Empirical Status of the Cognitive Model 89
often found in anxiety, whereas a negative bias in explicit memory is more likely found
in depression. In addition, MacLeod (1999) concluded that anxiety vulnerability is
characterized by implicit but not explicit memory bias for threat.
Presence of an explicit memory bias for threat is indicative of bias at the strate-
gic, elaborative phase of information processing. Contrary to earlier assertions, Coles
and Heimberg (2002) concluded in their review that explicit memory biases for threat-
relevant information is evident in panic disorder and, to a lesser extent, in PTSD and
OCD. However, explicit memory bias is less apparent in social phobia and GAD.
The self- referent encoding task (SRET) has been used most often to assess explicit
memory bias in anxiety and depression. Individuals are shown a list of positive, negative
(or threatening), and neutral self- relevant words and asked to indicate which words are
self- descriptive. After the endorsement task, individuals are given an incidental recall
exercise in which they write down as many words as they can remember. Based on this
experimental paradigm or various modifications, a negative or threat recall bias has
been found for social phobia (Gotlib et al., 2004); panic disorder (Becker, Rinck, &
Margraf, 1994; Cloitre et al., 1994; Lim & Kim, 2005; Nunn, Stevenson, & Whalan,
1984); PTSD (Vrana, Roodman, & Beckham, 1995); and GAD or high trait anxiety
(Mogg & Mathews, 1990). However, other studies have failed to find a negative cued or
free recall (or recognition) bias for GAD or high trait anxiety (Bradley, Mogg, & Wil-
liams, 1995; MacLeod & McLaughlin, 1995; Mathews, Mogg, et al., 1989; Mogg et al.,
1987, 1989; Richards & French, 1991); social phobia (Cloitre, Cancienne, Heimberg,
Holt, & Liebowitz, 1995; Lundh & Öst, 1997; Rapee et al., 1994, Experiments 1 and
2; Rinck & Becker, 2005); OCD (Foa, Amir, Gershuny, et al., 1997); and even panic
disorder (Baños et al., 2001).
Coles and Heimberg (2002) noted that explicit memory bias for threat was more
apparent when conceptual or “deep” processing of information was required at the
encoding stage, when individuals did not have to produce the stimuli they fear at the
retrieval stage, when recall rather than recognition is tested, and when externally valid
experiences are used that relate directly to the fear concerns of the individual. To this
end, some researchers have investigated memory for threatening experiences by expos-
ing individuals to imagined or real-life situations. Most of these studies involved socially
anxious individuals who were exposed to hypothetical or actual social encounters and
then assessed for encoding and retrieval of various elements of the experience. In the
majority of cases the high socially anxious group did not show an explicit threat recall
bias (e.g., Brendle & Wenzel, 2004; Rapee et al., 1994, Experiment 3; Stopa & Clark,
1993; Wenzel, Finstrom, Jordan, & Brendle, 2005; Wenzel & Holt, 2002). Radomsky
and Rachman (1999) found evidence for enhanced recall of prior contact with perceived
contamination objects (see also Radomsky, Rachman, & Hammond, 2001), but this
effect was not replicated in a later study of OCD patients with washing compulsions
(Ceschi, van der Linden, Dunker, Perroud, & Brédart, 2003).
A sufficient number of studies have found evidence of an explicit memory bias for
threat, especially when recall rather than recognition is assessed, to conclude that this
body of research provides a modest level of empirical support for Hypothesis 8. It would
appear that the conscious elaborative processing involved in the encoding and retrieval
of information may be biased toward threat in anxiety. However, an explicit memory
bias for threat has been most apparent in panic disorder and least evident in GAD and