Social Phobia 353
empiriCal status of the Cognitive moDelOver the last decade the volume of correlational and experimental research that has dem-
onstrated an information- processing bias for social threat information in social phobia
has grown exponentially (for reviews, see Alden & Taylor, 2004; Bögels & Mansell;
2004; D. M. Clark, 2001; D. M. Clark & McManus, 2002; Heimberg & Becker, 2002;
Heinrichs & Hofmann, 2001; Hirsch & Clark, 2004; Wilson & Rapee, 2004). Critics
argue that the role of biased cognitive processes in social anxiety remains unclear, espe-
cially as it pertains to etiological significance (Beidel & Turner, 2007; see also Stravyn-
ski, Bond, & Amado, 2004). In their review Heinrichs and Hofmann (2001) concluded
there was partial support for the cognitive model, especially for biased attention and
interpretation of self- referent social information. From our own review we believe there
is strong empirical support for cognitive dysfunction in social phobia, although we agree
that prospective studies and more sophisticated laboratory-based experimental research
is needed to determine the causal status of cognitive factors (see also D. M. Clark &
McManus, 2002). In the discussion that follows we focus on six hypotheses that are
central to the reformulated cognitive model of social phobia. Although considered a key
factor in social phobia, there is not enough research on involuntary inhibition to include
this concept in our review of the literature.
Hypothesis 1
Social phobia is characterized by an explicit elaborative interpretation and recall bias for
self- referent social threat information that is evident in the anticipatory, exposure, and
postprocessing phases of social anxiety.
If a deliberate and intentional information- processing bias for social threat is evident
in all three phases of social phobia, then we predict that socially phobic individuals will
exhibit a tendency to exaggerate the probability and severity of negative consequences in
socially relevant situations. Various studies have administered mildly positive and nega-
tive self- referent hypothetical social events and found that socially anxious individuals
overestimate the probability and/or the consequences of negative social events compared
to individuals with other anxiety disorders and nonclinical controls (e.g., Foa et al.,
1996; Lucock & Salkovskis, 1988; McManus, Clark, & Hackmann, 2000). Wilson
and Rapee (2005) used a more refined questionnaire of hypothetical social events to
show that the threat judgment bias is specifically related to social phobia and consists
of beliefs that others would perceive you negatively, that the event was an indication of
negative self- characteristics, and that the event would have long-term adverse conse-
quences (see also Wenzel, Finstrom, et al., 2005). This negative judgmental bias is spe-
cific to all social events whether positive or negative (Brendle & Wenzel, 2004; Voncken
et al., 2003) and is significantly reduced by CBT (Foa et al., 1996; Lucock & Salkovskis,
1988; McManus et al., 2000).
Individuals with social anxiety have a significantly greater tendency to choose neg-
ative interpretations to self- referent social but not nonsocial ambiguous hypothetical
events compared to people with other anxiety disorders or nonclinical groups (Amir et
al., 1998b; Constans et al., 1999; Stopa & Clark, 2000). However, more recent studies
suggest that when individuals with social phobia make “online inferences” (i.e., infer-