Empirical Status of the Cognitive Model 73
Biased Threat Estimations
One of the most consistent findings in cognitive research on anxiety is that anxious indi-
viduals tend to overestimate the probability that they will encounter situations that pro-
voke their specific anxiety state. In an early study Butler and Mathews (1983) gave clini-
cally anxious individuals, depressed individuals, and nonclinical controls 10 ambiguous
situations. The anxious group generated significantly more threatening interpretations
and rated these negative threatening events as significantly more probable and severe
(i.e., subjective cost) than nonclinical controls but not the depressed group. This finding
was later replicated with high trait- anxious students (Butler & Mathews, 1987). Biased
estimates of threat probability have been found in subsequent research in which social
phobics overestimate the probability of experiencing negative social events (Foa, Frank-
lin, Perry, & Herbert, 1996; Lucock & Salkovskis, 1988), claustrophobics exaggerate
the likelihood they will encounter closed spaces (Öst & Csatlos, 2000), individuals with
panic disorder interpret arousal- related scenarios and negative physical outcomes more
probable and costly (McNally & Foa, 1987; Uren, Szabó, & Lovibond, 2004), and wor-
riers generate higher subjective probabilities for future negative events (e.g., MacLeod,
Williams, & Bekerian, 1991). In this latter study increased access to reasons why the
negative event would happen and reduced access to why it would not happen (i.e., safety
features) predicted probability judgments.
Cognitive bias should be most evident during fear activation. The positive corre-
lation between heightened probability or severity (i.e., costly) estimates of threat and
intensity of anxious symptoms is consistent with this prediction (e.g., Foa et al., 1996;
Lucock & Salkovskis, 1988; Muris & van der Heiden, 2006; Öst & Csatlos, 2000;
Woods, Frost, & Steketee, 2002). Moreover, causal relations between anxiety and threat
perception have been found in fear provocation experiments. In various studies anxious
and phobic individuals predict that they will experience more fear and panic attacks
than actually happens when exposed to the fear situation (e.g., Rachman, Levitt, &
Lopatka, 1988b; Rachman & Lopatka, 1986; Rachman, Lopatka, & Levitt, 1988).
This tendency to overestimate the likelihood of threat has also been found in the worry
concerns of chronic worriers (Vasey & Borkovec, 1992) and in the exaggerated negative
appraisals of social performance generated by socially anxious individuals (Mellings
& Alden, 2000; Stopa & Clark, 1993). However, with repeated experience, individu-
als show a decrease in their overpredictions of fear so that their estimates more closely
match their actual fear level.
Maladaptive Looming Effect
Along with exaggerated estimates of threat probability and severity, inaccurate apprais-
als of the proximity of danger are also an aspect of biased cognitive processing in anxi-
ety. Riskind and Williams (2006) emphasize that “mental representations of dynamically
intensifying danger and rapidly rising risk” (pp. 178–179), termed looming maladaptive
style, are a key component of threat appraisal in anxiety. According to Riskind and col-
leagues, a critical feature of any threatening stimulus is the perception of the threat as
moving and intensifying in relation to the self in terms of physical or temporal proxim-
ity of real events but also in terms of the mental rehearsal of the potential time course
of future events (Riskind, 1997; Riskind, Williams, Gessner, Chrosniak, & Cortina,