410 TREATMENT OF SPECIFIC ANXIETY DISORDERS
difficulty disengaging from threat rather than by hypervigilance for threat, at least in
nonanxious individuals (Koster, Crombez, Verschuere, & De Houwer, 2004). Recently,
inviduals with GAD trained to attend to neutral rather than threat words showed a sig-
nificant decrease in anxious symptoms (Amir et al., 2009). It should be noted that like
the emotional Stroop task, the dot probe bias in GAD is not specific to threat stimuli but
to negative information more generally (Mogg, Bradley, & Williams, 1995).
In order to determine whether individuals with GAD have a tendency to impose
threatening interpretations in ambiguous situations, researchers have used a variety of
ambiguous stimuli. Studies employing homophones (for further discussion, see Chap-
ter 3, Hypothesis 8) have found that patients with GAD and high trait- anxious indi-
viduals produced significantly more threatening spellings than nonanxious individuals
(Mathews, Richards, & Eysenck, 1989; Mogg et al., 1994). Likewise a threat interpre-
tation bias has been detected when individuals with GAD are presented with ambigu-
ous sentences (Eysenck et al., 1991) or when speed of comprehending ambiguous sen-
tences is measured in high trait- anxious individuals (MacLeod & Cohen, 1993). There
was evidence that individuals with GAD exhibited a negative interpretation bias (i.e.,
facilitation effect) in an emotional priming task in which sentences depicting positive or
negative life events preceded positive and negative self- referent trait adjectives (Dalgleish
et al., 1995). Furthermore, Ken, Paller, and Zinbarg (2008) found that only high-trait
anxious individuals’ word-stem completion performance for threat words was affected
by a unconscious threat prime, again a finding consistent with presence of an automatic
hypervigilance for threat and subsequent facilitative interpretation of threat stimuli.
Although, negative results have also been reported in other studies (e.g., Hazlett- Stevens
& Borkovec, 2004), recent evidence suggests training in generating benign interpreta-
tion to threat can reduce anxious reactivity to a stressor (Hirsch et al., 2009).
In summary there is fairly strong and consistent evidence that GAD and its precur-
sor, high trait anxiety, are characterized by an automatic attentional bias for threat, as
predicted by the third hypothesis. Empirical evidence for a threat interpretation bias for
ambiguity is also moderately strong, especially in light of recent reports of the causal
effects of interpretation threat bias training (see discussion in Chapter 4, Hypothesis
12; also see MacLeod & Rutherford, 2004). However, it would appear that the process-
ing bias in GAD is not specific to threat but is sensitive to negative emotional stimuli
in general. Also the bias is not apparent in individuals who have recovered from GAD
and may be influenced by stressors that elevate state anxiety. It is also not clear whether
hypervigilance for threat or difficulty disengaging from threat is the primary feature of
the attentional bias. Finally, although the current model, like most cognitive theories
of GAD, argues that presence of threat encoding and interpretation biases are critical
processes that characterize worry, little is still known about the information- processing
biases that underlie worry per se. One development that might help in this regard would
be to use experimental stimuli that more closely resemble the idiosyncratic life concerns
of GAD worriers.
Clinician Guideline 10.11
Cognitive interventions must address the anxious person’s automatic tendency to assume a
more negative and threatening interpretation of ambiguous and uncertain life situations.