Professional Documents
Culture Documents
Updating Watson & Marks (1971): How Has Our Understanding of the
Mechanisms of Extinction Learning Evolved And Where Is Our Field Going
Next?
PII: S0005-7894(16)00019-8
DOI: doi: 10.1016/j.beth.2016.02.003
Reference: BETH 611
Please cite this article as: Asnaani, A., McLean, C.P. & Foa, E.B., Updating Wat-
son & Marks (1971): How Has Our Understanding of the Mechanisms of Extinction
Learning Evolved And Where Is Our Field Going Next?, Behavior Therapy (2016), doi:
10.1016/j.beth.2016.02.003
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
ACCEPTED MANUSCRIPT
PT
RI
Updating Watson & Marks (1971): How Has Our Understanding of the Mechanisms of
SC
Extinction Learning Evolved And Where Is Our Field Going Next?
NU
Anu Asnaani
MA
Carmen P. McLean
Edna B. Foa
D
TE
Abstract
J.P. Watson and I.M. Marks published a seminal article in Behavior Therapy entitled
“Relevant and Irrelevant Fear in Flooding – A Crossover Study of Phobic Patients” in 1971 that
PT
paved the way for important theoretical developments and empirical studies that examined the
RI
mechanisms underlying extinction learning. Indeed, in the 44 years since their article was
published, our knowledge about how exposure therapy works has increased considerably. In this
SC
review, we explore the progress our field has made in understanding extinction learning and how
NU
Watson and Marks’ important work has influenced this progress. We provide a brief summary of
the design and major findings of the Watson and Marks (1971) study, followed by a brief
MA
description of several theoretical conceptualizations of fear extinction that were developed
following the article’s publication. We also review empirical studies that illustrate the “state of
D
the science” with regard to the following key issues that were explored in Watson and Marks’
TE
paper: (1) the effect of specificity of exposure stimuli content in exposure therapy on outcome;
P
(2) fear activation as a mechanism of exposure; and (3) the associations between within- and
CE
between-session extinction learning and treatment outcome. The major findings of these three
AC
Keywords: exposure therapy; mechanisms; fear reduction; extinction learning; anxiety disorders
ACCEPTED MANUSCRIPT
Updating Watson & Marks (1971): How Has Our Understanding of the Mechanisms of
The paper published by J.P. Watson and I.M. Marks in Behavior Therapy entitled
PT
“Relevant and Irrelevant Fear in Flooding – A Crossover Study of Phobic Patients” (Watson &
RI
Marks, 1971) 44 years ago raised several fundamental questions about the efficacy and
SC
mechanisms of exposure techniques. Their important work set off a cascade of research that has
helped improve exposure therapy approaches to treating anxiety-related disorders, and has
NU
increased our understanding about the mechanisms underlying fear extinction. This article
MA
explores how our field has progressed since this seminal paper published in 1971. After a brief
summary of the design and major findings of this study, we describe several predominant
D
theoretical conceptualizations of fear extinction both at the time the article was written as well as
TE
subsequent theoretical developments. Next, we discuss empirical studies that illustrate the “state
of the science” with regard to the following key issues that were explored in Watson and Marks’
P
CE
paper: (1) the effect of specificity of content in exposure therapy on outcome; (2) fear activation
as a mechanism of exposure; and (3) the associations between within- and between-session
AC
The principal aim of Watson and Marks’ study was to examine the impact of flooding
(i.e., prolonged exposure, typically imaginal exposure) using cues that are relevant to the
patients’ fears during flooding versus flooding using cues that are irrelevant to patients’ fears.
Participants in the study were patients with specific phobias (N = 6; including “spiders, birds,
thunderstorms, wearing clean shirts and public violin playing”) and agoraphobia (N = 10). Half
of the participants first received 8 sessions of fear-relevant imaginal flooding, which entailed
ACCEPTED MANUSCRIPT
listening to the therapist repeatedly describe the patients’ feared situation or object for 50
minutes including the feared consequences of facing the feared situation, followed by 8 sessions
of fear-irrelevant imaginal flooding, which entailed situations that the authors defined as
PT
“normally frightening to anybody” including scenarios where the participant was eaten by a tiger
RI
(12 participants), burned to death (three participants), or drowned (1 participant). The other half
of the participants first received 8 sessions of fear-irrelevant imaginal flooding sessions followed
SC
by 8 sessions of fear-relevant imaginal sessions. Sessions were scheduled 2-3 times per week
NU
with a 4-8 day break between the two 8-session conditions, for an average treatment time of six
weeks. Half the participants (four in each order condition) were treated as inpatients due to the
MA
severity of their fears, and the remaining patients were treated as outpatients. Assessments of
participants’ symptoms (both with clinical and physiological measures) occurred just prior to the
D
TE
start of treatment, 2-4 days after sessions 8 and 16, and at 3-month and 6-month follow-up
during an imagery task whereby they were asked to imagine three different scenarios for 40
CE
seconds twice each, in counterbalanced order: (1) a neutral scene, (2) a control fear situation, and
AC
The study results indicated that although 8 sessions of either treatment alone led to
significantly improved clinical outcomes, the combination of the two treatments (i.e., 16
sessions) resulted in the most significant improvements in clinical and heart-rate measures
regardless of the order in which the treatment conditions were delivered. Interestingly, the fear-
irrelevant condition, but not the fear-relevant condition, produced significant improvement in
heart-rate and skin conductance during assessment. The authors also found that the fear-
irrelevant condition was significantly more effective than the fear-relevant condition in reducing
ACCEPTED MANUSCRIPT
subjective anxiety when individuals were imagining their feared phobic-specific situation during
the imagery assessment task. The gains from pre- to post-treatment were maintained at 6 months
PT
Other results reported by Watson and Marks (1971) suggested that differential
RI
mechanisms underlie success in each of the treatment conditions. Specifically, heightened
physiological activity during the imagery task before treatment predicted good outcome to the
SC
fear-relevant flooding, but not to the fear-irrelevant flooding. In addition, high levels of pre-
NU
treatment subjective anxiety during the imagery assessment predicted poor outcome to fear-
irrelevant flooding. Finally, high subjective anxiety reported during treatment sessions predicted
MA
good outcome to fear-irrelevant flooding, but was not associated with outcome in fear-relevant
flooding. The authors concluded that while flooding to both relevant and fear-irrelevant stimuli
D
TE
reduced phobic anxiety in participants, this reduction was achieved via different mechanisms
across conditions. They further hypothesized that these differing mechanisms may not be
P
mutually exclusive, and might also be additive in their effects on fear reduction.
CE
The study has several limitations. Of most concern is the nature of the “fear irrelevant”
AC
imaginal scenarios. Specifically, it is not clear whether these scenarios were actually irrelevant to
all participants’ fears. For instance, the drowning scenario may not tap into the core pathological
fear of someone with a fear of spiders, but for someone with agoraphobia who fears the physical
symptoms of anxiety (which would include an inability to breathe, which is a central feature of
drowning), such a scenario may be quite relevant to a core agoraphobic concern. These
confounding elements of the study manipulation make it difficult to draw conclusions about the
role of content specificity in exposure. Second, the authors used primarily two-tailed t tests to
test their hypotheses; while this is appropriate given the sample size and is consistent with the
ACCEPTED MANUSCRIPT
statistical approaches employed at the time the study was published, it limits the inferences and
types of questions (e.g., hypotheses about underlying mechanisms) that can be answered with
such a study design. Related to this choice of analytic approach, the sample of the different
PT
experimental groups (8 individuals in each, which were themselves subdivided by primary
RI
diagnosis of agoraphobia or specific phobia) was very small, and thus the findings need to be
interpreted with caution. Last, while the major focus of this study is on the immediate results
SC
after 8 and 16 sessions, the authors reported that the gains were maintained through 6-month
NU
follow-up, and yet they also noted that the majority of participants (14 of 16) received further
treatment during the follow-up period, including additional imaginal exposure, in vivo exposure
MA
(that is, situational exposure to avoided things, places or activities associated with the feared
stimulus), and antidepressant medications. The robustness or clinical relevance of this reported
D
TE
long-term maintenance of their treatment effects are therefore unclear given the high proportion
Theoretical underpinnings
Prior to and at the time of this study, the predominant conceptualization of how fear
AC
extinction occurred revolved around the assumption that repeated confrontation with a feared
stimulus in the absence of a feared consequence would lead to a reduction of fear, or habituation.
This assumption was adopted from the impactful work of previous learning theorists (e.g., Ivan
Pavlov and B.F. Skinner, among others) who had developed conditioning models to explain the
acquisition and extinction of fear (Pavlov, 1927; Skinner, 1938). Of note, Watson and Marks
used the terms “extinction” and “habituation” interchangeably, although they reported a
preference for the term habituation because it does not presuppose that the pathological fear
developed through conditioning. In addition, there was very little focus on the mechanisms
ACCEPTED MANUSCRIPT
underlying conditioning and extinction in these early conditioning theories. Watson and Marks’
study tried to elucidate mechanisms, and much of the theoretical and empirical work that
followed their study has sought to do the same. Thus, several theoretical models were proposed
PT
and subsequently adopted as the framework for empirical work in fear and anxiety in the decades
RI
following the 1971 study; we briefly described a few of these major theoretical models below.
SC
Wagner shortly after the Watson and Marks paper was published delineates a primarily
NU
behavioral model by which fear is maintained and extinguished (Rescorla & Wagner, 1972).
Rescorla and Wagner’s theory relies heavily on the early concepts of the unconditioned stimulus
MA
(US; a biologically potent stimulus that naturally brings on an affective response), conditioned
reflexive response) and conditioned response (CR; a taught response due to repeated US-CS
pairing), all of which are concepts adopted from previous classical and operant conditioning
P
paradigms.
CE
However, Rescorla and Wagner’s (1972) theory uniquely proposes two features not
AC
present in previous conditioning models: (1) organisms learn from the discrepancy between what
they expected to happen and what actually happens and the degree of expectancy violation when
encountering a US is dependent on the associative value of all cues present during this
confrontation with the US; and (2) excitation is the opposite of inhibition and a stimulus can be
one or the other, but not both. Rescorla and Wagner also incorporated principles already
prominent in the prevailing fear extinction theories during the time it was introduced, namely:
the perceived strength of a feared stimulus is directly measurable in the behavior it overtly elicits
or inhibits in an organism; the salience of the CS remains constant throughout old and new
ACCEPTED MANUSCRIPT
learning; and only the current associative value of a cue to a feared stimulus is relevant to the
amount of learning that occurs (regardless of the prior historical associative value of that cue).
Related to Watson and Marks’ study, Rescorla and Wagner’s model (along with previous
PT
conditioning models) would predict the reduction in fear over time following repeated
RI
confrontation with fear imagery, as the discrepancy between what is expected (feared outcome)
and what actually happens (no harm) continues with each repeated exposure. However, since
SC
Rescorla and Wagner’s theory was introduced after 1971, it is likely that the theoretical basis for
NU
the Watson and Marks study revolved around older, classical conditioning theories that would
predict habituation (i.e., fear reduction) with repeated exposure to the CS in absence to the UCS
MA
without presupposing a mechanism underlying fear reduction.
Wagner, 1972) was introduced and hypotheses emanating from the theory gained empirical
support, Lang (1979) proposed what he termed the bioinformational theory of emotional
P
imagery. Lang’s theory conceptualizes images in the brain as part of a conceptual fear network
CE
that includes both physiological and behavioral activity, patterns, and expression. The experience
AC
of fear would therefore be defined as a structure that is represented by these various parts of the
network. Importantly, the bioinformational theory suggests that physiological and behavioral
response following confrontation with imagery of a fearful stimulus depends on the type and
content of the imaginal exposure. Specifically, the responses depend upon whether there is a
match between the content of the exposure and the underlying fear structure. This point is
relevant to the main hypothesis in the Watson and Marks (1971) study, which was interested in
whether matching exposure content to patients’ fears would impact outcomes. However, Watson
and Marks’ findings were inconsistent with Lang’s hypothesis, because their results suggested
ACCEPTED MANUSCRIPT
that there was no need for matching the flooding imagery to the underlying phobic fear content
Lang also hypothesized that the imagery of the fear response to a particular fear stimulus,
PT
rather than the fear stimulus itself, is what activates the fear network. He further suggested that
RI
the structure of an emotional image and its consequent effects on fear arousal and behavior could
SC
responses to such fears. Because fear is activated by fear response representations and not by
NU
stimuli representations, Lang conceptualized the therapeutic process as emotional processing
because of its reliance on modifying the affective responses to fear imagery (Lang, 1977).
MA
Emotional processing theory. Another major theoretical development since the Watson
and Marks (1971) study was emotional processing theory (EPT; Foa & Kozak, 1986). Expanding
D
on Lang’s bio-informational model (Lang, 1977; 1979), EPT describes the representation of fear
TE
as a cognitive structure embedded in memory that includes information about fear stimuli, fear
P
responses, and their meaning. However, in contrast to the bioinformational theory’s emphasis on
CE
representations of fear response, EPT focuses on the meaning of the feared stimuli and responses
AC
represented within the structure. In addition, EPT suggests that the feared stimuli, responses, and
their meaning are interconnected, such that an input matching one part of the structure will
According to EPT, two conditions are necessary for emotional processing, the presumed
mechanism of recovery, to occur. First, the structure must be activated, so it is available for
change. Second, new information that is incompatible with the pathological fear structure must
be available and incorporated into the pathological memory structure. Inconsistent with the
findings of Watson and Marks, EPT suggests that there are specific pathological fear structures
ACCEPTED MANUSCRIPT
underlying different disorders. It would follow, therefore, that in order to be effective, exposure
must provide information that is inconsistent with the pathological elements of that specific
structure. The fact that Watson and Marks found that exposure to irrelevant-fear stimuli was as
PT
effective in reducing fear as relevant-fear would not be predicted by EPT.
RI
Since its inception, EPT has undergone two revisions to incorporate emerging empirical
evidence (Foa & McNally, 1996; Foa, Huppert, & Cahill, 2006). Influenced by modern learning
SC
theories that conceptualize extinction as the creation of new associations as opposed to the
NU
modification of old associations (e.g., Bouton & Swartzentruber, 1991), Foa and McNally (1996)
proposed that exposure therapy results in the formation of a new, non-pathological structure that
MA
competes with the original pathological fear structure. The old pathological structure and the
new, non-pathological structure contain overlapping elements, such that the same stimuli and
D
TE
responses are capable of activating both structures. Thus, exposures therapy is successful when
the new non-pathological structure is more readily activated, whereas relapse may occur when
P
the old pathological structure is activated. Foa, Huppert, & Cahill (2006) outlined a more
CE
detailed discussion on how EPT applies to the development and maintenance of PTSD and to
AC
social anxiety disorder, highlighting how EPT can guide exposure therapy and cognitive therapy
One of the major conclusions of the Watson and Marks (1971) article was that the
specificity of content for exposure therapy may not be as important as simply inciting non-
specific fear arousal in participants. As noted earlier, theoretical perspectives such as the
bioinformational theory and EPT are inconsistent with such findings, since both theories
emphasize the importance of activating the fear structure relevant to the major phobic concern
ACCEPTED MANUSCRIPT
and some level of matching in confronted fearful material with the fear structure (Lang, 1977;
Foa & Kozak, 1986). The empirical data that has emerged since this article was published is
discussed next, to assess the degree to which Watson and Marks’ finding that content specificity
PT
is not critical for good exposure therapy outcomes has been subsequently supported or refuted.
RI
Empirical data. Around the time the bioinformational and emotional processing theories
were being developed, several studies examining the necessity of confronting fear-relevant
SC
material during extinction learning were conducted. Specifically, in a follow-up to the Watson
NU
and Marks (1971) study, a two-part animal study examining fear acquisition and extinction
learning in rats was conducted by Baum and LeClerc (1974) that used a similar design to the
MA
human study by Watson and Marks. In the first experiment, extinction of avoidance behaviors
was observed during a fear-relevant stress condition with response prevention and during a fear-
D
TE
irrelevant stress condition, corroborating the notion that a match between the content of the
exposure stimuli and previously conditioned feared stimulus may not be necessary for good
P
outcomes. In this first experiment, extinction testing occurred immediately following the
CE
treatment conditions. However, in their second experiment, the authors inserted a 2-hour delay
AC
between the extinction conditions (relevant stress versus irrelevant stress exposure) and
extinction testing. This second experiment found evidence of extinction learning only in rats
exposed to relevant stress flooding, and not in the irrelevant stress condition. Such findings
suggest that the effects of extinction learning via exposure to fear-irrelevant stimuli might be
In another study, Mineka (1976) examined fear extinction in rats to better understand the
role of fear-irrelevant flooding in extinction learning. In the first experiment, both relevant and
degree, which is consistent with Marks and Watson (1971) and Baum and LeClerc (1974). The
second experiment showed that extinction to one conditioned fear response can facilitate
PT
learning and perhaps explaining the mechanism by which fear-irrelevant flooding was found to
RI
be effective in previous studies. However, inconsistent with the findings by Watson and Marks
(1971), the third experiment failed to find an effect of fear-irrelevant flooding using two stimuli
SC
significantly different from each other and not used in the previous experiments. These findings
NU
therefore support the hypothesis, subsequently supported in theoretical models of fear extinction,
that the match between the exposure stimulus and the individual’s feared stimulus is necessary in
MA
order to facilitate extinction learning to the target fear (Mineka, 1976).
Following these animal studies, an experimental study with humans also examined the
D
TE
role of exposure content specificity (Lang, Levin, Miller, & Kozak, 1983). In this study, non-
clinical participants reporting either high snake fear or social anxiety were exposed to both their
P
primary phobic fear, and the primary phobic fear of the other group (i.e., they completed either a
CE
snake exposure or a public speaking exposure). In addition, participants from both groups were
AC
asked to imagine both types of feared situations as well as control, non-fearful scenes. The
results were inconsistent and only partially in line with Watson and Marks (1971). Specifically,
response to snake exposure than the social anxious participants; however, both groups showed
similar and significant increases in physiological arousal during the speech exposure, even
though the socially anxious participants subjectively reported greater fear in this situation. In
addition, imagery assessments to both feared situations did not result in significant physiological
arousal in either group. However, given its design, this study could not provide insight into how
ACCEPTED MANUSCRIPT
Another study that did include outcome data examined individuals with obsessive
PT
compulsive disorder (OCD) who had both checking rituals and significant fears of future
RI
disasters, assigning half the sample to two hours of in vivo exposure (i.e., real life exposure), and
the other half to 90 minutes of imaginal exposure to feared future disasters plus 30 minutes of in
SC
vivo exposure (Foa, Steketee, Turner, & Fischer, 1980). Assuming that specificity of exposure
NU
content was important, the authors hypothesized that those receiving imaginal exposure would
fare better than those receiving in vivo exposure only since imaginal exposure allowed for a
MA
better match to the participants’ core fears. The results revealed that both groups improved
significantly and to a similar degree. However, rate of relapse was significantly lower in the
D
TE
group receiving imaginal exposure plus in vivo exposure, which provides some support for the
hypothesis that the greater the match between the individual’s fear, the better the outcome.
P
Following this initial flurry of studies that ensued in animal and human populations in the
CE
decade following the published findings by Watson and Marks (1971), little empirical
AC
investigation into the importance of the relevance of exposure content has been conducted. One
exception is a recent small study of individuals with OCD that found that acquisition of emotion
associated with decreases in OCD symptoms, and subsequent implementation of these skills in
clinically-relevant contexts were also associated with decreases in OCD. However, again, there
were significantly greater reductions in OCD symptoms when these skills were implemented in
2009). On the whole, there is evidence for the matching hypothesis, although the degree to which
ACCEPTED MANUSCRIPT
the content of exposure is similar to, or matches, the feared structure may be important. This
factor needs more empirical examination to determine its utility in exposure therapy.
PT
The study by Watson and Marks (1971) implied that, regardless of the relevance of
RI
feared exposure material, activation of fear (as measured clinically, subjectively, and
SC
these various fear indices. This idea has received support over the years, and there have been
NU
significant conceptual discussions about why such activation may serve as a crucial component
of later fear reduction and improved clinical outcomes. As mentioned previously, according to
MA
EPT, fear activation is conceptualized as one of the necessary conditions for successful recovery
from pathological fear, the other being incorporation of information that is incompatible with the
D
TE
pathological components of the fear structure. Other behavioral or learning theories related to
fear extinction have been less focused on the role of fear activation in impacting successful
P
treatment outcome.
CE
Typically, fear activation has been defined in both subjective and objective
AC
(physiological) terms. For subjective fear activation, this has been operationalized as peak (or
levels of subjective distress. Physiological fear activation (which often includes heart rate, skin
conductance, eye blink startle magnitude, and electromyogram) is measured similarly, by using
the difference between peak recordings on these measures and baseline physiological reactivity.
Empirical data. In line with subsequently developed theoretical models suggesting that
(specifically, EPT), Watson and Marks (1971) found that increased physiological responsiveness
ACCEPTED MANUSCRIPT
prior to the flooding of fear-relevant cues only was associated with better treatment outcome.
Similarly, other studies conducted around that time period found that increased heart rate during
initial exposure to the fearful stimulus in phobic individuals was associated with greater
PT
improvement in symptoms following systematic desensitization (Lang, Melamed, & Hart, 1970;
RI
Borkovec & Sides, 1979). Since then, many studies have examined the relationship between
increased physiological and subjective arousal during early confrontation to feared stimuli and
SC
treatment outcomes in anxiety-related disorders, some of which support the supposition that fear
NU
activation is a necessary, although not sufficient, condition for emotional processing to occur.
For example, Kozak, Foa & Steketee (1988) conducted a study with OCD patients who
MA
were treated with 15 sessions of imaginal and in vivo exposure therapy. The authors found that
greater subjective distress and higher physiological response (heart rate and electrodermal
D
TE
activity) during exposure predicted greater reductions in obsessional fear. In another small study
with female assault victims with PTSD, greater facial expression of fear (as coded by the Facial
P
Action Coding System by one of the reliably trained authors of the study) during the first
CE
imaginal exposure of prolonged exposure (PE) therapy was associated with greater improvement
AC
in treatment outcome (Foa, Riggs, Massie, & Yarczower, 1995). The study results revealed
significant associations between higher objectively measured fear activation and improved
outcome, and further found that fear activation appeared to mediate the relationship between
higher pre-treatment PTSD severity and better treatment outcome, implicating its importance in
impacting outcome.
Indirect evidence for the role of activation was found by Jaycox, Foa and Morral (1998),
who conducted cluster analyses using subjective distress ratings every five minutes during
imaginal exposure of PE therapy in female assault victims with PTSD. They found that patients
ACCEPTED MANUSCRIPT
reporting higher subjective fear activation (or emotional engagement) and gradual between-
session habituation (discussed later) showed greater improvements in PTSD symptoms than the
other patients. Similarly, in a sample of individuals with simple phobia, social phobia and panic
PT
disorder, Lang, Bradley, and Cuthbert (1998) found that greater physiological arousal during
RI
imaginal exposure was associated with better treatment outcome.
Further, an examination with Vietnam veterans with PTSD who received flooding
SC
therapy (prolonged exposure to a traumatic memory without the processing component of PE)
NU
found increased subjective (self-reported anxiety and fear) and physiological (heart rate, skin
conductance, and electromyogram) activation during flooding (Pitman, Orr, Altmann, &
MA
Longpre, 1996). However, in contrast to the studies discussed above, only heart rate activation in
the first flooding session was associated with a decrease in PTSD symptoms at post-treatment
D
TE
(specifically, daily number of intrusive combat memories); the other measures of fear activation
were not associated with outcome. It is important to note that patients in this study also only
P
for PTSD. The finding that heart rate was the only physiological measure of fear activation
AC
related to outcome is consistent with the findings reported by Watson and Marks (1971) with
fear-relevant cues.
Indeed, several recent studies have found only partial or mixed results regarding the
relationship between fear activation during extinction learning (exposure) and treatment
outcome. van Minnen and Hagenaars (2002) categorized a group of patients with PTSD as either
responders (defined as those showing at least a 50% reduction in PTSD symptoms after 9
sessions of imaginal exposure), non-responders (for those not meeting the improvement
criterion), or as drop-outs from treatment (defined as anyone who did not complete the 9 sessions
ACCEPTED MANUSCRIPT
of treatment, most of whom dropped out before session 2). The results were inconsistent. Overall
there were no differences between groups in mean and peak subjective anxiety ratings during the
first imaginal exposure. Contrast tests showed that the non-responders showed higher pre-
PT
imaginal ratings of subjective distress prior to starting their first imaginal exposure than those
RI
who improved over the course of treatment. On the other hand, the responder group showed a
greater increase (i.e., more fear activation) in subjective distress over the course of the first
SC
imaginal exposure compared to those who did not respond to the treatment. Only this latter
NU
finding is in line with the previous supporting studies of EPT. Initial or across-treatment
rate, carbon dioxide partial pressure, and respiration rate), as well as self-reported panic
D
TE
symptoms and subjective anxiety ratings, in patients with panic disorder and agoraphobia who
received three in vivo exposure sessions (Meuret, Seidel, Rosenfield, Hofmann, & Rosenfield,
P
2012). The results revealed that greater subjective anxiety, but not physiological measures of
CE
anxiety and panic during exposure were predictive of better improvement in panic symptoms.
AC
This finding for a significant relationship between subjective fear activation, and not
physiological activation, and treatment outcome is inconsistent with the previous findings by
Related to this, Baker et al. (2010) utilizing an analogue sample of college students
distress or initial heart rate activation and treatment outcome following two exposure sessions
conducted one week apart, or at two-week follow-up. Notably, these two more recent studies,
unlike the previous studies supporting the role of fear activation in improving treatment outcome,
ACCEPTED MANUSCRIPT
utilized either less-than-optimal exposure protocols (only 2-3 exposures conducted over a short
space of time) or sub-clinical populations. On the other hand, these studies used more advanced
statistical approaches such as stepwise regression or multilevel modeling, while the previous
PT
examinations relied primarily on correlational analyses or simple t tests.
RI
Overall, a number of other studies conducted in both analogue and clinical samples over
the past few decades have found either no relationship between fear activation and outcome
SC
(Bluett, Zoellner, & Feeny, 2014; Rauch, Foa, Furr, & Filip, 2004; Tsao & Craske, 2000; Sloan
NU
& Telch, 2002; Telch, 2004), or even a negative relationship between fear activation and
outcome, with higher activation being associated with poorer outcome (Foa et al., 1983;
MA
Kircanski et al., 2012). Taken together with the other studies described above, these findings
suggest a mixed picture with regards to importance of fear activation in treatment outcome.
D
TE
Watson and Marks (1971) discussed two hypothesized mechanisms through which
P
flooding may reduce pathological fear: 1) flooding works by blocking avoidance of the phobic
CE
stimuli and thereby permitting extinction to take place; 2) flooding works by allowing
AC
habituation1 of fear in general rather than the particular phobic fear. Watson and Marks plotted
the change in fear levels within and between exposure sessions and reported that earlier flooding
sessions were more anxiety-provoking for participants than were later sessions. They did not,
however, report on the significance of changes in anxiety within or between sessions, nor did
1
As mentioned previously, Watson and Marks used the terms “extinction” and
“habituation” somewhat interchangeably, while many researchers today generally consider the
term habituation inaccurate and argue that extinction more correctly connotes an active process
of new learning (see Bouton, 2004; Van Elzakker, Dahlgren, Davis, Dubois, & Shin, 2014). We
have therefore retained the terminology “extinction learning” to be consistent with both animal
and human models examining mechanisms of fear reduction.
ACCEPTED MANUSCRIPT
they examine how these changes differed across conditions or related to treatment outcomes.
Nonetheless, in light of their finding that both the irrelevant and relevant flooding reduced fear to
a similar extent but had different prognostic correlates, the authors concluded that the irrelevant
PT
flooding may reduce fear through habituation to anxiety in general while relevant flooding may
RI
reduce fear through blocking avoidance and thereby permitting extinction, and that these
processes may not be mutually exclusive. Experimental research on flooding thrived in the
SC
decades following the Watson and Marks (1971) study. Flooding had emerged as a powerful
NU
clinical tool to treat a host of anxiety-related conditions and there was much interest in
delineating the mechanisms and optimal parameters of flooding. The reduction of subjective and
MA
physiological indices of anxiety during flooding was seen as a key mechanism.
In terms of the terminology frequently used to describe such reductions, the degree to
D
TE
which anxiety decreases during a single extinction trial or session of exposure therapy is referred
to as within-session extinction (WSE). The degree to which peak anxiety decreases across
P
(BSE). To date, dozens of clinical and experimental studies have examined whether these
AC
processes are related to good outcomes following exposure therapy for anxiety-related conditions
(for a thorough review of WSE and BSE as potential mechanisms of exposure therapy, see
Craske et al., 2008). Both WSE and BSE have been measured by examining changes in
subjectively reported anxiety ratings using the subjective units of distress scale (SUDS), and
change in physiological measures of anxious arousal such as heart rate and skin conductance.
the beginning to the end of an exposure session, the role of WSE in treatment outcomes has not
been well supported overall. A small number of studies have found some support for a direct
ACCEPTED MANUSCRIPT
relationship between WSE and outcome. Foa et al. (1983) found that WSE of subjective fear was
receiving EX/RP for OCD. With PTSD, the previously mentioned study by Pitman et al., (1996)
PT
found that of several subjective and physiological indices of anxiety measured, there was a non-
RI
significant trend for WSE in heart rate to correlate with the number of intrusions per day, which
was one of several PTSD-related outcomes examined. In addition, a more recent study by
SC
Norton, Hayes-Skelton, and Klenck (2011) found that WSE in subjective distress during group
NU
exposure therapy sessions for mixed anxiety disorders was associated with superior treatment
outcomes. Similarly, among contamination fearful undergraduates, Kircanski et al. (2012) found
MA
that WSE in subjective distress was associated with significantly lower obsession severity. On
the other hand, WSE in subjective distress was not related to compulsion severity or reported
D
TE
distress during a behavioral avoidance task and WSE in heart rate was not related to any
outcomes.
P
Support for the importance of WSE has also been found by two clinical studies and three
CE
non-clinical studies that have indirectly examined the relationship between WSE and outcomes.
AC
Beckham, Vrana, May, Gustafson, & Smith (1990) found that flying phobia participants who
showed greater WSE in heart rate during an in vivo exposure session were more likely to elect to
fly in the subsequent 8 weeks than those who showed less WSE in heart rate. van Minnen and
Hagenaars (2002) found that responders to PE for PTSD showed greater WSE in subjective
distress than non-responders, although this was only true for imaginal exposures completed for
homework; there were no differences in WSE between responders and non-responders during
sessions. In a nonclinical sample of snake fearful women, Lang, Melamed and Hart (1970) found
that participants who achieved greater fear reduction following systematic desensitization
ACCEPTED MANUSCRIPT
showed greater WSE in heart rate than those with less post-treatment fear reduction. In their non-
clinical sample of speech phobic undergraduates, Borkovec and Sides (1979) found that
participants who received systematic desensitization showed greater WSE in subjective distress
PT
and superior post-treatment outcomes than participants who received either exposure with or
RI
without contiguous relaxation, or exposure only. Using a similar non-clinical sample, Chaplin
and Levine (1981) found that participants who received continuous exposure demonstrated
SC
greater WSE than those who received interrupted exposure, and that the continuous exposure
NU
group demonstrated lower anxiety on two of three post-treatment outcomes. Oliver and Page
(2003) found that blood-injection fearful participants who engaged in a distracting conversation
MA
during exposure showed greater WSE in subjective distress than those who engaged in a
conversation focused on the phobic stimuli during exposure or those who received exposure
D
TE
alone (with no conversation) and were also more likely to show superior outcomes at post-
A larger number of studies have failed to find any support for the relationship between
CE
WSE and clinical outcomes. Jaycox et al. (1998) found that WSE in self-reported anxiety was
AC
unrelated to PE treatment outcome among sexual assault survivors. In a small sample of PTSD
participants, Sripada and Rauch (2015) found that WSE in subjective distress was unrelated to
both symptom change and treatment responder status following PE for PTSD. In a
nonrandomized study by van Minnen and Foa (2006), participants with PTSD who received 60-
minutes of imaginal exposure evidence significantly greater WSE than those who received 30-
minutes. However, there were no differences between groups in terms of post-treatment PTSD
outcomes. This finding was replicated in a small RCT showing that WSE was greater for
participants with PTSD who received 40-minutes of imaginal exposures versus 20-minutes, but
ACCEPTED MANUSCRIPT
again, there were no differences in outcome between the two groups (Nacasch et al., 2015).
These latter studies indicate that although length of exposure does influence the degree of WSE,
PT
Null results have also been found in studies of exposure therapy for OCD and panic
RI
disorder. In a study designed to evaluate the role of attention versus distraction during exposure
for OCD, Grayson, Foa, and Steketee (1982) found that WSE in heart rate and subjective anxiety
SC
was evident in both treatment conditions, but was unrelated to return of subjective fear at the
NU
beginning of the next session. Kozak et al. (1988) found that WSE in subjective anxiety, heart
rate, and electrodermal activity were all unrelated to any of the clinical outcomes assessed
MA
among participants receiving EX/RP for OCD. There was also a trend for BSE in subjective
anxiety to predict improvement in rated fear and avoidance. Among panic disorder participants
D
TE
receiving exposure-based CBT plus either a low or moderate dose of alprazolam, Riley et al.
(1995) found that WSE of physiologic measures was unrelated to responder status at post-
P
treatment. Similarly, Meuret et al. (2012) found that neither WSE in physiological or subjective
CE
anxiety was related to clinical outcomes among panic disorder participants receiving exposure
AC
therapy.
Many studies with non-clinical samples have also failed to support the WSE-outcome
relationship. Among their sample of height-fearful participants, Baker et al. (2010) found that
WSE in subjective distress and heart rate during the in vivo exposure sessions were both
(2012) found that WSE in neither subjective anxiety nor heart rate was related to self-report,
speaking. Moreover, greater WSE in subjective anxiety was related to worse outcomes on a
ACCEPTED MANUSCRIPT
behavioral avoidance task. Results from non-clinical studies examining the relationship between
WSE and outcomes indirectly are consistent with the above null findings. In a sample of height-
fearful undergraduates, Lang and Craske (2000) found that although WSE in heart rate was
PT
greater among participants who received constant versus varied exposure, there were no
RI
differences in treatment outcomes between groups. Similarly, Kamphuis and Telch (2000) found
that undergraduates with claustrophobia fears who received one session of exposure with guided
SC
threat reappraisal showed superior fear reduction than those who received exposure with a
NU
distraction task, despite the fact that WSE was equal across groups. Using a similar sample,
Sloan and Telch (2002) compared exposure with guided reappraisal to exposure with safety-
MA
seeking behavior. Once again, WSE was equal across groups, but the reappraisal condition was
undergraduates with public-speaking anxiety, Tsao and Craske (2000) found that those who
received spaced exposure had superior follow-up outcomes than those who received massed
P
exposure, despite the fact that significant WSE in heart rate or subjective fear was not observed
CE
in either group.
AC
Empirical Data: Between-session extinction. In contrast to WSE, BSE has been shown
to predict exposure therapy outcomes more consistently. Several studies of PTSD have found
that extinction between imaginal exposure sessions is related to lower PTSD severity at post-
treatment. Among 69 participants with PTSD receiving PE, Rauch et al. (2004) found that
greater BSE in subjective distress over the course of six imaginal exposure sessions was
associated with more reduction in PTSD symptoms at post-treatment. In contrast to the null
findings on WSE, van Minnen and Foa (2006) found that BSE was significantly related to PTSD
symptoms at post-treatment among participants with PTSD, regardless of whether they received
ACCEPTED MANUSCRIPT
sessions with 60- or 30-minutes of imaginal exposure. Also in contrast to null WSE findings,
Sripada and Rauch (2015) found that BSE in subjective distress was related to both symptom
PT
Several clinical studies provide support for the role of BSE on some, but not all indices of
RI
BSE or across outcomes. In OCD patients, Kircanski et al. (2012) found that BSE in heart rate
was associated with lower obsession severity and lower reported distress during a behavioral
SC
avoidance task (but not compulsion severity). In this study, BSE in subjective distress was not
NU
related to any outcomes. Kozak et al. (1988) showed that BSE in heart rate (but not
subjective anxiety to predict improvement in rated fear and avoidance. Among OCD patients
D
TE
receiving EX/RP, Foa et al. (1983) found that greater BSE in subjective anxiety was associated
with superior outcomes at post-treatment, but not at follow-up. In a non-clinical sample, Baker et
P
al. (2010) found that BSE in subjective distress (but not heart rate) across two in vivo exposure
CE
sessions was related to lower reported distress during a behavioral avoidance task (but not to
AC
noted, Pitman et al. (1996) found null results for several process variables and most of the
subjective and physiological indices of anxiety measured, however, there was a non-significant
trend for BSE in heart rate to correlate with the number of intrusions per day.
A larger number of studies provide indirect support for the relationship between BSE and
exposure therapy outcomes by showing that those who show greater BSE also tend to have
superior outcomes. The cluster analysis conducted in the study by Jaycox et al. (1998) found that
sexual assault survivors with PTSD fit one of three different patterns of change during prolonged
ACCEPTED MANUSCRIPT
exposure therapy: 1) high initial fear activation and gradual BSE, 2) high fear activation and no
BSE, and 3) moderate fear activation and no BSE. Providing indirect support for the relationship
between BSE and outcome, participants in the first group showed greater improvement than
PT
those in the other two groups. van Minnen and Hagenaars (2002) found that responders to PE
RI
showed greater BSE in subjective distress compared to non-responders. However, in this study,
BSE was only measured between the first two sessions of prolonged exposure; change in
SC
subjective distress over all nine sessions of imaginal exposure was not assessed. In an RCT
NU
testing the effects of adding d-cycloserine (DCS) to virtual reality exposure therapy for
acrophobia, Ressler et al. (2004) showed that compared to those who received a placebo, those
MA
who received DCS had greater BSE in subjective anxiety and BSE in spontaneous fluctuations in
skin conductance, and showed significantly greater improvement at post-treatment and 3-month
D
TE
follow-up. In a large sample of PTSD patients receiving PE therapy, Bluett et al. (2014) showed
that the minority of participants (35.3%) who exhibited a reliable change in distress (i.e., reliable
P
BSE) across sessions had lower PTSD severity at post-treatment compared to those who
CE
exhibited a lack of reliable BSE. However, given that both groups of participants made marked
AC
improvements during treatment and there were no differences in PTSD diagnostic status, the
authors concluded that BSE is not necessary to achieve good PTSD outcomes.
Several studies examining the role of attentional processes on extinction with non-clinical
samples also provide some indirect support for the relationship between BSE and outcome.
Kamphuis and Telch (2000) found that participants with claustrophobia fears who received one
session of exposure with guided threat reappraisal showed greater fear reduction and less return
of fear than those who received exposure with a distraction task, their combination, or exposure
alone. Partial indirect support was found for the BSE-outcome relationship by the finding that
ACCEPTED MANUSCRIPT
BSE in subjective anxiety was significantly greater in those who received exposure with guided
threat reappraisal, either with or without a cognitive load task. In a related study, Sloan and
Telch (2002) found that participants who received exposure with guided reappraisal showed
PT
greater BSE in subjective anxiety and had superior outcomes on several measures at post-
RI
treatment (but not at follow-up) compared to those who received exposure with safety-seeking
SC
Oliver and Page (2003) found that those who engaged in distraction during exposure showed
NU
greater BSE in subjective distress relative to those who focused on the phobic stimuli during
exposure or those who received exposure alone. Compared to the other two groups, participants
MA
in the exposure plus distraction condition showed superior outcomes at post-treatment and
follow-up. Finally, Telch et al. (2004) found that claustrophobic participants who engaged in a
D
TE
distraction task during exposure had worse outcomes and less BSE in subjective fear than those
In contrast to the above findings, some studies have failed to find support for the
CE
relationship between BSE and outcomes. In participants with PTSD receiving PE, Nacasch et al.
AC
(2015) found that although BSE was greater for those who received 40-minutes of imaginal
exposures versus 20-minutes and greater BSE in the overall sample was significantly correlated
to greater reduction in PTSD symptoms, there were no differences in outcome between the
groups that received shorter or longer duration of imaginal exposure. Meuret et al. (2012) found
that despite significant BSE in measures of physiological and subjective anxiety over the course
of exposure therapy for panic disorder, none of the indices of BSE were associated with clinical
outcomes. Similarly, Riley et al. (1995) found that BSE of physiologic measures was unrelated
The role of BSE was also examined indirectly in several analogue studies by Craske and
colleagues. Rowe and Craske (1998) found that spider-fearful participants who were exposed to
varying spider stimuli showed greater BSE in subjective anxiety prior to and during approach to
PT
a novel spider than those exposed to the same spider. Despite this difference in BSE, there were
RI
no group differences in outcome at post-assessment, and those exposed to varying stimuli
showed superior outcomes during the 3-week follow-up. However, exposure in this study
SC
consisted of seven short exposure trials completed in a single experimental session, and therefore
NU
the results may not generalize to a typical course of exposure therapy.
Tsao and Craske (2000) compared four trials of exposure delivered in a single session
MA
(massed exposure), over an equal number of intervening days (uniformly-spaced exposure), and
fearful of public speaking. Although BSE in subjective anxiety was equal across groups and BSE
in heart rate was not observed in any group, the expanding-spaced exposure group showed
P
superior outcomes at the post-treatment follow up. Among height-fearful participants, Lang and
CE
Craske (2000) examined twelve 5-minute exposure trials delivered in a massed versus expanding
AC
spaced schedule as well as constant versus varied exposure. Participants who received massed
and constant exposure showed greater BSE than those who received expanding and varied
exposure; however, there were no consistent differences in outcomes across groups. Results of
these latter two studies are consistent with those of Chaplin and Levine (1981), who compared
continuous versus interrupted exposure for public speaking anxiety and found that continuous
exposure was associated with superior outcomes, despite the fact that both groups demonstrated
significant BSE.
ACCEPTED MANUSCRIPT
inconsistent findings, particularly with regard to BSE, is that the relevant mechanism is not fear
reduction, but rather, increased tolerance to fear or distress during phobic situations. Indeed,
PT
Craske et al. (2008), among others, have suggested that the ability to engage in goal-directed
behavior despite experiencing distress, and being able to tolerate one’s anxiety during exposure,
RI
ensures that extinction learning will occur, regardless of reduction in fear. Several empirical
SC
studies have found significant associations between distress tolerance and degree of anxiety
NU
symptoms (e.g., Schmidt, Richey, & Fitzpatrick, 2006; Marshall-Berenz, Vujanovic, Bonn-
Miller, Bernstein, & Zvolensky, 2010), but the role of distress tolerance as a potential moderator
MA
or mechanism of outcome in exposure therapy has been less explored than the other mechanisms
examined here.
D
TE
One study by McHugh et al. (2014) found that for patients in a partial hospitalization
CBT program, increases in distress tolerance were associated with lower anxiety symptoms (not
P
specific to a particular diagnosis) at post-treatment. However, this study did not examine anxiety
CE
symptoms specific to a particular anxiety disorder and did not appear to routinely utilize
AC
exposures. On the basis of their findings, Bluett et al. (2014) suggested that while distress/fear
reduction was not associated with outcome in individuals receiving exposure therapy for PTSD,
better treatment outcome may be associated with increase in distress tolerance (as evidenced by
note, this study did not measure distress tolerance directly. In a community sample of individuals
who scored low on a measure of distress tolerance, Macatee and Cougle (2015) introduced a
intervention successfully improved distress tolerance in the active condition compared to the
ACCEPTED MANUSCRIPT
waitlist control condition, and this improvement in distress tolerance mediated urges to
neutralize intrusive thoughts (mental compulsions). A potential shortcoming of this study was
the use of an analogue sample of participants who were not necessarily high in OCD symptoms;
PT
given the possibility that distress tolerance is more central to the maintenance of anxiety
RI
pathology, such an intervention would need to be tested further to determine whether it shows
SC
exposure therapy shows promise. However, given that the methods by which distress tolerance is
NU
acquired and adequately measured are still largely unknown, considerable additional theoretical
and empirical work is needed to fully understand how distress tolerance fits in with animal and
MA
human models of fear extinction.
D
Summary of Findings
TE
developments and the empirical evidence examining the importance of specificity of content in
P
CE
extinction learning since the Watson and Marks (1971) article was published taken together
suggest that: (1) the field generally veered away from the findings by Watson and Marks (1971),
AC
and continued to test protocols that included exposure content that is relevant to core fears of the
patients being tested; and (2) in fact, beyond a decade after their paper, the interest in examining
From a review of developments in the field since this time, it is not clear why this issue
was not pursued further, but it is likely because it was not in line with the forefront theories (e.g.,
EPT, bioinformational theory) about extinction learning that emerged after this study, which as
described previously, outline how important it is to tap into the relevant core fear in order to
affect meaningful fear reduction. Again, it is worth noting that the study by Watson and Marks
ACCEPTED MANUSCRIPT
(1971) arguably did not actually employ fear stimuli that were totally irrelevant to the
individual’s pathological fear structure, which extant theories of fear extinction would posit
might have led to some extinction effects since even with these irrelevant stimuli there was some
PT
degree of matching to the underlying fear structure (Foa & Kozak, 1986; Lang et al., 1977).
RI
Exploring the utility of tapping into irrelevant fears (more strictly defined than what was
done in the Watson and Marks study) may be a fruitful avenue to explore further. Such an
SC
endeavor would certainly be in line with the more recent transdiagnostic approaches being
NU
adopted for preliminary clinical efficacy trials (Farchione et al., 2012; Craske, 2012), which calls
for the use of assessment and intervention strategies that cut across diagnoses and specific fear
MA
domains. For example, the theory underlying one such transdiagnostic treatment (the Unified
Protocol, UP; Barlow et al., 2011) posits that there are more similarities in the etiology and latent
D
TE
structures (e.g. underlying common factors such as neuroticism and behavioral inhibition) of
various anxiety and mood disorders than there are differences. Thus, this treatment approach
P
focuses overall on increasing tolerance to strong emotions (not just specific fears) and correcting
CE
maladaptive emotion regulation strategies. However, without more vigorous and systematic
AC
empirical investigation, the utility of exposures to generally distressing stimuli that do not
specifically aim to match participants’ core fears is unclear, and currently unsupported.
Fear activation as a mechanism of exposure. Despite the many strides made by the
theoretical models discussed above in helping us understand how fear activation impacts
treatment outcome, there are many questions that have not been sufficiently examined, and doing
so may explain some of the inconsistencies found in the empirical evidence regarding the
relationship between fear activation and treatment outcome. For instance, Foa, Hembree, and
Rothbaum (2007) coined the terms over-engagement and under-engagement, suggesting that
ACCEPTED MANUSCRIPT
both too little activation and too much activation hinder emotional processing and therefore
treatment outcome. That is, it is possible that over-activation might actually hinder the corrective
learning and disconfirmation needed to result in successful reduction of fear. Similarly, McNally
PT
(2007) has suggested that there is an optimal level of fear activation required for fear extinction.
RI
To this end, distraction can be viewed as a means of titrating the level of fear activation, at least
SC
patients during exposure was shown to result in superior outcome, particularly in those reporting
NU
higher levels of initial baseline fear; Johnstone & Page, 2004). Indeed, Foa et al. (2006) modified
the original EPT to incorporate such findings and suggest that distraction may serve as a helpful
MA
way to moderate level of fear activation in certain anxiety disorders (such as specific phobias)
and as a maladaptive cognitive form of avoidance in other disorders (such as agoraphobia), but
D
TE
Other researchers (e.g., Craske et al., 2008) have noted that roughly an equal number of
P
studies have either supported or refuted the notion that fear activation (particularly initial fear
CE
activation in the first exposure session) meaningfully predicts treatment outcome across a range
AC
of fear-based disorders, thus its status as a mechanism of outcome is still unclear. The
methodological and assessment differences across these studies have also been purported to
account for the observed inconsistencies in findings; for example, some studies utilize
physiological indices of fear activation while others do not, different lengths and spacing of
exposures, use of clinical versus subclinical samples (Craske et al., 2008). In addition, it is
important to acknowledge that many of the earlier studies done on this area utilized smaller
samples (under 20 participants), although this is not unusual for initial studies examining
mechanisms of psychopathology, and effects of fear activation on outcome were noted even with
ACCEPTED MANUSCRIPT
these small sample sizes. Taken together, while the impact of adequate fear activation during
extinction learning on subsequent fear reduction is still in flux, it is crucial that this potential
mechanism is studied more systematically and vigorously. Indeed, the implications for exposure
PT
treatment of anxiety-related disorders are considerable based on the determination of whether
RI
fear activation is necessary for successful treatment outcomes.
WSE and BSE as mechanisms of treatment outcome. In terms of the role of WSE, it
SC
seems that while a small number of studies have found evidence that WSE is related to exposure
NU
therapy outcomes, the majority of studies do not support the idea that WSE is necessary for good
outcomes. Some researchers have suggested that null findings may be due to a limited range on
MA
this process variable, given that WSE occurs for most patients. However, this interpretation is
inconsistent with studies that find good reduction in symptoms from pre-to post-treatment in the
D
TE
absence of significant WSE. Thus, although patients may experience WSE during exposure
In contrast to findings on WSE, BSE has been associated with superior outcomes with
CE
some consistency. To date, however, the majority of supporting studies has not examined the
AC
relationship between BSE and outcome directly, but have instead examined whether group
differences in BSE match with group differences in outcome. For both WSE and BSE studies,
methodological differences in how process variables are measured (e.g., subjective ratings vs.
In light of the somewhat mixed findings described above, the necessity of WSE and BSE
has been the subject of much debate. In the most recent update to EPT, Foa et al. (2006) reported
that most studies do not find a direct relationship between WSE and treatment outcomes and
ACCEPTED MANUSCRIPT
suggest that WSE may not, in fact, be a reliable indicator of emotional processing. Researchers
such as Craske et al. (2008) and others have called to question the necessity of WSE as well as
BSE for therapeutic recovery. Craske et al. proposed that fear tolerance, defined as the ability to
PT
engage in goal-directed behavior despite experiencing distress is a more important clinical
RI
concept than fear reduction (i.e., WSE or BSE), and that as long as the participant is able to
tolerate their anxiety during exposure, extinction learning will occur regardless of change in
SC
distress levels. Indeed, modern learning theories of extinction (e.g., Bouton, 2004) propose that
NU
fear tolerance promotes extinction learning which has been considered a key mechanism of
References
Allen, L. B., & Barlow, D. H. (2009). Relationship of exposure to clinically irrelevant emotion
PT
http://dx.doi.org/10.1177/0145445509344180
RI
Baker, A., Mystkowski, J., Culver, N., Yi, R., Mortazavi, A., & Craske, M. G. (2010). Does
habituation matter? Emotional processing theory and exposure therapy for acrophobia.
SC
Behaviour Research and Therapy, 48(11), 1139-1143.
NU
http://dx.doi.org/10.1016/j.brat.2010.07.009
Barlow, D. H., Farchione, T. J., Fairholme, C. P., Ellard, K. K., Boisseau, C. L., Allen, L. B., &
MA
Ehrenreich-May, J. (2011). Unified protocol for transdiagnostic treatment of emotional
disorders: Therapist guide. New York, NY, US: Oxford University Press.
D
TE
Baum, M., & Leclerc, R. (1974). Irrelevant stress vs response prevention (flooding) interpolated
Beckham, J. C., Vrana, S. R., May, J. G., Gustafson, D. J., & Smith, G. R. (1990). Emotional
AC
http://dx.doi.org/10.1016/0005-7916(90)90002-3
Bluett, E. J., Zoellner, L. A., & Feeny, N. C. (2014). Does change in distress matter?
http://dx.doi.org/10.1016/j.jbtep.2013.09.003
Borkovec, T. D., & Sides, J. K. (1979). The contribution of relaxation and expectancy to fear
ACCEPTED MANUSCRIPT
reduction via graded, imaginal exposure to feared stimuli. Behaviour Research and
Bouton, M. E. (2004). Context and behavioral processes in extinction. Learning & Memory,
PT
11(5), 485-494. http://dx.doi.org/10.1101/lm.78804
RI
Bouton, M. E., & Swartzentruber, D. (1991). Sources of relapse after extinction in Pavlovian and
SC
http://dx.doi.org/10.1016/0272-7358(91)90091-8
NU
Chaplin, E. W., & Levine, B. A. (1981). The effects of total exposure duration and interrupted
Craske, M. G. (2012). Transdiagnostic treatment for anxiety and depression. Depression and
D
TE
Craske, M. G., Kircanski, K., Zelikowsky, M., Mystkowski, J., Chowdhury, N., & Baker, A.
P
(2008). Optimizing inhibitory learning during exposure therapy. Behaviour Research and
CE
Culver, N. C., Stoyanova, M., & Craske, M. G. (2012). Emotional variability and sustained
Farchione, T. J., Fairholme, C. P., Ellard, K. K., Boisseau, C. L., Thompson-Hollands, J., Carl, J.
http://dx.doi.org/10.1016/j.beth.2012.01.001
Foa, E. B., Grayson, J. B., Steketee, G. S., Doppelt, H. G., Turner, R. M., & Latimer, P. R.
ACCEPTED MANUSCRIPT
006X.51.2.287
PT
Foa, E. B., Hembree, E. A., & Rothbaum, B. O. (2007). Prolonged exposure therapy for PTSD:
RI
Emotional processing of traumatic experiences: Therapist guide. New York, NY: Oxford
University Press.
SC
Foa, E. B., Huppert, J. D., & Cahill, S. P. (2006). Emotional processing theory: An update. In B.
NU
O. Rothbaum (Ed.), Pathological anxiety: Emotional processing in etiology and
2909.99.1.20
Foa, E. B., McNally, R. J., & Williams, L. (1996). Mechanisms of change in exposure therapy.
P
In R. M. Rapee (Ed.), Current controversies in the anxiety disorders (pp. 329-343). New
CE
Foa, E. B., Riggs, D. S., Massie, E. D., & Yarczower, M. (1995). The impact of fear activation
and anger on the efficacy of exposure treatment for posttraumatic stress disorder.
Foa, E. B., Steketee, G., Turner, R. M., & Fischer, S. C. (1980). Effects of imaginal exposure to
Grayson, J. B., Foa, E. B., & Steketee, G. (1982). Habituation during exposure treatment:
http://dx.doi.org/10.1016/0005-7967(82)90091-2
Jaycox, L. H., Foa, E. B., & Morral, A. R. (1998). Influence of emotional engagement and
PT
Psychology, 66(1), 185-192. http://dx.doi.org/10.1037/0022-006X.66.1.185
RI
Johnstone, K. A., & Page, A. C. (2004). Attention to phobic stimuli during exposure: The effect
SC
Research and Therapy, 42(3), 249-275. http://dx.doi.org/10.1016/S0005-7967(03)
NU
00137-2
Kamphuis, J. H., & Telch, M. J. (2000). Effects of distraction and guided threat reappraisal on
MA
fear reduction during exposure-based treatments for specific fears. Behaviour Research
Kircanski, K., Mortazavi, A., Castriotta, N., Baker, A. S., Mystkowski, J. L., Yi, R., & Craske,
Kozak, M. J., Foa, E. B., & Steketee, G. (1988). Process and outcome of exposure treatment with
Lang, A. J., & Craske, M. G. (2000). Manipulations of exposure-based therapy to reduce return
http://dx.doi.org/10.1016/S0005-7967(99)00031-5
495-512. http://dx.doi.org/10.1111/j.1469-8986.1979.tb01511.x
Lang, P. J., Bradley, M. M., & Cuthbert, B. N. (1998). Emotion, motivation, and anxiety: Brain
PT
mechanisms and psychophysiology. Biological Psychiatry, 44(12), 1248-1263.
RI
http://dx.doi.org/10.1016/S0006-3223(98)00275-3
Lang, P. J., Levin, D. N., Miller, G. A., & Kozak, M. J. (1983). Fear behavior, fear imagery, and
SC
the psychophysiology of emotion: The problem of affective response integration. Journal
NU
of Abnormal Psychology, 92(3), 276-306. http://dx.doi.org/10.1037/0021-843X.92.3.276
Lang, P. J., Melamed, B. G., & Hart, J. (1970). A psychophysiological analysis of fear
MA
modification using an automated desensitization procedure. Journal of Abnormal
intervention for low distress tolerance and its effect on performance on a neutralization
P
doi:10.1016/j.jbtep.2015.01.007
AC
Marshall-Berenz, E. C., Vujanovic, A. A., Bonn-Miller, M. O., Bernstein, A., & Zvolensky, M.
doi:10.1002/jts.20568
McHugh, R. K., Kertz, S. J., Weiss, R. B., Baskin-Sommers, A. R., Hearon, B. A., &
doi:10.1016/j.beth.2013.11.002
ACCEPTED MANUSCRIPT
psychological treatments for anxiety disorders. Clinical Psychology Review, 27(6), 750-
759. http://dx.doi.org/10.1016/j.cpr.2007.01.003
PT
Meuret, A. E., Seidel, A., Rosenfield, B., Hofmann, S. G., & Rosenfield, D. (2012). Does fear
RI
reactivity during exposure predict panic symptom reduction? Journal of Consulting and
SC
Mineka, S. (1976). Effects of flooding an irrelevant response on the extinction of avoidance
NU
responses. Journal of Experimental Psychology: Animal Behavior Processes, 2(2), 142-
153. http://dx.doi.org/10.1037/0097-7403.2.2.142
MA
Nacasch, N., Huppert, J. D., Su, Y., Kivity, Y., Dinshtein, Y., Yeh, R., & Foa, E. B. (2015). Are
Norton, P. J., Hayes-Skelton, S., & Klenck, S. C. (2011). What happens in session does not stay
CE
http://dx.doi.org/10.1016/j.janxdis.2011.02.006
Oliver, N. S., & Page, A. C. (2003). Fear reduction during in vivo exposure to blood-injection
stimuli: Distraction vs. attentional focus. British Journal of Clinical Psychology, 42(1),
13-25. http://dx.doi.org/10.1348/014466503762841986
Pavlov, I. (1927). Conditioned reflexes. New York, NY: Oxford University Press.
Pitman, R. K., Orr, S. P., Altman, B., & Longpre, R. E. (1996). Emotional processing and
http://dx.doi.org/10.1016/S0010-440X(96)90024-3
Rauch, S. A. M., Foa, E. B., Furr, J. M., & Filip, J. C. (2004). Imagery vividness and perceived
PT
anxious arousal in prolonged exposure treatment for PTSD. Journal of Traumatic Stress,
RI
17(6), 461-465. http://dx.doi.org/10.1007/s10960-004-5794-8
Rescorla, R. A., & Wagner, A. R. (1972). A theory of Pavlovian conditioning: variations in the
SC
effectiveness of reinforcement and nonreinforcement. In A. H. Black & W. F. Prokasy
NU
(Eds.), Classical conditioning II: current research and theory (pp. 64-99). New York:
Appleton-Century-Crofts.
MA
Ressler, K. J., Rothbaum, B. O., Tannenbaum, L., Anderson, P., Graap, K., Zimand, E., . . .
Riley, W. T., McCormick, M. G. F., Simon, E. M., Stack, K., Pushkin, Y., Overstreet, M. M., . . .
CE
Rowe, M. K., & Craske, M. G. (1998). Effects of varied-stimulus exposure training on fear
reduction and return of fear. Behaviour Research and Therapy, 36(7-8), 719-734.
http://dx.doi.org/10.1016/S0005-7967(97)10017-1
Schmidt, N. B., Richey, J. A., & Fitzpatrick, K. K. (2006). Discomfort intolerance: Development
Appleton-Century-Crofts.
Sloan, T., & Telch, M. J. (2002). The effects of safety-seeking behavior and guided threat
PT
reappraisal on fear reduction during exposure: An experimental investigation. Behaviour
RI
Research and Therapy, 40(3), 235-251. http://dx.doi.org/10.1016/S0005-7967(01)
00007-9
SC
Sripada, R. K., & Rauch, S. A. M. (2015). Between-session and within-session habituation in
NU
prolonged exposure therapy for posttraumatic stress disorder: A hierarchical linear
Telch, M. J., Valentiner, D. P., Ilai, D., Young, P. R., Powers, M. B., & Smits, J. A. J. (2004).
D
TE
Fear activation and distraction during the emotional processing of claustrophobic fear.
http://dx.doi.org/10.1016/j.jbtep.2004.03.004
CE
Tsao, J. C. I., & Craske, M. G. (2000). Timing of treatment and return of fear: Effects of massed,
AC
http://dx.doi.org/10.1016/S0005-7894(00)80026-X
Van Elzakker, M. B., Dahlgren, M. K., Davis, F. C., Dubois, S. & Shin, L. M. (2014). From
Pavlov to PTSD: The extinction of conditioned fear in rodents, humans, and anxiety
van Minnen, A., & Foa, E. B. (2006). The effect of imaginal exposure length on outcome of
http://dx.doi.org/10.1002/jts.20146
ACCEPTED MANUSCRIPT
van Minnen, A., & Hagenaars, M. (2002). Fear activation and habituation patterns as early
PT
Watson, J. P., & Marks, I. M. (1971). Relevant and irrelevant fear in flooding: A crossover study
RI
of phobic patients. Behavior Therapy, 2(3), 275-293. http://dx.doi.org/10.1016/S0005-
7894(71)80062-X
SC
NU
MA
D
P TE
CE
AC
ACCEPTED MANUSCRIPT
Highlights:
Many theoretical and empirical developments have occurred since Watson & Marks
(1971).
The match between the underlying fear and exposure content seems to be important.
Role of fear activation in outcome is mixed and needs more investigation.
PT
Within-session extinction is a limited predictor of outcome.
Between-session extinction has more, but still mixed, support as mechanism in
RI
extinction.
SC
NU
MA
D
P TE
CE
AC