PDF Archive

Easily share your PDF documents with your contacts, on the Web and Social Networks.

Share a file Manage my documents Convert Recover PDF Search Help Contact



Memory biases in anxiety disorders .pdf


Original filename: Memory biases in anxiety disorders.pdf
Title: PII: S0272-7358(01)00113-1

This PDF 1.2 document has been generated by 3B2 Total Publishing 6.06b/W / Acrobat Distiller 4.05 for Windows, and has been sent on pdf-archive.com on 16/03/2017 at 21:13, from IP address 205.178.x.x. The current document download page has been viewed 350 times.
File size: 208 KB (41 pages).
Privacy: public file




Download original PDF file









Document preview


Clinical Psychology Review 22 (2002) 587 – 627

Memory biases in the anxiety disorders:
Current status
Meredith E. Coles*, Richard G. Heimberg1
Adult Anxiety Clinic of Temple, Department of Psychology, Temple University, 1701 North 13th Street,
Philadelphia, PA 19122-6085, USA

Abstract
Information-processing models of emotional disorders suggest that anxious individuals may be
characterized by a memory bias for threat-relevant information. This paper reviews and synthesizes
evidence for explicit (conscious) and implicit (unconscious) memory biases in the anxiety disorders.
Our review suggests variations among the anxiety disorders for explicit memory biases. Specifically,
there is support for explicit memory biases for threat-relevant information in panic disorder (PD),
particularly when information has been deeply encoded, but not in social phobia (SP) or generalized
anxiety disorder (GAD). The few available studies suggest the presence of explicit memory biases in
posttraumatic stress disorder (PTSD) and obsessive – compulsive disorder (OCD), but further research
is needed. In contrast, some degree of support for implicit memory biases has been demonstrated for
each of the anxiety disorders. Inconsistencies in the existing literature, topics worthy of future research
attention, and directions for revising existing information-processing models of anxiety are discussed.
D 2002 Elsevier Science Ltd. All rights reserved.
Keywords: Memory; Anxiety disorders

1. Introduction
The examination of memory bias in anxiety disordered individuals is important from both
clinical and basic science perspectives. Intrusive memories are a common symptom of several
anxiety disorders. Individuals with posttraumatic stress disorder (PTSD), for example, often
* Corresponding author.
E-mail addresses: mcoles@astro.ocis.temple.edu (M.E. Coles), rheimber@nimbus.ocis.temple.edu
(R.G. Heimberg).
1
Also corresponding author.
0272-7358/02/$ – see front matter D 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 2 7 2 - 7 3 5 8 ( 0 1 ) 0 0 11 3 - 1

588

M.E. Coles, R.G. Heimberg / Clinical Psychology Review 22 (2002) 587–627

experience unwanted memories through flashbacks, nightmares, and intrusive recollections of
their traumatic event. Individuals with social phobia (SP) can often recount the vivid details
of their self-perceived public humiliations. Individuals with panic disorder (PD) frequently
experience terrifying thoughts of heart attacks, impending insanity, or even death, which may
be fueled by salient memories of their first, or worst, attack. These examples suggest that
anxiety disorders are characterized by heightened accessibility of threatening information. If
so, they should be associated with a memory bias, or preferential memory, for threatening
information. As reasonable as this hypothesis appears, when examined across anxiety
disorders and types of memory, the relevant evidence is mixed.
The anxiety disorders share the primary features of anxiety and avoidance. Despite their
similarities, however, there are many distinctions among them (e.g., the focus of the anxiety,
comorbidity patterns, age of onset), and this heterogeneity may have contributed to
inconsistent findings regarding memory biases. It is also likely that differences in methods
used to study memory processes have contributed to differences in findings. Indeed, the
methods that have been utilized have been as diverse as the populations studied. To date,
there has not been a comprehensive examination of the empirical support for different types
of memory bias in the anxiety disorders. While some reviews have been presented
(e.g., Becker, Roth, Andrich, & Margraf, 1999; Williams, Watts, MacLeod, & Mathews,
1988, 1997), a more comprehensive review will allow for better assessment of current
knowledge and more complete delineation of areas for future exploration.
This paper is comprised of three main sections. First, we provide a general context for
understanding the literature review to follow. This context includes summaries of information-processing models in general and specific to the anxiety disorders. Second, we review
the evidence for explicit and implicit memory biases in each of the anxiety disorders,2
focusing on studies that have examined preferential memory for threat-relevant information
among clinically anxious individuals. Each anxiety disorder is considered separately to best
examine whether different clinical characteristics are related to different memory processes.
Finally, possible reasons for inconsistencies in the literature are addressed and directions for
future research on memory biases in the anxiety disorders are discussed. Directions for
revising existing information-processing models of anxiety are considered as well.

2. Information-processing theories of emotional disorders
Information-processing models of emotional disorders suggest that anxious individuals are
characterized by a memory bias for threat-relevant information. That is, they may be more
likely to recall threat-relevant stimuli, either in comparison to their recall of nonthreat stimuli
or in comparison to the recall of nonanxious persons. Two theories that stimulated much of
the research on information processing in the anxiety disorders are A. T. Beck’s theory of
cognitive schemata and Bower’s theory of associative networks.
2

Due to the extreme heterogeneity of specific phobias and the limited research that has focused on individuals
with clinically severe specific phobias, this class of anxiety disorders is not included in our review.

M.E. Coles, R.G. Heimberg / Clinical Psychology Review 22 (2002) 587–627

589

Beck, Emery, and Greenberg (1985) and Beck, Rush, Shaw, and Emery (1979) proposed
that cognitive functioning is directed by schemata, or cognitive structures, that influence how
one attends to, interprets, and remembers events. If these schemata are distorted, then
information processing will also be distorted (Beck et al. 1985, 1979). Anxiety and
depression are distinguished by the focus of the prominent schemata, with anxiety characterized by schemata concerned with danger and vulnerability and depression characterized by
schemata concerned with loss. Both encoding and recall of mood-congruent information (e.g.,
threat information) should be facilitated when the relevant schema is activated. Bower’s
(1981, 1987) network theory (see Johnson-Laird, Herrmann, & Chaffin, 1984, for a review of
semantic network models) proposes that events are represented in memory as groupings of
associative pathways between various concepts, or nodes, related to the person’s construction
of the event. Further, distinct emotions have specific nodes in memory. One’s mood at
encoding biases the formation of connections, therefore favoring encoding of moodcongruent information. If mood at recall is the same as it was at encoding, recall of moodcongruent information should also be enhanced.
Although these early theories distinguished between encoding and retrieval, they did not
distinguish between explicit and implicit memory (or between strategic and automatic
processes more generally). However, a more recent model proposed by Williams et al.
(1988) has done so, and, it is important to consider these distinctions in various types of
processing before turning to this model. Explicit memory represents conscious, effortful
retrieval of previously learned information and is often examined via tests of free-recall or
recognition. In contrast, implicit memory represents retrieval of information that is learned as
an unintended effect of experience and is tested indirectly (i.e., without specific instructions
for participants to search their memory). For example, after seeing a list of words, participants
may be asked to complete word stems (e.g, tr___) with the first word that comes to mind. In
this example, priming (a form of implicit memory) is demonstrated when participants
complete more stems for words to which they have been previously exposed than stems
for new words. There are two strong lines of support for the distinction between explicit and
implicit memory. First, studies have revealed a number of manipulations that powerfully
affect performance on explicit recall or recognition tasks but that have negligible, or
opposing, effects on implicit memory (see Roediger & McDermott, 1993, for a review).
Second, there is a large body of evidence suggesting that these types of memory are mediated
by different neuroanatomical substrates (medial temporal lobe and explicit memory vs. right
occipital lobe and implicit memory; see Gabrieli, Fleischman, Keane, Reminger, & Morrell,
1995; Schacter, 1998).
Neither Beck’s nor Bower’s models predicted different types of memory bias for the
different emotional disorders. However, Williams et al.’s (1988) review revealed differences
in the nature of cognitive biases in anxiety vs. depression. More studies provided evidence for
explicit memory biases in depressed individuals than in anxious individuals, but there
appeared to be more support for implicit memory biases in anxious individuals. Williams
et al. proposed an integrative model to account for these differences.
Williams et al. (1988) made three main proposals: (1) both encoding and retrieval involve
automatic and strategic components, (2) a bias in one aspect does not entail a bias in the other,

590

M.E. Coles, R.G. Heimberg / Clinical Psychology Review 22 (2002) 587–627

and (3) apparently discrepant results from the existing literature could be explained if
different emotional states had their effects on different aspects of information processing. The
second proposal meant that there could be biases in preattentive priming without biases in
elaboration. Indeed, Williams et al. interpreted the results of existing studies to show that, in
early (automatic) stages of processing, anxious individuals tend to orient their attention
towards the location of threat, thereby increasing priming of this information. However, in
later elaborative (or strategic) stages of processing, anxious individuals tend to orient their
attention away from threat, thereby decreasing the degree to which this information would be
retrievable. Therefore, anxious individuals were proposed to be uniquely characterized by a
pattern of initial vigilance to threat followed by avoidance of further elaboration of this
material, thereby making threatening information more accessible but less retrievable.
The proposal of a vigilance–avoidance pattern of processing in anxiety was clearly
consistent with the work of Mogg, Mathews, and Weinman (1987), who found poorer
memory for threatening than nonthreatening material despite other findings of greater
attentional bias towards threat in anxious individuals. This pattern of increased initial
automatic attention to threat (vigilance) followed by strategic efforts to distance from it
(avoidance) may play a role in the maintenance of anxiety as it might prevent habituation, or
objective evaluation, of threatening material, and allow threat-relevant information to retain
its anxiety-eliciting properties. This conceptualization of anxiety is consistent with models of
emotional disorders that posit a lack of emotional processing as a factor in pathological
anxiety (Foa & Kozak, 1986; Rachman, 1980).
In 1997, Williams et al. suggested some revisions to their thinking based on the intervening
decade of research. Although the main tenets of their model remained intact, they emphasized
the distinctions between memorial and nonmemorial elaboration processes and conceptual vs.
perceptual processing as well as the importance of encoding (e.g., prepriming) for retrieval
(for more detailed discussion of all these issues, see Williams et al., 1997). However, it is
worth noting that Williams et al. did not make any predictions about potential differences in
preferential recall of threatening material according to the nature or type of anxiety in either
their original or revised models.
We now turn to an examination of the evidence for and against memory biases in each
anxiety disorder.

3. Review of the empirical literature by disorder
3.1. Review of studies in Panic Disorder
Panic Disorder (PD) is characterized by recurrent and unexpected periods of intense fear
or discomfort in which the person experiences a number of symptoms (e.g., palpitations,
choking, nausea, fears of dying) that develop abruptly and reach their peak rapidly. Panic
attacks are followed by persistent concern about having additional attacks, worry about the
implications or consequences of the attack, or result in a significant change in behavior.
PD can be accompanied by agoraphobia, in which the person experiences anxiety about

M.E. Coles, R.G. Heimberg / Clinical Psychology Review 22 (2002) 587–627

591

being in situations from which escape might be difficult and avoids these situations
(e.g., malls, buses).
An explicit memory bias towards threat stimuli would be consistent with PD patients’
tendency to report vivid memories of previous threatening experiences and sensations. Such a
memory bias for threatening information may serve to fuel the vicious cycle of panic attacks.
In fact, an explicit bias for threatening memories may interact with catastrophic misinterpretations of bodily sensations that are proposed to be a core feature of PD (Clark, 1986;
Clark et al., 1988). Information-processing models do imply an implicit memory bias for
threat in this group. Memories of past panic symptoms in a particular setting are certainly not
effortfully retrieved but may engender great distress when entering situations in which an
attack has previously occurred. While 15 studies have addressed explicit memory biases in
PD, only 5 have examined implicit memory biases in PD.
3.1.1. Explicit memory bias in Panic Disorder
Support for an explicit memory bias in PD has been relatively strong, with supportive
data reported in 9 (60%) of 15 studies (see Table 1 for a summary of studies included in
this review). In the first relevant comparison (Nunn, Stevenson, & Whalan, 1984),
individuals with PD (and agoraphobia) and nonanxious controls (NACs) read three
passages containing potentially threatening information and two passages containing only
neutral information. Patients subsequently recalled more propositions from the threatening
passages than did NACs. In a second experiment, participants studied lists containing both
threatening (e.g., travel, danger, crowd) and neutral (e.g., letter, family, plant) words.
Patients later recalled more threatening than neutral words, whereas the reverse was true of
control participants.
Explicit memory bias for threat in PD has also been demonstrated with other encoding
tasks, including having participants rate the self-descriptiveness of stimulus words (McNally,
Foa, & Donnell, 1989), perform lexical decision tasks (‘‘Is this a word?’’; Cloitre &
Liebowitz, 1991), and imagine a scene involving themselves and the referent word (Becker,
Rinck, & Margraf, 1994; Becker et al., 1999). Cloitre and Liebowitz (1991) showed that
individuals with PD both recalled and recognized more panic-relevant words than neutral or
positive words. Becker et al. (1999) found that, after a self-referential encoding task, patients
recalled more anxiety symptom words (e.g., palpitations, dizziness) and frightening situation
words (e.g., malls, crowds) than controls, yet performed similarly to controls when recalling
catastrophic cognition words (e.g., helplessness, crazy), positive words (e.g., love, gift), or
neutral words (e.g., chin, dry).
Cloitre, Shear, Cancienne, and Zeitlin (1994) presented individuals with PD, matched
NACs, and clinician controls (therapists who worked with PD patients) with word pairs
(threat, positive, and neutral pairs) and asked them to rate the relatedness of each word pair.
The clinician group was utilized to control for familiarity effects without the related fear. On
an explicit cued-recall task, PD patients recalled more threat words than the clinicians or
NACs. PD patients recalled more threat than neutral or positive word pairs, whereas
clinicians recalled more positive than neutral or threat word pairs, and NACs did not differ
on their recall of the three word types.

592

Anxious
participants’ bias
Authors (year)

Sample

Methodology

PD
Nunn et al. (1984)
(Study 1)

female: 9 PDs and
9 NACs

Nunn et al. (1984)
(Study 2)

female: 9 PDs and
9 NACs

Ehlers et al. (1988)

24 PDs and 24 NACs

Pickles and
van den Broek
(1988) (Study 1)

female: 12 PDs,
13 GAD,
and 12 NACs

followed along reading
5 passages (3 threatening,
2 neutral) as
experimenter read them
aloud, instructed to
imagine that they were
main character, 5-min
filler task followed
by free-recall test
listened to 20 words
(10 threatening and
10 neutral) followed by
free-recall test
recognition test for 72
words seen in a Stroop
task (12 physical threat,
12 separation threat,
12 embarrassment threat,
36 neutral)
heard 4 passages
(2 threatening, 2 neutral),
2-min filler task
followed by
free-recall test

Explicit
memory

Implicit
memory

Y

N/A

PD patients recalled more
propositions from threatening
passages than NACs

Y

N/A

PD patients recalled more
threatening words than neutral
words, reverse for NACs

N

N/A

both groups endorsed more threat
than control words

N

N/A

NACs remembered more
information from passages overall,
groups did not differ based on type
of passage

Results

M.E. Coles, R.G. Heimberg / Clinical Psychology Review 22 (2002) 587–627

Table 1
Investigations of memory bias among patients with anxiety disorders

female: 12 PDs,
13 GAD, and 12 NACs

McNally et al.
(1989)

24 PDs and 24 NACs

Cloitre and
Liebowitz
(1991)

14 PDs and 14 NACs

Beck et al. (1992)

20 PDs and 20 NACs

Becker et al.
(1994)

30 PDs and 20 NACs

listened to 20 words
(10 threatening and
10 neutral) followed
by free-recall test
15 anxiety words and
15 control words, rated
how much apply to self,
high and low arousal,
free-recall test
perceptual processing
encoding task, followed
by free-recall test and
high-speed recognition test
dot-probe task with
physical threat, social
threat, nonthreat, and
positive emotional words,
followed by
recognition test
incidental learning task
(10 panic-relevant,
10 positive, and 10
negative words) and
imagined self-referent
scene, followed
by free-recall test

N

N/A

differences in recall of neutral and
threatening words were
the same for each group

Y

N/A

PDs recalled more anxiety
words than control words,
whereas NACs showed the
opposite pattern

Y

N/A

N

N/A

PDs displayed better perceptual
and semantic memory for panic
words than positive or neutral
words as compared to NACs
both groups showed better
recognition memory for social
threat stimuli than other
word types

Y

N/A

PDs recalled more panic-related
words than NACs, but the
groups did not differ on their
recall of positive or
neutral words

(continued on next page)

M.E. Coles, R.G. Heimberg / Clinical Psychology Review 22 (2002) 587–627

Pickles and
van den Broek
(1988) (Study 2)

593

594

Table 1 (continued )
Anxious
participants’ bias
Sample

Methodology

Cloitre et al.
(1994)

24 PDs, 24 clinician
controls, and 24
NACs

presented with word pairs
(threat, positive, and
neutral pairs) and
participants rated
relatedness of words;
after filler task,
presented with word
stem completion task in
which they were given
one word and a
cued-recall task

Otto et al. (1994)

12 PDs, 12 GADs,
and 12 NACs

presented with panic
words, general threat
words, positive words,
and neutral words and
were then asked to rate
the personal emotional

Explicit
memory

Implicit
memory

Y

Y

N

N/A

Results
cued-recall—PDs recalled more
threat words than clinician
controls or NACs, PDs recalled
more threat than neutral or
positive word pairs, clinicians
recalled more positive than
neutral or threat word pairs,
and NACs did not differ on
their recall for the three
word types
word stem completion—PDs
completed more threat word
stems than clinician
controls or NACs, PDs
completed more threat
words stems than positive
word stems, clinicians did
not show differences in
completions by word type,
and NACs completed more
neutral than positive or
threat word stem
All groups recalled panic-threat,
general-threat and positive
words better than neutral words.
Patients with greater right-ear
(left hemisphere) advantage
showed a greater bias

M.E. Coles, R.G. Heimberg / Clinical Psychology Review 22 (2002) 587–627

Authors (year)

significance of each;
after a 5-min filler
task, participants completed a cued-recall test

Amir, McNally,
Riemann, &
Clements (1996)

19 PDs and 20 NACs

14 PDs and 14 NACs

social threat, physical
threat, and neutral words
followed by free-recall
and stem completion
tests in a counterbalanced order

white noise judgment
paradigm—rated volume
of noise accompanying
panic-relevant and
neutral sentences

N

N/A

N

N across noise
levels, Y in
low-noise
condition

stem completion—both groups
correctly completed more
stems of neutral words than
either physical threat words or
social threat words
across noise levels PDs did not
show a priming bias based on
sentence type but NACs
showed a priming bias for
neutral sentences, when
examining the low-noise
condition alone, PDs rated
the noise accompanying
old panic-relevant sentences as
less loud than the noise
accompanying neutral sentences,
but NACs did not
(continued on next page)

M.E. Coles, R.G. Heimberg / Clinical Psychology Review 22 (2002) 587–627

Rapee (1994)

toward recall of general threat
and panic threat words relative
to neutral words. Those with a
lower right-ear advantage
showed apparent cognitive
avoidance of threat material.
There was no relationship
between perceptual asymmetry
and memory bias in NACs.
free-recall—both groups correctly
recalled more neutral words than
physical threat words, and more
physicals words than social
threat words

595


Related documents


memory biases in anxiety disorders
integrated cognitive model
veale2014
art 1
like a phoenix from the ashes
eutapticsisthebestanxietyremedytohelpordinarypeople


Related keywords