Biological Psychology 92 (2013) 51–58
Contents lists available at SciVerse ScienceDirect
Biological Psychology journal homepage: www.elsevier.com/locate/biopsycho
Extinction, generalization, and return of fear: A critical review of renewal research in humans Bram Vervliet ∗ , Frank Baeyens, Omer Van den Bergh, Dirk Hermans Department of Psychology, University of Leuven, Belgium
a r t i c l e
i n f o
Article history: Received 18 April 2011 Accepted 12 January 2012 Available online 26 January 2012 Keywords: Fear Conditioning Human Renewal
a b s t r a c t The main behavioral signature of fear extinction is its fragility. This is exemplified by the renewal effect, where a change in the background context produces recovery of fear to a conditioned-and-extinguished stimulus. Renewal is the backbone of a widely accepted theory of extinction in animal research, as well as an important experimental model to screen novel treatment techniques. This has led to an explosion of fear renewal research in humans. However, the mere observation of return of fear in a renewal procedure is not sufficient to validate this particular theory of extinction in the tested sample/procedure. Here, we systematically outline a set of experimental tests that aid in evaluating alternative extinction/renewal mechanisms. We examine published renewal studies in human fear conditioning and conclude that the prevailing theory of extinction is often taken for granted, but critical tests are lacking. Including these tests in future research will not only reveal the fear extinction mechanism in humans, but also inspire further developments in extinction treatment research. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Human conditioning research is strongly inspired by animal conditioning findings and theories. One of the central aims is to bridge the translational gap between pre-clinical animal research and patient studies, by testing the animal-based empirical findings and theoretical models in a human (healthy and/or patient) sample using comparable experimental methodology. Validating animal-based theories in human experimental protocols provides the necessary ground for generalizing innovative applications from animals to patients. Fear extinction is a major research focus in animal conditioning, and has culminated in a widely used theory of extinction (Bouton, 1994). It is obviously also very important in the clinical domain. Fears constitute an important part of anxiety disorders; the extinction of fears contributes importantly to treatment success (Craske et al., 2008). Currently, Bouton’s theory of extinction is also widely applied in human clinical and pre-clinical research. It serves as an important guideline for investigating fear extinction on different levels (e.g., emotion, cognition, brain imaging) and for developing and screening extinction-enhancing techniques (see Vervliet, 2008). Most of these studies index the effects of extinction training by assessing contextual renewal, which is also the central
∗ Corresponding author at: Department of Psychology, University of Leuven, Tiensestraat 102, B-3000 Leuven, Belgium. E-mail address:
[email protected] (B. Vervliet). 0301-0511/$ – see front matter © 2012 Elsevier B.V. All rights reserved. doi:10.1016/j.biopsycho.2012.01.006
phenomenon on which Bouton’s theory of extinction is based. In this paper, we (1) describe the renewal phenomenon and outline Bouton’s theory of extinction, (2) summarize the critical tests that have validated this extinction theory in animal conditioning research, and (3) evaluate its empirical basis in human fear conditioning. It will become clear throughout the paper that, theoretically, different mechanisms of extinction may (co-)exist and lead up to renewal, but that critical tests are currently lacking in the human domain. Hence, detailed behavioral analysis of extinction in the human fear conditioning paradigm is necessary to further develop human theories of fear extinction, to guide interpretations of brain activity patterns underlying extinction, and to design and screen new behavioral and/or pharmacological enhancers of extinction.
1.1. Fear conditioning and extinction In general, the term ‘fear conditioning’ refers to an experimental procedure in which (1) a neutral stimulus is arranged as a reliable predictor of an aversive stimulus (mostly an electrical stimulation), and (2) the predicting stimulus starts eliciting cognitive, emotional and/or behavioral reactions in anticipation of the aversive shock. In contemporary research, psychophysiological recordings and subjective ratings are often combined in order to get a clear picture of the conditioned fear response. These recordings frequently include skin conductance responses and/or fear-potentiated startle reflexes. Ratings often ask for the expectancy of the shock, the level
52
B. Vervliet et al. / Biological Psychology 92 (2013) 51–58
of fear, or the acquired valence of the predicting stimulus (see Lipp, 2006). The conditioning process is usually conceptualized as follows. Paired presentations of a neutral stimulus (the conditional stimulus, CS) and an aversive stimulus (the unconditional stimulus, US) result in the formation of an association between the memory representations of the neutral CS and the aversive US. Future confrontations with the CS will activate its own representation and, by virtue of the association, the memory of the US as well. This ‘thinking of the aversive US’ elicits anticipatory fear. Hence, the basic associative learning framework is a sort of ‘spreading of activation’ framework. Fear extinction occurs when repeated presentations of the CS, in the absence of the US, lead to a gradual decay of anticipatory fear reactions. This apparently simple phenomenon has proven difficult to explain. Since Pavlov’s original observations (Pavlov, 1927), the crucial question has been whether behavioral extinction reflects unlearning of the original CS–US association (e.g., Rescorla and Wagner, 1972), or whether it reflects an inhibition of the original CS–US association (e.g., Pavlov, 1927; Konorski, 1967). In the former case, extinction is considered a passive process by which the CS returns to a neutral state (as if no conditioning has occurred). However, numerous studies demonstrate that fear reactions can easily return to a conditioned-and-extinguished CS (Bouton, 2002). This strongly suggests that an extinguished CS is not at all neutral: The fear reactions can only recover if the original CS–US association somehow survives extinction. Arguably, new learning occurs during extinction that inhibits, but not erases, the CS–US association (Pavlov, 1927; Konorski, 1967). This new learning is often conceptualized as the formation of an inhibitory CS–US association. Henceforth, the CS both activates and deactivates the representation of the US, so that the CS will sometimes elicit fear and sometimes not. The circumstances at test largely determine which association will determine the behavioral outcome of the CS (fear/no fear). This is convincingly demonstrated by the renewal effect.
2. Contextual renewal: extinction, generalization, and return of fear
association requires input from the CS and the extinction context in concert. Hence, presentations of the CS in any other context will lack the necessary context-input and therefore fail to activate the inhibitory association. This leaves the way open for the excitatory association to activate the US representation in memory, which then leads to a recovery of the conditioned reaction. The bottom line of the model is that the context is merged in the representation of the extinction memory (CS–noUS), but not the acquisition memory (CS–US). This is in line with the observation that conditioned fears are generally context-independent, whereas extinction effects are generally context-specific. Importantly, the theory states that the extinction context does not directly activate or inactivate the US representation; it modulates the inhibitory CS–US association. From another perspective, the extinction context retrieves the specific CS extinction memory (the inhibitory CS–US association). This has important implications for the conditions under which extinction and renewal are predicted to occur. These conditions represent the critical tests of the theory. 3. The theory of extinction: critical tests The extinction theory of Bouton (1994) has been validated in animal conditioning research by systematically rejecting a set of alternative hypotheses that are simpler in nature (and hence, more parsimonious). We briefly introduce these alternative hypotheses and present the critical tests. 3.1. Incomplete extinction The hypothesis. A fear reaction to a conditioned-andextinguished stimulus after a context change may simply be due to incomplete extinction. The critical tests. First, a control group without context change between extinction and test provides a between-groups comparison at test. Second, the fear recovery has to be compared to the level of fear on the last extinction trial(s) in the context change group. Finally, these between- and within-group comparisons should produce a significant 2 × 2 interaction between group (ABA versus AAA) and moment (end of extinction versus test) (TEST 1). 3.2. Generalization decrement
Contextual renewal refers to a situation where changes in the background context evoke a recovery of fear to a conditioned-andextinguished stimulus (see Bouton, 2002). For example, if CS–US pairings have occurred in context A, and CS-only extinction presentations have occurred in context B, renewal tests can consist of presenting the CS in context A again or in a novel context C (ABArenewal and ABC-renewal, respectively; Bouton and Bolles, 1979). The typical renewal observation is an increase of fear reactions elicited by the CS. A third variant comprises CS–US pairings and CS-only extinction presentations in the same context A, followed by tests in a novel context B (AAB-renewal; Bouton and Ricker, 1994). Control conditions typically consist of AAA (all phases in the same context) and/or ABB procedures (final test in the extinction context). Although all three types of renewal have been documented in the literature, ABC-renewal is often smaller than ABA-renewal (e.g., Harris et al., 2000) and AAB-renewal is not always observed (e.g., Bouton and King, 1983). For excellent comprehensive reviews on renewal in animal fear conditioning research, readers are referred to papers by Bouton (e.g., 1994, 2002) for behavioral analysis, and to Myers and Davis (2002) and Quirk and Mueller (2008) for neural analysis. The behavioral theory of extinction (Bouton, 1994) is strongly inspired by these contextual renewal phenomena. The observed sensitivity of extinction to contextual changes has led to the assumption that activating the second-learned, inhibitory CS–US
The hypothesis. Under the assumption that a discrete stimulus (the CS) can be perceived differently in distinct contexts, stimulus generalization processes may underlie the renewal effect (see Lovibond et al., 1984). Consider ABA-renewal. The test occurs in the exact situation of acquisition (CS-in-A), but the extinction effect has to be generalized from CS-in-B to CS-in-A. It is well known that generalization decreases to progressively dissimilar stimuli (e.g., Lissek et al., 2008). Hence, the generalization of extinction from B to A is likely to be smaller than the original acquisition in A. The net result is an increase of fear evoked by the CS when presented in A. This is a very important explanation, because it is very parsimonious: it follows directly from the oldest principles in conditioning research. The critical tests. If stimulus generalization is at the heart of the renewal phenomenon, an equal generalization decrement should emerge on the transition from A to B after acquisition. Bouton and colleagues routinely do not observe a fear decrement on the first extinction trial(s) in B (e.g., Bouton and King, 1983). Apparently, the context change (from A to B) leaves acquisition intact, whereas the opposite context change (from B to A) disrupts extinction. This asymmetry has been the crucial challenge in renewal research, as it does not follow from a simple generalization mechanism. But what if one observes a relative asymmetry, with a decrement after acquisition that is smaller than the recovery at test? At first sight, this seems to prove that fear acquisition generalizes more
B. Vervliet et al. / Biological Psychology 92 (2013) 51–58
easily than fear extinction. However, contemporary models of stimulus generalization can also account for such relative asymmetry based on simple principles. For example, Pearce (1987) proposed that generalization is proportional to stimulus similarity, and this equally so for excitatory (acquisition) and inhibitory (extinction) associations (hence, a basic assumption of generalization symmetry). In ABA-renewal, the conditioned response may, for example, generalize 80% to the generalization stimulus (CS-in-B). During extinction, inhibition will accumulate to CS-in-B in order to counter this 80% generalized excitation. The end point after extinction is 80% inhibition acquired by CS-in-B. Importantly, this acquired inhibition will have to generalize back to CS-in-A at test, following the same 80% generalization rule. This leaves only 64% of the inhibition active at renewal test in A, while the excitation is 100%. The net result is therefore a recovery of 36% of conditioned responding (100–64%), while the initial decrement after acquisition was 20%. Hence, a relative asymmetry in the ABA-design does not imply intrinsically weaker generalization of extinction than acquisition. There is one exception. In case of complete generalization of acquisition (from A to B), the observation of renewal does imply a weaker (incomplete) generalization of extinction (from B to A). The strength of the generalization of acquisition should therefore always be tested via a simple t-test comparing the last trial(s) of acquisition and the first trial(s) of extinction (TEST 2). A nonsignificant result can suggest a (nearly) complete generalization of acquisition. Further evidence can be acquired by comparing the rate of extinction with a control group without context change. Again, the absence of a significant difference can suggest complete generalization of acquisition. It is advisable, if not necessary, to report the effect size of the observed difference, to document further the absence of a significant effect (as the absence of evidence is in itself not evidence of absence). 3.3. The test context: simple summation of excitation The hypothesis. During CS–US pairings in acquisition, the surrounding context A can also enter the conditioning process, because it is always around when the USs occur. Hence, a context-US association may develop, leading to contextual excitation and fear (e.g., Rescorla and Wagner, 1972). The return to the acquisition context may activate this contextual excitation, which can summate with the CS and produce a ‘recovery’ of fear. The critical tests. If context A acquires associative strength during acquisition, it should contribute to the level of conditioned fear elicited by the CS (summation). This implies a certain fear decrement when the CS is presented in a different context B (cf. TEST 2). In addition, the level of contextual excitation may be measured directly (a) through direct tests of context-induced fear (e.g., the baseline fear-potentiated startle reflex in that context), or (b) through behavioral preference tests by giving free access to the acquisition context A and a neutral context and measuring the proportion of time spent in either context (TEST 3). In addition, the ABC-renewal effect shows that the acquisition context A is not necessary for the recovery effect: renewal can occur in the absence of summation of contextual excitation (TEST 4a). The same rationale applies to AAB-renewal (TEST 4b). Finally, presenting a weakly conditioned CS+ in the test context can test for summation effects directly. However, because this test has not been applied routinely in animal conditioning research, we do not include it further in our analysis here. 3.4. The extinction context: protection-from-extinction The hypothesis. During extinction in B, the context can enter the extinction process and acquire some inhibitory strength. This context-US inhibition will then also suppress the fear reactions
53
otherwise elicited by the CS. With more inhibition accruing to the context, there is less need for inhibitory learning about the CS itself. Eventually, the CS-excitation will be countered by the summation of CS-inhibition and context-inhibition. When a context change removes this contextual inhibition, a recovery of CS-fear is elicited (see, e.g., Lovibond et al., 1984). This mechanism can account for all three types of renewal (ABA, ABC, AAB). The difference with Bouton’s theory of extinction is that the extinction context does not retrieve a CS-specific extinction memory. Rather, it is a general safety signal that reduces fear to any “danger” stimulus that is presented in that context. Critical tests. An extra stimulus can be conditioned in context A or in another context and then presented in the extinction context B (TEST 5). Contextual inhibition is revealed if the level of fear to this stimulus is reduced in this extinction context (compared to a neutral context via a t-test). Bouton and colleagues ran this test in rats but found no evidence for contextual inhibition (Bouton and King, 1983). 3.5. Selective fear suppression to other extinguished stimuli This last test is based on a unique prediction from Bouton’s model of extinction. It is stated that extinction contexts have a general ability to activate extinction memories (Bouton, 1994). Hence, an extinction context is predicted to decrease fear to other stimuli that have been conditioned and extinguished (in any other context). The only requirement is that the stimulus has built up an extinction memory (an inhibitory CS–US association). Critical tests. Presenting a conditioned-and-extinguished stimulus in the extinction context of another stimulus should lead to less renewal than in a neutral context. This can be verified by a t-test comparing the test trial(s) in the extinction context versus a neutral context (TEST 6). This test is only meaningful if simple contextinhibition has been ruled out as the mechanism of extinction (see TEST 5). 3.6. Summary of the critical tests - TEST 1: A significant 2 × 2 interaction between group (ABA versus AAA) and moment (end of extinction versus test) proves that the level of fear at test is not merely due to incomplete extinction. - TEST 2: The absence of a significant t-test comparing the last acquisition trial(s) and the first extinction trial(s) after the context change suggests that renewal is more than simple stimulus generalization. In combination with TEST 3, it also suggests that renewal is more than a simple summation effect in an ABA-design. - TEST 3: In combination with TEST 2, the absence of fear elicited by the test context suggests that renewal is more than a simple summation effect (through fear measures or behavioral preference tests). - TEST 4a: The observation of ABC-renewal suggests that renewal is more than a simple summation effect. - TEST 4b: The observation of AAB-renewal suggests that renewal is more than a simple summation effect. - TEST 5: The absence of a significant t-test comparing the level of fear elicited by a separately conditioned stimulus in the CS extinction context versus a neutral context suggests that renewal is more than simple protection-from-extinction by the extinction context. - TEST 6: A significant t-test comparing the level of fear to a separately conditioned-and-extinguished stimulus in the CS extinction context versus a neutral context strongly suggests that extinction contexts act as retrieval cues for extinction memories. TEST 1 and TEST 2 merely require extra analyses of specific test moments in a normal ABA-renewal experiment (and
54
B. Vervliet et al. / Biological Psychology 92 (2013) 51–58
potentially TEST 3, if one has a direct measure of contextual fear). The other tests require more investment: The inclusion of ABCand/or AAB-renewal groups, and the inclusion of specific inhibition and selective transfer tests. In the following section, we evaluate the current state of renewal studies in human fear conditioning studies in reference to these tests.
renewal effect (fear to the extinguished CS+ in a different context). Although Bouton’s extinction theory has not been systematically tested in one and the same paradigm, different studies provide clues that are relevant in this regard. The test results are summarized in Table 2 and discussed below. 4.2. TEST 1: end-of-extinction < renewal
4. Contextual renewal studies in human fear conditioning In human fear conditioning research, it is conventional to include a control stimulus throughout the experiment. This control stimulus (CS−) is equally often presented as the CS+ during the acquisition phase, but never paired with the aversive US. Hence, the only difference between the CS+ and the CS− is their contingency with shock. Comparing the response to the CS+ with the response to the CS− reflects the actual conditioning effect, controlling for non-associative processes. This is particularly important in human fear conditioning research, because most physiological indices that are used to track emotional changes are very sensitive to such non-associative processes, like orienting, habituation, or dishabituation. Presentations of the CS− are normally continued throughout extinction and at the time of testing. A ‘true’ recovery of fear should be differential: an increase of fear to the CS+ that is not observed (or at least, significantly less) to the CS−. Measures of fear reactions typically include both subjective ratings and physiological indices (skin conductance responding and/or startle reflex modulation) to the CS+ and CS−. All twenty-three human fear conditioning studies included in this review use the renewal procedure, but only a few are focused on examining the renewal effect per se (Alvarez et al., 2007; Effting and Kindt, 2007; Milad et al., 2005a; Neumann and Kitlertsirivatana, 2010; Vansteenwegen et al., 2005). Other studies have tested treatments to reduce the renewal effect, such as presenting extinction retrieval cues during test (Dibbets et al., 2008; Dibbets and Maes, 2011; Vansteenwegen et al., 2006), conducting extinction in multiple contexts (Bandarian Balooch and Neumann, 2011; Neumann et al., 2007), manipulating context similarity and the fear-relevance of contexts (Bandarian Balooch and Neumann, 2011; Neumann and Longbottom, 2008), adding unsignaled USs during extinction (Vervliet et al., 2010), and manipulating the time interval between acquisition and extinction (Huff et al., 2009). In addition, a number of MRI and fMRI studies have focused on the implication of brain structures and brain activity patterns in extinction recall and renewal (Kalisch et al., 2006; Milad et al., 2005b, 2007), also in anxiety disorder patients (Milad et al., 2009, 2008; Rougemont-Bücking et al., 2010). Finally, the influence of sex and gonadal hormones on renewal has been examined (Milad et al., 2006, 2010), and the test–retest reliability of the contextual renewal procedure has been evaluated (Zeidan et al., in press). The current paper focuses on the basic renewal effect in these studies and evaluates the information that these studies provide with regard to the behavioral mechanism of extinction/renewal in human fear conditioning. 4.1. Evaluation of the critical tests Table 1 summarizes the procedural parameters of the twentyfour published human fear conditioning studies on contextual renewal. As can be seen, these studies include a variety of stimuli (CSs, USs), procedures (e.g., reinforcement schedule; timing; within- versus between-subjects testing), renewal designs (ABA, ABC, AAB), fear measures (ratings, skin conductance, fearpotentiated startle) and context manipulations (entire room switching, room lights switching, switching of room pictures or background colors on the screen). All studies report the basic
A genuine renewal effect requires a significant Group (ABA, AAA) × Moment (last extinction, first test) × Stimulus (CS+, CS−) interaction. This was tested in twelve of the twenty-four published studies, and revealed the predicted interaction. In addition, two studies reported significant increases to the CS+, without comparison to the CS− (see Table 2, columns 1 and 2). Hence, only fifty percent of the reported studies analyzed TEST 1, but the available evidence strongly suggests that the renewal effect reflects a stimulus-specific recovery of the conditioned response that is induced by the context change after extinction. 4.3. TEST 2: end-of-acquisition = start-of-extinction The effect of the context change after acquisition informs about the possible involvement of simple stimulus generalization processes or summation by contextual excitation. Including the CS− in the analysis, this boils down to testing the Group (ABA, AAA) × Moment (last acquisition, first test) × Stimulus (CS+, CS−) interaction. Bouton’s theory predicts the absence of a significant triple interaction. This was tested in eight studies: two studies were in line with the prediction (plus one that only tested changes to the CS+), the other five studies revealed a significant interaction (i.e. a generalization decrement). Eleven studies tested whether the conditioning effect actually survived the context change by comparing responses to the CS+ and the CS− on the first extinction trial alone; all but one revealed a significant effect (see Table 2, columns 3 and 4). Taken together, fewer than half of the studies explicitly tested for the generalization (decrement) of acquisition. A majority of these studies evidenced generalization of acquisition, but only a minority revealed the absence of a detectable decrement. Of the latter studies, only one reported the effect size: Cohen’s d with regard to the CS+ was smaller than 0.06 (Bandarian Balooch and Neumann, 2011). 4.4. TEST 3: excitation by test context Alvarez et al. (2007) provided the only direct test by comparing a subgroup of participants that showed significant fear of the test context versus a subgroup that did not show fear of that context. They observed no significant difference in the size of the renewal effect (CS+ versus CS−), suggesting that the ABA-renewal effect was not driven by simple summation. In addition, the observation of differential (CS+ versus CS−) fear recovery also goes against simple summation by the test context, as that should equally augment fear responding to the CS−. This was found in twelve studies (see TEST 1). 4.5. TEST 4a: ABC A genuine renewal effect requires a significant Group (ABC, AAA) × Moment (last extinction, first test) × Stimulus (CS+, CS−) interaction effect. Neumann and Kitlertsirivatana (2010) observed this interaction, whereas Effting and Kindt (2007) only observed a Group × Moment interaction, indicating an equal increase to the CS+ and CS− upon the context change from B (extinction) to C (test). Hence, there is evidence of renewal in a neutral test
Table 1 Overview of the crucial parameters of the reported studies on renewal of human fear conditioning. Abbreviations: PTSD = post-traumatic stress disorder; pic = picture; stim = stimulation; SCR = skin conductance response; BOLD fMRI = blood-oxygination level dependent functional magnetic resonance imaging. Participants
CS+/CS−
Aversive US
Contexts
Fear measures
Test
Alvarez et al. (2007) Bandarian Balooch and Neumann (2011) Dibbets et al. (2008) Dibbets and Maes (2011) Effting and Kindt (2007) Huff et al. (2009) Kalisch et al. (2006) Milad et al. (2005a) Milad et al. (2005b) Milad et al. (2006) Milad et al. (2007) Milad et al. (2008) Milad et al. (2009) Milad et al. (2010) Neumann et al. (2007) Neumann and Longbottom (2008) Neumann and Kitlertsirivatana (2010) Rauch et al. (2005) Rougemont-Bücking et al. (2010) Vansteenwegen et al. (2005) Vansteenwegen et al. (2006) Vervliet et al. (2010) Zeidan et al. (in press)
13 (healthy) 103 (healthy) 75 (healthy) 183 (healthy) 125 (healthy) 66 (healthy) 17 (healthy) 30 (healthy) 14 (healthy) 42 (healthy) 17 (healthy) 14 monozygotic male twins 16 PTSD and 15 trauma-controls 54 (healthy) 64 (healthy) 134 (healthy) 60 (healthy) 14 (healthy) 16 PTSD and 15 trauma-controls 40 (healthy) 32 (healthy) 32 (healthy) 18 (healthy)
A high and low tone Geometric figures Geometrical figures Pics of neutral faces Drawings of faces Spider and snake pics Pics of angry faces Coloring of lamp pic. Coloring of lamp pic. Coloring of lamp pic. Coloring of lamp pic. Coloring of lamp pic. Coloring of lamp pic. Coloring of lamp pic. Geometric figures Fear-relevant/fear-irrelevant Pics of real-life objects Coloring of lamp pic. Coloring of lamp pic. Drawings of faces Drawings of faces Geometric figures Coloring of lamp pic.
Electrical stim. Electrical stim. Loud scream Loud scream Electrical stim. Electrical stim. Electrical stim. Electrical stim. Electrical stim. Electrical stim. Electrical stim. Electrical stim. Electrical stim. Electrical stim. Electrical stim. Electrical stim. Electrical stim. Electrical stim. Electrical stim. Loud noise Loud noise Electrical stim. Electrical stim.
Virtual reality contexts Room lighting level Screen color Screen color Colors room lighting Experimental rooms Screen color + sound Room pics on screen Room pics on screen Room pics on screen Room pics on screen Room pics on screen Room pics on screen Room pics on screen Colors room lighting + sounds Fear-relevant/fear-irrelevant Real contexts pics (diff. angles) Room pics on screen Room pics on screen Light/dark room Light/dark room Light/dark room Room pics on screen
Startle reflex; SCR Online US-expectancy Online US-expectancy Online US-expectancy SCR; online US-expectancy SCR SCR; BOLD response (fMRI) SCR SCR; structural MRI SCR SCR; BOLD response (fMRI) SCR; BOLD response (fMRI) SCR SCR Online US-expectancy SCR; online US-expectancy Probed US-expectancy response times Skin conductance SCR; BOLD response (fMRI) SCR; retrospective US-expectancy SCR; retrospective US-expectancy SCR; online US-expectancy Skin conductance
Immediate Immediate Immediate Immediate Immediate Immediate/delayed Delayed Delayed Delayed Delayed Delayed Delayed Delayed Delayed Immediate Immediate Immediate Delayed Delayed Immediate Immediate Immediate Delayed
B. Vervliet et al. / Biological Psychology 92 (2013) 51–58
Study
55
56 Table 2 Overview of the critical tests of contextual renewal in the reported studies in human fear conditioning. Only the first author of the studies is mentioned. Abbreviations: WS = within-subjects test; BS = between-subjects test; SCR = skin conductance response; RT = reaction time task; CS+ = the conditioned stimulus; CS− = a nonconditioned control stimulus. TEST 2 generalization of acquisition
TEST 1 incomplete extinction (diff. renewal)
TEST 3 fear of test context
TEST 4a and 4b neutral test context
TEST 5 extinction context safety signal
TEST 6 selective tranfser
Increase CS+ versus CS−
Generalization
Decrement
Fear
Summation
ABC
AAB
Alvarez et al. (2007) Bandarian Balooch and Neumann (2011) Dibbets et al. (2008) Dibbets and Maes (2011) Effting (2007, Exp. 1)
yes yes yes (BS) – yes
yes yes yes yes –
– yes (due to CS−) yes (CS + ) yes (CS + ) yes (CS + )
yes – – – –
no – – – –
– – – – –
– – – – –
– – – – –
– – – – –
Effting (2007, Exp. 2)
yes
–
yes (CS + )
–
–
–
–
yes yes (RT, not SCR) yes See Milad et al. (2005a) – – – – – – – – – – yes yes yes (BS) yes (BS) yes (BS) yes (BS) See Milad et al. (2005a) – – yes (BS) yes (WS), no (BS) yes (WS) yes (WS) yes (WS) yes (WS)
– – yes
– – –
– – –
– – –
yes, but also for CS− – – –
–
Huff et al. (2009) Kalisch et al. (2006) Milad et al. (2005a) Milad et al. (2005b) Milad et al. (2006) Milad et al. (2007) Milad et al. (2008) Milad et al. (2009) Milad et al. (2010) Neumann et al. (2007) Neumann and Longbottom, 2008 Neumann and Kitlertsirivatana (2010) Rauch et al. (2005) Rougemont-Bücking et al. (2010) Vansteenwegen et al. (2005) Vansteenwegen et al. (2006) Vervliet et al. (2010)
yes (WS) yes (BS) – – yes (BS) [ratings, not SCR] yes (BS) [ABA, not ABC] yes (BS) yes (RT, not SCR) –
– – –
– – –
– – –
– – – – – yes yes (CS+, Exp1) yes
– – – – – – – no
– – – – – – – –
– – – – – – – –
– – – – – (partial test) – yes
– – – – – – – –
– partial test – partial test – – – –
– – – – – – – –
– no – yes (WS)
– – – –
– – – –
– – – –
– – – –
– – – –
– – – –
Zeidan et al. (in press)
–
– yes no yes (ratings, not SCR) –
–
–
–
–
–
–
–
–
B. Vervliet et al. / Biological Psychology 92 (2013) 51–58
Increase CS+
B. Vervliet et al. / Biological Psychology 92 (2013) 51–58
context, but it remains to be seen under what conditions this increase is stimulus-specific. 4.6. TEST 4b: AAB There are currently no reports of AAB-renewal tests in human fear conditioning. 4.7. TEST 5: inhibition by extinction context Milad et al. (2007, 2009) have included test presentations of a separately conditioned stimulus in the extinction context of the CS+ (see Table 2, column 9). They routinely observed higher levels of fear to this stimulus as compared to the extinguished CS+. This indicates that the extinction effect to the CS+ is not entirely due to general inhibition by the extinction context. However, it is still possible that part of the extinction effect was due to context inhibition (which would be removed during a renewal test and produce recovery). This can only be tested by comparing the level of fear to the extra conditioned stimulus in the extinction context versus a neutral context. It remains to be seen whether contextual inhibition contributes to extinction, and hence, to renewal. 4.8. TEST 6: selective transfer There are currently no reports of selective transfer tests in human fear conditioning. 4.9. Summary of the critical tests The basic renewal effect is observed in all reported studies. Overall, the results show that this effect is not simply due to incomplete extinction (TEST 1) or to summation by the test context (TEST 3 and TEST 4a). The exact mechanism, however, is still unclear. This is due to (a) the absence of some crucial tests, and (b) two patterns of results that may be at odds with Bouton’s extinction theory. The following tests are missing in human fear conditioning research on renewal: AAB-renewal (TEST 4b), a critical test for the inhibition of the extinction context (TEST 5), and tests of selective transfer (TEST 6). Especially the latter two tests are absolutely necessary for a correct evaluation of Bouton’s theory of extinction. b. A majority of the studies report a decrement of generalizationof-acquisition and/or a non-differential recovery effect at renewal test (CS+ and CS−). The first result leaves the possibility open that the renewal effect is based on a simple stimulus generalization mechanism. The second result suggests that the observed increase of fear is not a genuine ‘recovery’ (as the CS− was never fear conditioned). Nevertheless, three studies did show a “clean” renewal effect: a differential recovery after extinction, and no generalization decrement after acquisition (Bandarian Balooch and Neumann, 2011; Neumann and Kitlertsirivatana, 2010; Vansteenwegen et al., 2005). This is obviously a very small minority of the number of reported studies (24). In the next section, we provide recommendations for optimizing the renewal procedure and measurement in human fear conditioning. 5. Recommendations for future research We propose that studies on contextual renewal should routinely include analyses of the generalization of acquisition (by comparing the end of acquisition and the beginning of extinction) and the recovery of fear (by comparing the end of extinction with the beginning of test). These analyses should also include the responses to the CS−, in order to distinguish fear recovery from a general increase of fear. Possibly, an extinction retention index can be used that
57
includes the end levels of acquisition (cf. Milad et al., 2007), preferably with the inclusion of the end level of extinction as well. In addition, measuring the baseline level of fear at test can evaluate the level of contextual excitation (cf. Alvarez et al., 2007). This can be done by looking at startle reflexes to probe-alone trials (Alvarez et al.), by looking at tonic skin conductance levels (cf. Marschner et al., 2008) or by asking fear- or expectancy-ratings during contextalone intervals (cf. Fonteyne et al., 2009). Finally, the inclusion of an extra conditioned stimulus during the acquisition phase can evaluate the level of contextual inhibition, by comparing fear reactions to that stimulus in the extinction context versus a neutral context. Based on these considerations, we propose an optimized renewal procedure. This consists of an ABC-renewal design that includes an extra conditioned stimulus during acquisition in A. Next, the CS+ is extinguished in B (alternated with CS− trials). Towards the end of extinction training, the extra conditioned stimulus is presented in the extinction context B as well and subsequently, all stimuli are presented in the novel context C. Based on Bouton’s theory, we expect an increase of fear to the CS+ from B to C (TEST 1 and TEST 4a), but equal levels of fear to the extra conditioned stimulus in B and C (TEST 5). This test excludes contextual inhibition. Finally, generalization-of-acquisition (TEST 2) and contextual excitation (TEST 3) should also be evaluated. Responses to the CS− should always be included in the analyses. The test of specific transfer (TEST 6) is less stringent, because it is not aimed at refuting competing hypotheses. However, it is important if one aims to fully establish Bouton’s model of extinction in a research program. In that case, a separate experiment is needed. There is another issue that has not yet been addressed here, but that was put forward by Lovibond et al. (1984) as a critique to the research line of Bouton and colleagues. Namely, the contexts are often not equated for exposure time during the different phases in the typical ABA- and ABC-renewal designs. Hence, it may be important to alternate acquisition trials in A with context-alone presentations of B, and alternate extinction trials in B with contextalone presentations of A. For one thing, this may reduce confounds from orienting reactions when the context changes at test. For example, Neumann and Kitlertsirivatana (2010) have included this potentially important methodological control, and observed the typical renewal effects. To date, no study has addressed the influence of cognitive processes on fear renewal in humans. It is very likely that such processes (e.g., induced by verbal instructions or hindered by mental load tasks) will have an influence on the learning and/or expression of fear and extinction in humans (Lovibond, 2004). This has obvious importance for the application of fear extinction findings to patient samples. On a final note, we would like to return to the issue of the CS− and the various observations of non-differential increase of fear during renewal. It should be noted that this is not unique to the renewal phenomenon in humans; it is a quite common observation in human reinstatement studies (Dirikx et al., 2009) and spontaneous recovery (Norrholm et al., 2008). In renewal, increases to the CS− may reflect summation of contextual excitation. This is unlikely, however, based on the counter-evidence from direct tests reported in this review (Alvarez et al., 2007). Alternatively, increased responding (both SCR and startle) during CS− after a context change may reflect on dishabituation and/or sensitization (Groves and Thompson, 1970). A more interesting possibility, however, is that the CS− may not be a neutral control stimulus, but actively involved in associative learning processes throughout the acquisition phase. Namely, the CS− may acquire an inhibitory association with the US, in order to counter contextual excitation (fear) and/or generalized excitation (fear) from the CS+. The CS− now predicts the absence of the US and turns into a safety signal. The context change at renewal may disrupt this inhibition, just like it
58
B. Vervliet et al. / Biological Psychology 92 (2013) 51–58
disrupts the extinction-inhibition of the CS+. In this case, the nondifferential increase of fear does not exclude a genuine recovery effect to the CS+. Rather, the problem is that the CS− may not be the best control stimulus for an extinguished stimulus, as it may share the basic process of extinction: inhibition. The literature currently lacks direct, empirical tests of the associative status of the typical CS− in order to evaluate its relevance as a control stimulus for extinction and recovery. Such tests would inform about the status of the traditional CS− as a control stimulus for extinction effects. In summary, contextual renewal represents a major pathway to the study of extinction and fear recovery, and has been established as a robust phenomenon in human fear conditioning. The prevailing theory in animal conditioning research (Bouton, 1994) is often taken for granted in this human literature, but critical tests are currently lacking. We have summarized these tests and their rationales, thereby indicating the wealth of alternative mechanisms that may be involved in fear extinction. Routinely including these tests in renewal studies is expected to greatly expand our understanding of the fear extinction process in humans. Renewal is likely to be governed by a multitude of factors, the relative weight of which depends on the experimental paradigm, the tested sample, and the renewal design employed. A clear view on the extinction process under different circumstances has important implications for human conditioning theories, brain imaging studies, and clinical innovations on fear extinction. Acknowledgement This research was supported by University of Leuven grants GOA/2007/03 and PF/10/005. References Alvarez, R.P., Johnson, L., Grillon, C., 2007. Contextual-specificity of short-delay extinction in humans: renewal of fear-potentiated startle in a virtual environment. Learning and Memory 14, 247–253. Bandarian Balooch, S., Neumann, D.L., 2011. Effects of multiple contexts and context similarity on the renewal of extinguished conditioned behaviour in an ABA design with humans. Learning and Motivation 42, 53–63. Bouton, M.E., 1994. Conditioning, remembering, and forgetting. Journal of Experimental Psychology: Animal Behavior Processes 20, 219–231. Bouton, M.E., 2002. Sources of relapse after behavioral extinction. Biological Psychiatry 52, 976–986. Bouton, M.E., Bolles, R.C., 1979. Contextual control of the extinction of conditioned fear. Learning and Motivation 10, 445–466. Bouton, M.E., King, D.A., 1983. Contextual control of the extinction of conditioned fear: tests for the associative value of the context. Journal of Experimental Psychology: Animal Behavior Processes 9, 248–265. Bouton, M.E., Ricker, S.T., 1994. Renewal of extinguished responding in a second context. Learning & Behavior 22, 317–324. Craske, M.G., Kircanski, K., Zelikowsky, M., Mystkowsky, J., Chowdhurry, N., Baker, A., 2008. Optimizing inhibition learning during exposure therapy. Behaviour Research and Therapy 46, 5–27. Dibbets, P., Havermans, R., Arntz, A., 2008. All we need is a cue to remember: the effect of an extinction cue on renewal. Behaviour Research and Therapy 46, 1070–1077. Dibbets, P., Maes, J.R., 2011. The effect of an extinction cue on ABA-renewal: does valence matter? Learning and Motivation 42, 133–144. Dirikx, T., Vansteenwegen, D., Eelen, P., Hermans, D., 2009. Non-differential return of fear in humans after a reinstatement procedure. Acta Psychologica 130, 175–182. Effting, M., Kindt, M., 2007. Contextual renewal of human fear associations in a renewal paradigm. Behaviour Research and Therapy 45, 2002–2018. Fonteyne, R., Vervliet, B., Hermans, D., Baeyens, F., Vansteenwegen, D., 2009. Reducing chronic anxiety by making the threatening event predictable: an experimental approach. Behaviour Research and Therapy 47, 830–839. Groves, P.M., Thompson, R.F., 1970. Habituation: a dual-process theory. Psychological Review 70, 49–450. Harris, J.A., Jones, M.L., Bailey, G.K., Westbrook, F., 2000. Contextual control over conditioned responding in an extinction paradigm. Journal of Experimental Psychology: Animal Behavior Processes 26, 174–185. Huff, N.C., Hernandez, J.A., Blanding, N.Q., La Bar, K.S., 2009. Delayed extinction attenuates conditioned fear renewal and spontaneous recovery in humans. Behavioral Neuroscience 123, 834–843.
Kalisch, R., Korenfeld, E., Stephan, K.S., Weiskopf, N., Seymour, B., Dolan, R.J., 2006. Context-dependent human extinciton memory is mediated by a ventromedial prefrontal and hippocampal network. The Journal of Neuroscience 26, 9503–9511. Konorski, J., 1967. Integrative Activity in the Brain: An Interdisciplinary Approach. University Press of Chicago, Chicago, IL. Lipp, O.V., 2006. Human fear learning: contemporary procedures and measurement. In: Craske, M.G., Hermans, D., Vansteenwegen, D. (Eds.), Fear and Learning: From Basic Processes to Clinical Implications. American Psychological Association, Washington, DC, US, pp. 37–51. Lissek, S., et al., 2008. Generalization of conditioned fear-potentiated startle in humans: experimental validation and clinical relevance. Behaviour Research and Therapy 46, 678–687. Lovibond, P.F., 2004. Cognitive processes in extinction. Learning and Memory 11, 495–500. Lovibond, P.F., Preston, G.C., Mackintosh, N.J., 1984. Context specificity of conditioning, extinction, and latent inhibition. Journal of Experimental Psychology: Animal Behavior Processes 10, 360–375. Marschner., A., Kalisch, R., Vervliet, B., Vansteenwege, D., Büchel, C., 2008. Dissociable roles for the hippocampus and the amygdala in human cued versus context fear conditioning. The Journal of Neuroscience 28, 9030–9036. Milad, M.R., Orr, S.P., Pitman, R.K., Rauch, S.L., 2005a. Context modulation of memory for fear extinction in humans. Psychophysiology 42, 456–464. Milad, M.R., Quinn, B.T., Pitman, R.K., Orr, S.P., Fischl, B., Rauch, S.L., 2005b. Thickness of ventromedial prefrontal cortex in humans is correlated with extinction memory. In: Proceeding of the National Academy of Sciences, 102, pp. 10706–10711. Milad, M.R., et al., 2006. Fear conditioning and extinction: influence of sex and menstrual cycle in healthy humans. Behavioral Neuroscience 120, 1196–1203. Milad, M.R., Wright, C.I., Orr, S.P., Pitman, R.K., Quirk, G.J., Rauch, S.L., 2007. Recall of fear extinction in humans activates the ventromedial prefrontal cortex and hippocampus in concert. Biological Psychiatry 62, 446–454. Milad, M.R., Orr, S.P., Lasko, N.B., Chang, Y., Rauch, S.L., Pitman, R.K., 2008. Presence and acquired origin of reduced recall for fear extinction in PTSD: results of a twin study. Journal of Psychiatry Research 42, 515–520. Milad, M.R., et al., 2009. Neurobiological basis of failure to recall extinction memory in posttraumatic stress disorder. Biological Psychiatry 66, 1075–1082. Milad, M.R., et al., 2010. The influence of gonadal hormones on conditioned fear extinction in healthy humans. Neuroscience 168, 652–658. Myers, K.M., Davis, M., 2002. Behavioral and neural analysis of extinction. Neuron 36, 567–584. Neumann, D.L., Kitlertsirivatana, E., 2010. Exposure to a novel context after extinction causes a renewal of extinguished conditioned responses: implications for the treatment of fear. Behaviour Research and Therapy 48, 565–570. Neumann, D.L., Lipp, O.V., Cory, S.E., 2007. Conducting extinction in multiple contexts does not necessarity attenuate the renewal of shock expectancy in a fear-conditioning procedure with humans. Behaviour Research and Therapy 45, 385–394. Neumann, D.L., Longbottom, P.L., 2008. The renewal of extinguished conditioned fear with fear-relevant and fear-irrelevant stimuli by a context change after extinction. Behaviour Research and Therapy 46, 188–206. Norrholm, S.D., et al., 2008. Timing of extinction relative to acquisition: a parametric analysis of fear extinction in humans. Behavioral Neuroscience 122, 1016–1030. Pavlov, I.P., 1927. Conditioned Reflexes: An Investigation of the Physiological Activity of the Cerebral Cortex. Oxford University Press, London. Pearce, J.M., 1987. A model for stimulus generalization in pavlovian conditioning. Psychological Review 94, 61–73. Quirk, G.J., Mueller, D., 2008. Neural mechanisms of extinction learning and retrieval. Neuropsychopharmacology 33, 56–72. Rauch, S.L., Milad, M.R., Orr, S.P., Quinn, B.T., Fischl, B., Pitman, R.K., 2005. Orbitorfrontal thickness, retention of fear extinction, and extraversion. Cognitive Neuroscience and Neuropsychology 16, 1909–1912. Rescorla, R.A., Wagner, A.R., 1972. A theory of pavlovian conditioning: variations in the effectiveness of reinforcement and nonreinforcement. In: Black, A.H., Prokasy, W.F. (Eds.), Classical Conditioning II. Appleton Century Crofts, New York, pp. 64–99. Rougemont-Bücking, A., et al., 2010. Altered processing of contextual information during fear extinction in PTSD: an fMRI study. CNS Neuroscience and Therapeutics 00, 1–10. Vansteenwegen, D., Hermans, D., Vervliet, B., Francken, G., Beckers, T., Baeyens, F., Eelen, P., 2005. Return of fear in a human differential conditioning paradigm caused by a return to the original acquisition context. Behaviour Research and Therapy 43, 323–336. Vansteenwegen, D., Vervliet, B., Hermans, D., Beckers, T., Baeyens, F., Eelen, P., 2006. Stronger renewal in human fear conditioning when tested with an acquisition retrieval cue than with an extinction retrieval cue. Behaviour Research and Therapy 44, 1717–1725. Vervliet, B., 2008. Learning and memory in conditioned fear extinction: effects of d-cycloserine. Acta Psychologica 127, 601–613. Vervliet, B., Vansteenwegen, D., Hermans, D., 2010. Unpaired shocks during extinction weaken the contextual renewal of a conditioned discrimination. Learning and Motivation 41, 22–31. Zeidan, M.A., et al., in press. Test–retest reliability during fear acquisition and fear extinction in humans. CNS Neuroscience and Therapeutics.
