenberg, E., Chen, E., Craske, M., Bohn, P., & Bystritsky, A. (1996). Specify of attentional bias in panic disorder and social phobia. Journal of Anxiety Disorders, 10, 529 ± 541. Mansell, W., & Clark, D. M. (1999). How do I appear to others? Social anxiety and processing of the observable self. Behaviour Research and Therapy, 37, 419 ± 434. Martin, M., Williams, R. M., & Clark, D. M. (1991). Does anxiety lead to selective processing of threatrelated information? Behaviour Research and Therapy, 29, 147 ± 160. Mathews, A., & Mackintosh, B. (1998). A cognitive model of selective processing in anxiety. Cognitive Therapy and Research, 22, 539 ± 560. Mathews, A., & MacLeod, C. (1994). Cognitive approaches to emotion and emotional disorders. Annual Review of Psychology, 45, 25 ± 50. Mattia, J. I., Heimberg, R. G., & Hope, D. A. (1993). The revised stroop color-naming task in social phobics. Behaviour Research and Therapy, 31, 305 ± 313. McNally, R. J. (1994). Panic disorder: a critical analysis. New York, NY: Guilford Press. McNally, R. J., Foa, E. B., & Donnell, C. D. (1989). Memory bias for anxiety information in patients with panic disorder. Cognition and Emotion, 3, 27 ± 44. McNeil, D. W., Ries, B. J., Taylor, L. J., Boone, M. L., Carter, L. E., Turk, C. L., & Lewin, M. R. (1995). Comparison of social phobia subtypes using stroop tests. Journal of Anxiety Disorders, 9, 47 ± 57. Mellings, T. M. B., & Alden, L. E. (2000). Cognitive processes in social anxiety: the effects of self-focus, rumination, and anticipatory processing. Behaviour Research and Therapy, 38, 243 ± 257. Mogg, K., Bradley, B. P., 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. Niekerk, J. K., Moeller, A. T., & Nortje, C. (1999). Self-schema in social phobia and panic disorder. Psychological Reports, 84, 843 ± 854. Nugent, K., & Mineka, S. (1994). The effect of high and low trait anxiety on implicit and explicit memory tasks. Cognition and Emotion, 8, 147 ± 164. O'Banion, K., & Arkowitz, H. (1977). Social anxiety and selective memory for affective information about the self. Social Behavior and Personality, 5, 321 ± 328. O'Craven, K. M., Downing, P. E., & Kanwisher, N. (1999). FMRI evidence for objects as the units of attentional selection. Nature, 401, 584 ± 587.

770

N. Heinrichs and S. G. Hofmann

Rapee, R. M. (1995). Descriptive psychopathology of social phobia. In: R. G. Heimberg, M. R. Liebowitz, D. A. Hope, & F. R. Schneier (Eds.), Social phobia: diagnosis, assessment, and treatment ( pp. 41 ± 66). New York, NY: Guilford Press. Rapee, R. M., & Heimberg, R. G. (1997). A cognitive ± behavioral model of anxiety in social phobia. Behaviour Research and Therapy, 35, 741 ± 756. Rapee, R. M., & Lim, L. (1992). Discrepancy between self and observer ratings of performance in social phobics. Journal of Abnormal Psychology, 101, 728 ± 731. Rapee, R. M., McCallum, S. L., Melville, L. F., Ravenscroft, H., & Rodney, J. M. (1994). Memory bias in social phobia. Behaviour Research and Therapy, 32, 89 ± 99. Riemann, B. C., & McNally, R. J. (1995). Cognitive processing of personally relevant information. Cognition and Emotion, 9, 325 ± 340. Roediger, H. L., & Guynn, M. J. (1996). Retrieval processes. In: E. L. Bjork, & R. A. Bjork (Eds.), Memory ( pp. 197 ± 236). San Diego, CA: Academic Press. Roediger, H. L., & Thorpe, L. A. (1978). The role of recall time in producing hypermnesia. Memory and Cognition, 6, 296 ± 305. Rogers, T. B., Kuiper, N. A., & Kirker, W. S. (1977). Self-reference and the encoding of personal information. Journal of Personality and Social Psychology, 35, 677 ± 688. Rusting, C. L. (1998). Personality, mood, and cognitive processing of emotional information: three conceptual frameworks. Psychological Bulletin, 124, 165 ± 196. Sanz, J. (1996). Memory biases in social anxiety and depression. Cognition and Emotion, 10, 87 ± 105. Schacter, D. L. (1992). Understanding implicit memory: A cognitive neuroscience approach. American Journal of Psychology, 4, 559 ± 569. Smith, T. W., Ingram, R. E., & Brehm, S. S. (1983). Social anxiety, anxious self-preoccupation, and recall of self-relevant information. Journal of Personality and Social Psychology, 44, 1276 ± 1283. Spielberger, C. D., Gorsuch, R. L., Lushene, R. E., Vagg, P. R., & Jacobs, G. A. (1983). Manual for the StateTrait Anxiety Inventory. Palo Alto, CA: Consulting Psychologists Press. Stopa, L., & Clark, D. M. (1993). Cognitive processes in social phobia. Behaviour Research and Therapy, 3, 255 ± 267. Stopa, L., & Clark, D. M. (2000). Social phobia and interpretation of social events. Behaviour Research and Therapy, 38, 273 ± 283. Stroop, J. R. (1938). Factors affecting speed in serial verbal reactions. Psychological Monographs, 50, 38 ± 48. Tyron, W. W. (1999). A bidirectional associative memory explanation of posttraumatic stress disorder. Clinical Psychology Review, 19, 789 ± 818. Tulving, E. (1983). Elements of episodic memory. New York, NY: Oxford University Press. Vrana, S. R., Roodman, A., & Beckham, J. C. (1995). Selective processing of trauma-relevant words in posttraumatic stress disorder. Journal of Anxiety Disorders, 9, 515 ± 530. Wallace, S. T., & Alden, L.-E. (1997). Social phobia and positive social events: The price of success. Journal of Abnormal Psychology, 106, 416 ± 424. Wells, A., Clark, D., & Ahmad, S. (1998). How do I look with my minds eye: Perspective taking in social phobic imagery. Behaviour Research and Therapy, 36, 631 ± 634. Wilhelm, S., McNally, R. J., Baer, L., & Florin, I. (1996). Directed forgetting in obsessive ± compulsive disorder. Behaviour Research and Therapy, 34, 633 ± 641. Williams, J. M. G., Mathews, A., & MacLeod, C. (1996). The emotional Stroop task and psychopathology. Psychological Bulletin, 120, 3 ± 24. Williams, J. M. G., Watts, F. N., MacLeod, C., & Mathews, A. (1988). Cognitive psychology and emotional disorders. Chichester, England: Wiley. Woody, S. R., Chambless, D. L., & Glass, C. R. (1997). Self-focused attention in the treatment of social phobia. Behaviour Research and Therapy, 35, 117 ± 129. Yuen, P. K. (1994). Social anxiety and the allocation of attention: Evaluation using facial stimuli in a dot-probe paradigm. Unpublished research project, Department of Experimental Psychology, University of Oxford, UK.