88 COGNITIVE THEORY AND RESEARCH ON ANXIETY
that it can not simply be dismissed as response bias. Interpretative biases have been
demonstrated in panic disorder for body sensation information and in social phobia for
ambiguous social scenarios (see Hirsch & Clark, 2004). In addition, the interpretative
bias training studies provide evidence of a possible causal role in anxiety (see also Chap-
ter 4). Although much remains to be understood about the specificity of the interpreta-
tive bias, we believe the findings are sufficiently well advanced to conclude that it plays
a contributory role in anxiety and so warrants a “strongly supported” designation.
Threat- Related Expectancies
If anxiety is characterized by a threat bias in elaborative processing, then anxious indi-
viduals should be more likely to hold heightened expectations for future threat or danger
that are relevant to their anxious concerns. MacLeod and Byrne (1996) reported that
anxious students anticipated significantly more negative personal future experiences
than nonanxious controls. In a 6-month follow-up of New York City workers after the
9/11 terrorist attacks, individuals who reported more PTSD symptoms also appraised
the threat of future terrorist attacks more likely (Piotrkowiski & Brannen, 2002).
Research on covariation bias indicates that heightened expectations of negative
experiences can bias perceptions of environmental contingencies (MacLeod, 1999). In
this experimental paradigm, individuals are presented fear- relevant or neutral slides that
are randomly associated with a mild shock (aversive response), a tone (neutral response),
or nothing. Participants are asked to pay attention to the stimulus– response associations
and determine whether or not there was a particular relationship between type of stimu-
lus and response. Tomarken, Mineka, and Cook (1989) found that high fearful women
consistently overestimated the percentage of times that the fear slides were associated
with an electric shock, which reflects a processing bias for threat. This overestimation
of threat as indicated by exaggerated judgments of fear stimuli and shock associations
was replicated in spider- phobic individuals (de Jong et al., 1995), although prior fear
may have a greater effect on future covariation expectancies rather than post hoc esti-
mates of past covariation (de Jong & Merckelbach, 2000). Covariation bias for threat
has also been demonstrated in panic-prone individuals exposed to slides of emergency
situations (Pauli, Montoya, & Martz, 1996) and, more recently, in generalized social
phobia when estimating the contingency between negative outcomes and ambiguous
social events (Hermann, Ofer, & Flor, 2004; see Garner, Mogg, & Bradley, 2006, for
contrary results). Although it is unclear whether the covariation bias is as prominent in
the anxiety disorders as it is in specific phobic states, it is evident that negative expectan-
cies can bias judgments of contingencies that characterize anxiety- relevant situations.
Explicit Memory Bias
Information- processing research has also investigated whether anxiety is characterized
by a biased recall of threat- congruent information. If threat- relevant schemas are acti-
vated in anxiety, one would expect increased access to schema- congruent memories.
However, evidence that anxious individuals exhibit a mnemonic advantage for threat-
relevant information has not been compelling (Mathews & MacLeod, 1994; MacLeod,
1999). Williams et al. (1997) concluded that biased implicit memory for threat is more