Panic Disorder 299
PBQ in order to determine if physiological and mental threat schemas play a critical role
in the development of panic disorder.
If beliefs about the threatening nature of internal states are preexisting cognitive
structures, then individuals prone to panic disorder should evidence biased processing of
panic- relevant information even during nonanxious or nonpanic states. In fact there is a
large body of information- processing research that is consistent with activation of physi-
ological or mental threat schemas in panic disorder. Experiments employing the emo-
tional Stroop task have shown that compared to nonclinical control groups, individuals
with panic disorder exhibit a specific color- naming interference for physical threat or
catastrophe words (Hayward et al., 1994; McNally et al., 1994; Teachman, Smith-Janik,
& Saporito, 2007) even at subliminal presentation rates (Lim & Kim, 2005; Lundh et
al., 1999). However, some studies have found that the interference effect in panic is evi-
dent for all threat words in general (Ehlers, Margraf, Davies, & Roth, 1988; Lundh et
al., 1999; McNally, Kaspi, Riemann, & Zeitlin, 1990) or even all emotionally valenced
words (Lim & Kim, 2005; McNally et al., 1992). A few studies have reported no specific
color- naming interference for physical threat words in panic disorder (Kampman, Kei-
jsers, Verbraak, Näring, & Hoogduin, 2002; McNally et al., 1992). Nevertheless, the
general findings from the emotional Stroop experiments are consistent with the presence
of prepotent physiological and mental threat schemas in panic disorder.
Evidence has also been found for an interpretation bias for internal stimuli in panic
disorder. Harvey et al. (1993) found that panic disorder patients chose threat explana-
tions for ambiguous interoceptive scenarios more than social phobics, although there
was no statistical significance between the groups in how often they made threat inter-
pretations. In a covariation bias experiment Wiedemann, Pauli, and Dengler (2001)
found that individuals with panic disorder but not healthy controls overestimated the
association between emergency room pictures (i.e., panic- relevant stimuli) and a negative
consequence (i.e., harmless shock to forearm). However, this finding was not replicated
in a later study (Amrhein, Pauli, Dengler, & Wiedemann, 2005), although electrophysi-
ological evidence for a covariation bias in panic disorder was found. Coles and Heim-
berg (2002) in their review concluded that panic disorder is characterized by an explicit
but not an implicit memory bias for threatening information, especially when deep pro-
cessing is encouraged at the encoding stage. Moreover, the explicit memory bias may
be especially pronounced with physical threat information (Becker et al., 1994; Cloitre
et al., 1994; Pauli, Dengler, & Wiedermann, 2005), although others have failed to find
a specific memory bias (Baños et al., 2001; Lim & Kim, 2005). Finally, Teachman et
al. (2007) found that individuals with panic disorder produced faster response times to
self- evaluative panic- relevant associations on an Implicit Association Test, which reflects
involuntary processing of stimuli congruent with underlying threat schemas.
Overall there is strong empirical support for the schema vulnerability hypothesis
from the information- processing literature. Findings of an automatic threat- processing
bias in nonpanic states are consistent with our contention of a prepotent, enduring sche-
matic threat organization in panic disorder. However, it is still unclear whether the sche-
matic content in panic disorder is highly specific to physiological and mental sensations
or more reflective of general threat themes, and whether activation of these schemas is
responsible for the catastrophic interpretation of bodily sensations. We also await the
development of a more specific panic belief questionnaire that can test the predictive
validity of the schema vulnerability hypothesis in prospective research designs.