DEPRESSION AND ANXIETY 26:22–27 (2009)
EXTINCTION RETENTION PREDICTS IMPROVEMENT
IN SOCIAL ANXIETY SYMPTOMS FOLLOWING
Angela C. Berry, M.A., David Rosenfield, Ph.D., and Jasper A. J. Smits, Ph.D?
Background: Several researchers have argued that basic research on extinction
learning can guide efforts to enhance the efficacy of exposure-based therapy. At
the basis of this translational research paradigm is the assumption that
extinction retention is important to the outcome of exposure-based therapy. This
study is the first to examine the relationship between extinction retention, which
comprises the amount of fear reduction that is retained between two exposure
sessions and improvement in anxiety symptoms following exposure treatment.
Methods: Adults (N590), participating in two separate studies, who received
three sessions of repeated exposure to public speaking provided ratings of peak
fear during exposure treatment and completed the Liebowitz Social Anxiety
ScaleSelf—Report version, LSAS-SR, Baker et al. [2002: Behav Res Ther
40:701–715] at baseline, posttreatment, and follow-up. Results: After control-
ling for within-session extinction, extinction retention accounted for significant
variance in the improvement of LSAS-SR scores over time. Conclusions: Our
findings suggest that the consolidation of extinction learning into long-term
memory is associated with improvements in fear and avoidance related to social
situations following exposure therapy. Implications for exposure therapy
augmentation studies are discussed. Depression and Anxiety 26:22–27, 2009.
& 2008 Wiley-Less, Inc.
Key words: exposure therapy; social anxiety; social phobia; fear extinction;
Exposure therapy refers to the extended contact with
the events or stimuli that are presumed to elicit the
anxiety.The particular procedures for exposure
exercises vary considerably; some involve repeated
presentation of the stimulus within and across sessions,
whereas others involve prolonged exposure within a
session. Although research suggests that these various
forms of exposure exercises are associated with fear
reduction,[2,3]not all patients who undergo exposure
therapy experience relief of their symptoms.[4–6]
Accordingly, there is still considerable room for
improvement in exposure-based treatments for anxiety
Emphasizing the current priority of the National
Institute of Mental Health to support translational
research, Anderson and Inseldiscussed the promise
of basic extinction research, particularly that which
examines extinction retention, for improving treatment
outcome for people with anxiety disorders. As others
Anderson and Insel liken the
procedures employed in animal studies of extinction
learning to procedures used in exposure therapy.
Indeed, both extinction training and exposure therapy
comprise repeated exposure to the feared stimulus in
Published online 9 September 2008 in Wiley InterScience
Received for publication 25 February 2008; Accepted 31 March
?Correspondence to: Jasper A. J. Smits, Ph.D., Department of
Psychology, Southern Methodist University, Dedman College,
P.O. Box 750442, Dallas, TX 75275. E-mail: email@example.com
Department of Psychology, Southern Methodist University,
rrrr 2008 Wiley-Liss, Inc.
the absence of the associated negative outcome.[9,10]
Given this parallel, parameters of extinction training in
animals that are associated with greater retention of
extinction learning may well translate to strategies that
can augment the efficacy of exposure therapy.[11,12]
Central to this premise is the assumption that
extinction retention, or the consolidation of extinction
learning into long-term memory, is predictive of
clinical outcome of exposure treatment (e.g., reduc-
tions in fear and avoidance of phobic situations). To our
knowledge, there has been no research examining the
relationship between extinction retention observed
during exposure therapy and clinical improvements
following exposure therapy. The present study provides
the first test of the assumption that extinction retention
predicts exposure therapy outcome.
For this exploratory analysis, we used data from two
samples of individuals who received three treatment
sessions consisting of repeated exposure to public
speaking. As part of the study protocol, participants
provided ratings of peak fear during the exposure
sessions and completed a measure of social anxiety
severity at pretreatment, posttreatment, and follow-up.
Consistent with the definition of extinction retention
employed in animal studies (K.M. Myers, personal
communication, July 9, 2007), we operationalized
extinction retention as the rebound in peak fear
observed from the end of an exposure session to the
beginning of the subsequent session (see Fig. 1). In
addition to extinction retention, we examined the
effects of within-session extinction, conceptualized as
the change in peak fear from the beginning to the end
of each session (see Fig. 1), on anxiety reduction. This
variable was included because extinction retention, by
definition, can only happen if within-session extinction
has occurred. Therefore, it was necessary to examine
the relationship between extinction retention and
improvements in social anxiety severity while control-
ling for within-session extinction.
We used data collected as part of two randomized-controlled studies
examining potential augmentation strategies for exposure treatment of
social anxiety disorder.[13,14]Although these studies differed with
respect to specific aims, they shared a common research paradigm. In
both studies, we randomly assigned participants to either no treatment
(e.g., waitlist or placebo) or to a three-session exposure treatment
protocol comprising repeated exposure to public speaking with or
without augmentation (e.g., videotape feedback of performance,
videotape feedback of audience reactions, ingestion of 200mg of
niacin to induce facial flushing). Additionally, participants receiving
exposure therapy completed peak fear ratings following each speech
exposure and completed the Liebowitz Social Anxiety Scale – Self
Report version (LSAS-SR)immediately before treatment and at two
posttreatment assessments. As reported elsewhere,[13,14]neither the
videotape feedback nor blushing induction procedures impacted the
outcome of the exposure therapy. Thus, we did not consider the effects
of condition assignment in the present analyses. Data analyses included
only participants who received at least two treatment sessions and
completed at least one of the posttreatment assessments. All study
procedures were approved by the institutional review boards of the
respective universities and all participants provided written informed
consent following explanation of study procedures.
University of Texas at Austin and the Austin community, met the
following inclusion criteria: (a) Axis-I diagnosis of social phobia,
determined by the Composite International Diagnostic Interview
(CIDI-auto)(b) significant public speaking anxiety, measured by
an impromptu speech task; (c) fluent in English; (d) negative for
current psychosis, bipolar disorder, or actively suicidal or history of
seizure; and (e) no recent change in psychotropic medications.
Sample characteristics are presented in Table 1.
Participants (N537) were recruited from Southern
Methodist University and met the following entry criteria: (a)
significant public speaking anxiety, measured by an impromptu
speech task; (b) no recent change in psychotropic medications; (c)
elevated level of social cost bias (Z80) on the Social Cost
Questionnaire;(d) negative for history of alcohol use disorders,
allergy to aspirin, history of diabetes, current use of anticoagulants or
blood pressure medications; (e) negative for pregnancy and using an
acceptable means of birth control if sexually active; and (f) no current
psychotherapy. Sample characteristics are presented in Table 1.
Participants (N553), who were recruited from the
Figure 1. Definitions of extinction retention and within-session
23 Research Article: Extinction Retention
Depression and Anxiety
Immediately following eligibility assessment, participants received
a videotaped rationale explaining the procedures of the study and the
efficacy of repeated exposure to public speaking for treatment of
Participants attended three exposure sessions within
a 1-week period. At the beginning of each session, participants
selected a topic on which they were to give a 3-minute speech. They
were allowed 10min to prepare a speech outline and then delivered
the speech five times without use of their outline. All speeches were
video recorded and delivered in front of a small audience. Between
speech repetitions, participants in the video feedback conditions
received video feedback and participants in the no-video feedback
condition watched a video on a neutral topic. Additionally,
participants completed measures of peak fear (05no fear; 1005ex-
treme fear) immediately following each speech.
Procedures matched those of Study 1 with the
following exceptions:(1) participants completed three 5-min speeches
rather than five 3-min speeches during each session; (2) sessions were
completed within a 3-week rather than 1-week period; and (3);
instead of videotape feedback, 18 of 37 participants took 200mg of
niacin before each treatment session.
by subtracting the first peak fear rating of each session from the last
peak fear rating of the previous session (see Fig. 1). These two values
(e.g., Sessions 1–2, and sessions 2–3) were then averaged to yield an
extinction retention index. Complete retention would result in a 0 for
this index, with lesser retention yielding more negative scores.
was calculated by subtracting, for each session separately, the last
peak fear rating from the first peak fear rating. These three values
were then averaged to yield a within-session extinction index. Greater
scores are reflective of greater within-session extinction.
Liebowitz Social Anxiety Scale-Self Report Ver-
The LSAS-SR consists of 24 items measur-
Extinction retention was calculated
ing social anxiety: 13 items describe performance and 11 describe
social interaction situations. Using a 0 (none/never) to 3 (severe/
usually) Likert scale, items are rated on anxiety and avoidance and
summed to obtain a total score. The LSAS-SR shows good internal
consistency (a5.95), retest reliability (r5.83), and sensitivity to
treatment change.Study 1 participants completed the LSAS-SR at
baseline and at 1 and 4 weeks following the last session. Study 2
participants completed the LSAS-SR at baseline, immediately
following and 2-weeks following the last session.
A three-level hierarchical linear modeling approach
(HLM);was employed to examine change in social
anxiety symptoms (LSAS-SR) over time (level 1) within
individuals (level 2). First, we entered time (pretreat-
ment, posttreatment, follow-up) as a level 1 predictor
of LSAS-SR. Differences between individuals with
regard to extinction retention and within-session
extinction (level 2 variables, which were grand mean
centered) were then modeled as determinants of the
level 1 slope of LSAS-SR over time. This analytic
strategy allowed us to examine the effect of extinction
retention on the slope of LSAS-SR over time while
controlling for the effect of within-session extinction.
Finally, individuals were nested within Studies (a level 3
variable that was coded 1 for Study 1 and 0 for Study 2)
to investigate whether the effects of extinction reten-
tion and within-session extinction on the change of
LSAS-SR over time varied between studies, thereby
allowing us to examine whether the hypothesized
relationships replicated across the two data sets.1
Our analytic approach is consistent with recent
recommendations for comparing and aggregating data
TABLE 1. Sample characteristics
VariableStudy 1 (n553) Study 2 (n537)
M SDM SD
Generalized social phobia statusa
aParticipants in Study 2 were not assessed for DSM-IV social phobia.
Following Mennin et al.,
assigned to participants who scored greater than 60 on the LSAS-SR
generalized social phobia status was
1The regression equations for the three-level HLM model were
Level 1: LSAS-SRhij 5b0ij1b1ij*Timehij1ehijwhere h represents the
three assessment times (pre , posttreatment, follow-up)
within individuals, i represents individuals within studies, and j
represents the two studies. Thus, b1ijis the slope of change in LSAS-
SR over time for individual i in Study j. Level 2: b0ij5g00j1Z0ij
b1ij5g10j1g11j* Extinction Retentionij1g12j*Within-Session Extinc-
tionij1Z1ijwhere Extinction Retentionijis the extinction retention
score for subject i in Study j (centered at the grand mean), and
Within-Session Extinctionijis the within-session extinction index for
subject i in Study j (grand mean centered). Thus, g11jis the change in
the slope of LSAS-SR (b1ij) per unit change in extinction retention for
Study j, and g12jis the change in the slope of LSAS-SR (b1ij) per unit
change in within-session extinction for Study j. And g10jis the slope
of LSAS-SR over time for a subject in Study j whose extinction
retention and within-session extinction scores are at the grand mean
for each variable. Level 3: g00j5g0001g001*Studyj1m00jg10j5g1001
g121*Studyj1m12jwhere Studyjis coded 1 for participants in Study 1
and 0 for participants in Study 2. Thus, the significance of g111and
g121indicate if the effect of extinction retention and within-session
extinction, respectively, is different for Studies 1 and 2. And as Study
2 is coded 0, g100is the slope of LSAS-SR for those in Study 2 whose
scores are at the grand mean of extinction retention and within-
session extinction. Similarly, g110and g120are the per unit effect of
extinction retention and within-session extinction, respectively, on
the slope of LSAS-SR for those in Study 2.
24Berry et al.
Depression and Anxiety
from separate experiments.
advantages of this approach. First, HLM calculates
intercepts and slopes separately for each participant,
which can then be modeled as a function of individual
differences on relevant dimensions (e.g., extinction
retention). Second, HLM provides unbiased estimates
of standard errors even when the data are noninde-
pendent (e.g., data from individuals within studies are
related because the studies have slightly different
characteristics). Indeed, failure to allow for this
possibility could lead to underestimation of error terms
and increased Type 1 error rates.[20,21]Third, HLM
can easily handle incomplete data.
There are several
Changes in Study Variables
Participants in Study 2 experienced greater extinc-
tion retention compared to participants in Study 1
(t (85)54.11, Po.001; see Table 2 and Fig. 2). The
degree of within-session extinction was also signifi-
cantly different between the studies, with greater
within-session extinction observed among participants
in Study 1 relative to those in Study 2 (t (86)5?4.24,
Po.001; see Table 2 and Fig. 2).2There was a small,
but nonsignificant difference in the intercepts (pre-
treatment LSAS-SR levels) between the two studies,
with higher scores among Study 1 relative to Study 2
participants (t (89)51.66, P5.099; see Table 3 and
Fig. 2). The average slope of change on the LSAS-SR
over time was significant for participants in both
studies (Study 1: b5?9.52, t (243)5?6.96, P o.005;
Study 2: b5?3.15, t (243)5?2.07, P o.05). This
difference between studies in change of LSAS-SR over
time was significant (Po.01; see Table 3).
Improvement in Social Anxiety Symptoms as a
Function of Extinction Retention
change of LSAS-SR over time (see Table 3); higher
TABLE 2. Means and standard deviations of the study
Study 1 Study 2
LSAS-SR5Liebowitz Social Anxiety Scale—Self Report version.
Figure 2. Extinction retention and within-session extinction by
TABLE 3. Three-level HLM analyses of the effects of
extinction retention, within-session extinction, and study
on change in social anxiety symptoms (LSAS-SR)
EffectCoefficient SEt dfP
Effect of study
Effect of Study
Effect of study
Effect of study
1.52 ?2.07 243
2.05 ?3.11 243
.11 ?3.71 243 o .001
.18 .80 243
.17 ?2.01 243
.22 .18 243
89 o .001
The ‘‘Study’’ variable was coded 1 for Study 1 and 0 for Study 2.
Assessment was coded 0 (pre), 1 (post), and 2 (follow-up). Extinction
retention and within-session extinction were grand mean centered.
This coding scheme creates the following interpretation for each
regression coefficient in the model: The coefficient for each effect is
the relationship for Study 2 as Study 2 was coded 0 (e.g., the effect of
within-session extinction was ?0.34 for Study 2), whereas the effect
of study is the difference between Study 2 and Study 1 on that
coefficient (e.g., the effect of within-session extinction was ?0.341
0.045?0.30 for Study 1).
2We examined the relationship between severity of illness (i.e.,
pretreatment LSAS-SR score) and extinction retention (r5?.030,
P5.78) and within-session extinction (r5.008, P5.94), respectively.
These findings suggest that severity of social anxiety did not affect
relationship between either within-session extinction or extinction
retention and symptom reduction.
25Research Article: Extinction Retention
Depression and Anxiety
extinction retention was associated with steeper (more
negative) slopes of LSAS-SR over time (i.e., more
improvement; b5?.41, t (243)5?3.71, P o.001).
The effect of extinction retention on the change in
LSAS was not significantly different for the two studies
(P5.42). Within-session extinction was also signifi-
cantly associated with changes in LSAS-SR over time
(i.e., greater within-session extinction was related to
greater LSAS improvement; b5?.34, t (243)5?2.01,
Po.05). This association also was not significantly
different between the two samples (P5.86). Thus, both
extinction retention and within-session extinction
predicted improvement on the LSAS, and these
relationships were observed in both studies.
This study fills an important gap in the literature by
demonstrating that adults receiving exposure treatment
evidence greater improvements in anxiety and avoid-
ance of social situations when they consolidate extinc-
tion learning into long-term memory. These results are
persuasive because they were replicated across two
studies. Our findings provide support for the argument
that basic research on extinction learning may inform
the development of augmentation strategies for ex-
posure-based therapies because extinction retention is
important to the outcome of these therapies. In
addition to this finding, we also found that within-
session extinction was predictive of changes in social
One clinical implication of these findings is that
therapists may best help their patients by employing
strategies that not only maximize fear reduction during
a session, but also aid the patient in consolidating this
newly learned information. For example, patients may
derive more benefit from treatment when exposure
exercises are followed by cognitive debriefing, which is
aimed at helping patients solidify session gains. A
potential complimentary strategy for enhancing con-
solidation of extinction memory is the prescription of
cognitive enhancers like D-cycloserine (DCS). Indeed,
results of a series of animal studies suggest that DCS
facilitates extinction consolidation in animalsand
preliminary data from human studies indicate that
acute administration of DCS earlier to exposure
therapy sessions enhances treatment outcome among
patients with anxiety disorders.[23–25]It should be noted
that these studies do not report on extinction retention
per se or the relationship between extinction retention
and treatment outcome. Hopefully, the findings of
study will encourage investigators to examine whether
the improved clinical outcome associated with these
cognitive enhancers is indeed mediated by enhanced
Previous research on processes of exposure treatment
change, considered within the emotional processing
framework,has identified between-session habitua-
tion (i.e. reduction in peak fear from one session to the
next; see Fig. 3), as opposed to within-session habitua-
tion (i.e. decrease in fear during exposure exercises or
‘‘within-session extinction’’), as an important predictor
of clinical improvement.[27–29]Our findings are com-
patible with the results of this previous work, as
between-session habituation is effectively the combina-
tion of fear reduction during treatment and consolida-
tion of this learning into long-term memory. This
combination of two factors, both of which we have
shown to be related to clinical improvement, should be
a better predictor of outcome than either one alone.
Thus, the previous results highlighting the importance
of between-session habituation support our overall
recommendation that maximum improvement will
occur when both extinction retention and within-
session extinction are maximized.
Several factors may limit the generalizability of the
study results to clinical settings. First, our sample
included self-selected individuals who volunteered to
participate in a brief exposure-based treatment. Sec-
ond, the sample was relatively homogenous with
respect to age. Third, because the treatment dose was
fairly low, few participants showed evidence of
clinically significant improvement on the LSAS-SR.
Fourth, our study relied solely on self-report measures.
Future work may include physiological measures of
extinction retention and within-session extinction to
determine whether the observed relationships extend
beyond subjective indices. Despite these limitations,
our study provides initial support for the importance of
enhancing extinction retention during exposure ther-
apy. These findings underscore the utility of basic
research focusing on the identification of parameters of
extinction training that are associated with extinction
1. Todd JT, Pietrowski JL. Animal models of exposure therapy: a
selective review. In: Richard DCS, Lauterbach DL, eds. Hand-
Figure 3. Definition of between-session habituation.
26Berry et al.
Depression and Anxiety
book of Exposure Therapies. Oxford: Academic Press:2006; Download full-text
2. Foa EB, Chambless DL. Habituation of subjective anxiety during
flooding in imagery. Behav Res Ther 1978;16:391–399.
3. Watson JP, Gaind R, Marks IM. Physiological habituation to
continuous phobic stimulation. Behav Res Ther 1972;10:
4. Barlow DH, Gorman JM, Shear MK, Woods SW. Cognitive-
behavioral therapy, imipramine, or their combination for panic
disorder: a randomized controlled trial. J Am Med Assoc 2000;
5. Davidson JRT, Foa EB, Huppert JD et al. Fluoxetine, compre-
hensive cognitive behavioral therapy, and placebo in generalized
social phobia. Arch Gen Psychiatry 2004;61:1005–1013.
6. Foa EB, Liebowitz MR, Kozak MJ, et al. Randomized, placebo
controlled trial of exposure and ritual prevention, clomipramine,
and their combination in the treatment of obsessive-compulsive
disorder. Am J Psychiatry 2005;162:151–161.
7. Anderson KC, Insel TR. The promise of extinction research for
the prevention and treatment of anxiety disorders. Biol Psychia-
8. Bouton ME. Context, ambiguity, and unlearning: sources of
relapse after behavioral extinction. Biol Psychiatry 2002;52:
9. Myers KM, Davis M. Mechanisms of fear reduction. Mol
10. Pavlov IP. Conditioned Reflexes. London: Oxford Press:1927.
11. Ledgerwood L, Richardson R, Cranney J. D-cycloserine and the
facilitation of extinction of conditioned fear: consequences for
reinstatement. Behav Neurosci 2004;118:505–513.
12. Santini E, Hong G, Keqin R, de Ortiz SP, Quirk GJ.
Consolidation of fear extinction requires protein synthesis in
the medial prefrontal cortex. J Neurosci 2004;24:5704–5710.
13. Berry AC. 2006. Facilitating changes in catastrophic thinking:
can niacin challenges reduce social cost bias? Unpublished
master’s thesis, Southern Methodist University.
14. Smits JAJ, Powers MB, Buxkamper R, Telch MJ. The efficacy of
videotape feedback for enhancing the effects of exposure-based
treatment for social anxiety disorder: a controlled investigation.
Behav Res Ther 2006;44:1773–1785.
15. Baker SL, Heinrichs N, Kim HJ, Hofmann SG. The Liebowitz
Social Anxiety Scale as a self-report instrument: a preliminary
psychometric analysis. Behav Res Ther 2002;40:701–715.
16. World Health Organization. Composite International Diagnos-
tic Interview (CIDI) Researcher’s Manual. Washington, DC:
American Psychiatric Press Inc: 1997.
17. Mennin DS, Fresco DM, Heimberg RG, Schneier FR, Davies
SO, Liebowtiz MR. Screening for social anxiety disorder in the
clinical setting: using the Liebowitz Social Anxiety Scale.
J Anxiety Disord 2002;16:661–673.
18. Foa EB, Franklin ME, Perry KJ, Herbert JD. Cognitive biases
in generalized social phobia. J Abnorm Psychol 1996;105:
19. Raudenbush SW, Bryk AS, Cheong YF, Congdon RT. HLM 6:
Hierarchical Linear and Nonlinear Modeling. Lincolnwood, IL:
Scientific Software International; 2004.
20. Hox JJ. Multilevel Analysis: Techniques and Applications.
Mahwah, NJ: Lawrence Erlbaum Associates; 2002.
21. Raudenbush SW, Bryk AS. Hierarchical linear models: applica-
tions and data analysis methods (2nd ed.). Thousand Oaks, CA:
22. Davis M, Ressler K, Rothbaum BO, Richardson, R. Effects of D-
cycloserine on extinction: translation from preclinical to clinical
work. Biol Psychiatry 2006;60:369–375.
23. Hofmann SG, Meuret AE, Smits JAJ, et al. Augmentation of
exposure therapy with D-cycloserine for social anxiety disorder.
Arch Gen Psychiatry 2006;63:298–304.
24. Kushner MG, Kim SW, Donahue C, et al.
augmented exposure therapy for obsessive-compulsive disorder.
Biol Psychiatry 2007;62:835–838.
25. Ressler KJ, Rothbaum BO, Tannenbaum L, et al. Cognitive
enhancers as adjuncts to psychotherapy: use of D-cycloserine in
phobic individuals to facilitate extinction of fear. Arch Gen
26. Foa EB, Kozak MJ. Emotional processing of fear: exposure to
corrective information. Psychol Bull 1986;99:20–35.
27. Foa, EB, Huppert JD, Cahill SP. Emotional processing theory:
an update. In: Rothbaum BO, ed. Pathological Anxiety:
Emotional Processing in Etiology and Treatment. New York:
Guilford Press. 2006;3–24.
28. Jaycox LH, Foa EB, Morral AR. Influence of emotional
engagement and habituation on exposure therapy for PTSD. J
Consult Clin Psychol 1998;66:185–192.
29. Kozak MJ, Foa EB, Steketee G. Process and outcome of exposure
treatment with obsessive-compulsives: Psychophysiological in-
dicators of emotional procession. Behav Ther 1988;19:157–169.
27 Research Article: Extinction Retention
Depression and Anxiety