Beck and Emery’s (1985) cognitive model of anxiety

Beck and Emery propose that anxiety disorders are maintained by a cognitive–affective–physiological interaction that is fueled by self-knowledge stored in long-term memory. This knowledge stems from past experiences and forms a cognitive set of assumptions (schemas) composed of rigid and inflexible beliefs about the self, others, and the world. Beck and Emery’s cognitive model therefore takes a schema-based information-processing perspective. They suggest that individuals with anxiety have hyperactivated negative schemas, leading to preferential processing of threat-consistent information, and schemas that are hyposensitive to cues signaling safety, leading to underestimations of personal coping recourses and of the level of safety in the environment.

For individuals with SAD, social situations involving perceived social scrutiny activate these dysfunctional schemas (Beck & Emery, 1985). At the deepest level of the cognitive system are unconscious, unconditional beliefs about the self (e.g., “I am boring”) and negative views of others (e.g., “people are judgmental”). At the surface level are more conscious conditional rules (e.g., “if I make a mistake, others will laugh at me”). Information consistent with these self-beliefs (e.g., “people think I am a failure”) becomes selected preferentially and information inconsistent with these beliefs (e.g., positive social interaction) is ignored or discarded. Such selective processing of negative self and social information increases the sense of vulnerability, heightening anxiety and confirming and perpetuating negative views of the self.

Clark and Wells’ (1995) cognitive model of SAD

According to Clark and Wells, heightened self-focused attention, negative observer-perspective images of the perceived self, and negative self-evaluation all contribute to maintaining the dysfunctional patterns of social anxiety. Derived from self-presentation models (see Schlenker & Leary, 1982), a core feature of the model is that socially anxious individuals hold a strong desire to convey a favorable impression of themselves to others in social situations, but believe that they may not have the ability to do so. These negative performance expectations trigger a processing mode termed “processing of the self as a social object” (Clark, 2001, p. 401). When engaged in this processing mode, socially anxious individuals shift their attention from the environment to detailed monitoring of themselves (called self-focused attention). Self-focused attention increases awareness of feared anxiety responses and perpetuates the belief that other people perceive the individual in the same negative manner in which the individual sees themselves.

This negative impression of the “observable self” (p. 71) is often described by individuals with social anxiety as a “compelling feeling” (p. 71); however, Clark and Wells (1995) suggest that this negative representation of the self can also include negative imagery. Negative self-images are conceptualized as distorted mental pictures representing an individual’s feared outcome (e.g., being embarrassed in social situations). These images tend to be idiosyncratic, are viewed from the observer perspective, and can hijack attentional resources. Focusing on this self-image in social situations increases the perception that others are noticing anxiety symptoms consistent with the image. In this way, self-focused attention directed toward somatic symptoms, thoughts, and/or images is said to increase anxiety in social situations as well as to bias processing of information.

Consistent with Beck and Emery’s (1985) model, the tendency of socially anxious individuals to interpret situations in a threatening manner is also linked to dysfunctional beliefs by Clark and Wells (1995). The authors distinguish between three categories of beliefs: beliefs in excessively high standards (e.g., “I must get everyone’s approval”), conditional beliefs about social evaluation (e.g., “If someone does not approve of me it must be my fault”), and unconditional beliefs about the self (e.g., “I am uninteresting”). The model also suggests that the use of safety behaviors and engagement in post-event rumination following a social event not only exacerbate anxiety, but increase self-focused attention and prevent disconfirmation of these negative self-beliefs.

Rapee and Heimberg’s (1997) cognitive model of SAD

Similar to Clark and Wells (1995), Rapee and Heimberg (1997) suggest that when social situations are encountered, individuals with SAD focus on an internal mental representation of the self as seen by the audience. This mental representation is described as a distorted image that may be based on past negative social experiences that are consistent with negative core beliefs and self-schemas (as described by Beck & Emery, 1985; the importance of imagery is further emphasized in the updated model by Heimberg, Brozovich, & Rapee, 2010). Along the lines of self-presentation theory (Schlenker & Leary, 1982), a core feature of the model is the comparison of this mental representation of self with the perceived expectation of the audience. According to the model, individuals with SAD have a strong desire to be accepted by others. They also believe that others hold high expectations for their social performance, but at the same time assume that they will be unable to live up to these standards. As a result, individuals with SAD expect greater negative evaluation from others (the updated model also emphasizes the role of positive evaluation, see Heimberg et al., 2010), which in turn produces heightened anxiety.

Whereas Clark and Wells’ (1995) model suggests that self-focused attention is the central attentional process maintaining anxiety, Rapee and Heimberg (1997) assert that socially anxious individuals attend to both this internal mental representation of the self and to external threat cues (e.g., behaviors indicating negative evaluation from audience members). Rapee and Heimberg argue that this attentional monitoring is not done in isolation; rather, there exists an interactive relationship between monitoring the mental representation of self and the monitoring of attention toward social threat in the environment. For example, biased detection of audience behaviors (e.g., yawning) results in greater focus on internal self-representations (e.g., cognitions regarding how boring one is). Focusing on internal self-representations also leads to an increase in internal anxiety sensations, as well as an increase in the detection of negative audience behaviors (those that are in line with negative self-appraisals). Accordingly, both monitoring of mental representations of the self and external cues in the environment can heighten anxiety, hinder social performance, and preclude the perception of information inconsistent with social fears.

Hofmann’s (2007) cognitive model of SAD

Consistent with previous conceptualizations of SAD, Hofmann (2007) suggests that individuals with social anxiety experience apprehension in social situations because they perceive the social standards (i.e., expectations and social goals) of performance to be excessively high and doubt their ability to meet those standards. In this way, the model incorporates theory described in self-presentation models of the disorder (as in Clark & Wells, 1995; Rapee & Heimberg, 1997). According to Leary and Kowalski (1995), the goal for most socially anxious individuals in a social situation is to make a desired impression on others. However, Hoffman argues that socially anxious individuals have a deficiency in their ability to define attainable social goals. Moreover, they experience difficulty in selecting achievable behavioral strategies to reach these goals. These processes lead to increases in social apprehension and self-focused attention. This attention is directed at both internal and external threat cues (as in the Rapee and Heimberg model); however, the model emphasizes the role of heightened self-focused attention in SAD (consistent with the model of Clark and Wells).

In discussing the role of negative self-perception as a maintaining factor in the disorder, Hofmann (2007) explicitly incorporates research relating to self-discrepancy theory (Higgins, 1987), making a conceptual departure from previous cognitive-behavioral models (Beck & Emery, 1985; Clark & Wells, 1995; Rapee & Heimberg, 1997). Self-discrepancy theory postulates that people compare themselves to internalized standards, or domains of the self. Three basic domains of the self exist: the actual self (i.e., perceived attributes that either themselves or others believe they possess), the ideal self (i.e., perceived attributes that either themselves or others hope, or wish they possessed), and the ought self (i.e., perceived attributes that themselves or others believe it is their duty or responsibility to possess). The model emphasizes that discrepancies among socially anxious individuals’ self-domains may underlie their fear that they will be unable to convey a desired impression to others in social situations.

Moscovitch’s (2009) cognitive model of SAD

Moscovitch argues that previous cognitive-behavioral models of SAD (Clark & Wells, 1995; Rapee & Heimberg, 1997; Hofmann, 2007) are unsatisfactory because they confuse feared stimuli (i.e., the focus of anxiety) with feared consequences (i.e., feared outcomes when stimuli are present; Moscovitch, 2009, p. 2). Instead, he contends that practitioners involved in assessing and treating socially anxious individuals should re-focus their attention from targeting patients’ generic feared social situations (e.g., fear of evaluation) to targeting the core feared stimuli in SAD; namely, specific self-attributes that individuals with social anxiety perceive to be flawed or deficient. This view is built on the consensus that a negative, distorted self-view is central to SAD (e.g., Clark & Wells, 1995; Rapee & Heimberg, 1997).

Moscovitch (2009) proposes a typology of self-fears that include concerns about social skills and behaviors (e.g., “I will do something stupid”), showing signs of anxiety (e.g., “I will sweat”), physical appearance (e.g., “I am ugly”) and character (e.g., “I am boring”). During the development of the Negative Self-Portrayal Scale (used to assess perceived deficiencies in self-attributes; Moscovitch & Huyder, 2011), these four deficiencies were reduced to three subscales, concerns about: social competence, physical appearance, and showing signs of anxiety. As the core fear or threat in SAD, the model suggests that it is the activation of these self-attributes in anticipation of or during social situations that leads to emotional distress and maladaptive behavioral responses, including the use of safety behaviors designed to conceal perceived flaws in self-attributes and to prevent feared consequences. Thus, whereas Clark and Wells (1995) assert that people with social anxiety fear they will behave in a socially inept fashion, Moscovitch (2009) emphasizes broader feared self-dimensions.

Stopa’s (2009) model of SAD

Stopa argues that despite consensus that self-related constructs are an important maintaining factor in SAD (Clark & Wells, 1995; Hofmann, 2007; Rapee & Heimberg, 1997), previous conceptualizations of the disorder take a limited view of the self into consideration. These models often recognize only one or two aspects of the self and do not capture the construct’s full complexity. Stopa (2009) acknowledges that this oversimplification may be due to the models’ aims to provide theoretical frameworks from which treatment may be derived. However, in doing so, she states that we may be ignoring important information regarding aspects of the self and self-processes that may significantly contribute to the continuation of social anxiety.

Stopa (2009) proposes that the self can be organized into three broad aspects: content, structure, and process. Content refers to information about the self and the way this information is represented. For example, knowledge about oneself can exist in the form of verbal statements (e.g., “I am boring”) or be represented visually as images. Structure describes the way information about the self is organized, which can determine what aspects of self-knowledge are accessed at any given time. For example, individuals may have a more compartmentalized sense of self (i.e., little to no overlap between different self-attributes across self-aspects) or an integrated self-organization (i.e., duplication of different self-attributes across self-aspects; see Showers, 1992). Process refers to how attention is allocated to self-relevant information and the strategies that are used to evaluate and monitor information about the self.

To date, Stopa (2009) argues that cognitive-behavioral models have primarily focused on the content of the self-concept (e.g., the way that self-images, self-schemas, and negative thoughts and beliefs about the self contribute to SAD) and on one aspect of self-related processes, attentional biases. In contrast, relatively few researchers have incorporated the aspect of self-structure into models of SAD; yet knowing how self-knowledge and information is stored and organized could lead to more targeted and effective interventions.

Self-imagery and imagery rescripting

Several cognitive-behavioral models of SAD posit that negative self-images play a role in the maintenance of the disorder (Clark & Wells, 1995; Hofmann, 2007; Rapee & Heimberg, 1997). Research has found that individuals with social anxiety report experiencing negative self-images in social situations (for a review see Ng, Abbott, & Hunt, 2014). These self-images are often recurrent as they tend to occur in different social situations and are linked in meaning and content to prior unpleasant social events (Hackmann, Clark, & McManus, 2000). Research has also demonstrated the deleterious effects of negative-self imagery. For example, negative self-images tend to increase the perceived visibility of anxiety symptoms, poor performance appraisal (Hirsch, Mathews, Clark, Williams, & Morrison, 2003; Stopa & Jenkins, 2007), negative thoughts, and self-focused attention (Makkar & Grisham, 2011). This relationship between self-focused attention and negative self-images is consistent with the models’ emphases on the role of heightened self-focused attention (Clark & Wells, 1995; Hofmann, 2007), and preferential allocation of attention (Rapee & Heimberg, 1997) in social anxiety. Focusing on these distorted self-images has also been found to increase post-event rumination following a speech task (Makkar & Grisham, 2011), as predicted by Hofmann (2007), and increase the use of safety behaviors (Hirsch, Meynen, & Clark, 2004), as predicted by Clark and Wells (1995). Focusing on these negative self-images has also been found to decrease explicit self-esteem (Hulme, Hirsch, & Stopa, 2012).

A number of contemporary CBT treatment programs for SAD (e.g., Clark et al., 2003; Rapee, Gaston, & Abbott, 2009) have been developed that include techniques (e.g., video feedback, behavioral experiments, and surveying other people’s observations) to modify distorted self-images. To correct negative self-images using video feedback, clients are asked to (1) visualize how they think they will appear prior to watching the video, (2) specifically operationalize what their negative behaviors will look like (e.g., how red the blushing will be), and (3) watch the video from an observer point-of-view, ignoring the negative thinking and feelings that may bias the perception of performance (e.g., Harvey, Clark, Ehlers, & Rapee, 2000). In this way, video feedback offers clients the opportunity to discover that they do not appear as they think they do, consequently learning that their self-impressions are inaccurate (e.g., Rapee & Hayman, 1996). Studies documenting the effectiveness of video feedback as a tool to modify distorted self-images during CBT have found that videotaped feedback decreases social anxiety ratings, perceived social cost ratings, and increases positive appraisals of performance (e.g., Laposa & Rector, 2014; see also Rapee et al., 2009). Having video feedback across CBT sessions also tends to increase performance ratings and to decrease self-focused attention for the following feedback session (Laposa & Rector, 2014).

Despite support for the efficacy of video feedback in modifying distorted images and social anxiety, these techniques are primarily present-focused and do not directly modify the early memories and experiences that tend to be linked to self-images. While this may not be an issue for some clients (e.g., those who have an image unrelated to a past event), for clients who experience negative imagery that is linked to a past traumatic event, Wild and Clark (2011) suggest that use of only present-focused techniques may produce modest treatment responding. The authors argue that for these clients, new advances in CBT treatment techniques, such as “imagery rescripting” (i.e., a pre-existing negative self-image is identified, challenged, and then transformed into a more benign mental image or a new positive mental image, see Wild & Clark, 2011), may prove more efficacious.

Investigators have begun to examine the efficacy of using imagery rescripting as an intervention to correct distorted self-images and to improve social anxiety symptoms (e.g., Wild, Hackmann, & Clark, 2008). CBT for SAD including imagery rescripting has been found to be superior to in vivo exposure with applied relaxation (Clark et al., 2006). Other smaller trials have found that imagery rescripting is associated with reductions in fear of negative evaluation, negative core beliefs, and social anxiety (e.g., Frets, Kevenaar, & Heiden, 2014;Lee & Kwon, 2013; Nilsson, Lundh, & Viborg, 2012; Wild, Hackmann, & Clark, 2007, 2008). However, these studies examine the efficacy of imagery rescripting as a standalone procedure, rather than within the context of CBT. To overcome this issue, McEvoy and Saulsman (2014) developed the imagery-enhanced cognitive group behavioral therapy protocol (IE-CGBT). IE-CGBT includes techniques such as video feedback and imagery rescripting, and utilizes imagery-based techniques in all components of the CBT program (i.e., cognitive restructuring, behavioral experiments, and attention training). In comparing IE-CGBT to a control treatment without the self-imagery enhancements, McEvoy, Erceg-Hurn, Saulsman, and Thibodeau (2015) found that more clients completed treatment in the IE-CGBT condition (91% vs. 65%). Furthermore, while effect sizes were large for both treatment protocols, they were significantly higher for IE-CGBT. A higher proportion of the IE-CGBT clients also achieved clinically significant change according to the reliable change index. These findings suggest that modifying distorted self-images through IE-CGBT may be an efficacious treatment protocol for SAD. However, these findings are preliminary and more research is needed to replicate and extend these results.