Massive extinction treatment attenuates the renewal effect

Two experiments with rats as subjects investigated whether massive extinction can attenuate the renewal effect. Experiment 1 investigated whether moderate or massive extinction could prevent the return of conditioned responding following Pavlovian conditioning in Context A, extinction in Context B, and subsequent testing in Context C (i.e., ABC renewal). Experiment 2 examined whether massive extinction could prevent renewal following training in Context A, extinction in Context B, and testing in Context A (i.e., ABA renewal). Both experiments observed attenuated renewal following massive, but not moderate extinction. Results are discussed in terms of contemporary theories of extinction. The processes responsible for the experimental extinction of Pavlovian associations have been a focal point of researchers for many years (e.g., Pavlov, 1927). Numerous researchers have reported that, following experimental extinction, behavior indicative of extinction (i.e., weak conditioned responding) tends to be restricted to the context in which the extinction treatment was implemented (e.g., Bouton & Bolles, 1979; Bouton & Ricker, 1994; Chelonis, Calton, Hart, & Schachtman, 1999; Gunther, Denniston, & Miller, 1998; Lovibond, Preston, & Mackintosh, 1984; Rauhut, Thomas, & Ayres, 2001). This observation which is suggestive of the context specificity of extinction, has been termed the renewal effect and has attracted a great deal of attention due to its potential to illuminate the mechanisms underlying experimental extinction and because of its clinical implications (i.e., that exposure-based therapies might be restricted to the place in which the therapy is provided; Bouton and Bouton; Rodriguez, Craske, Mineka, & Hladek, 1999). The renewal effect has been demonstrated using three different procedures. One form of renewal occurs when acquisition training is provided in Context A, extinction treatment in Context B, and subsequent behavioral testing in Context C (i.e., ABC Renewal; Bouton & Bolles, 1979). Renewal in an ABC preparation is evidenced by enhanced conditioned responding to the previously extinguished conditioned stimulus (CS) when testing is conducted in Context C, relative to subjects that were tested in the context used for extinction treatment (e.g., Context B). Another form of renewal is that obtained with an ABA design, in which acquisition training is conducted in Context A, extinction treatment in Context B, and behavioral testing in Context A, relative to subjects that receive training, extinction, and testing in the same context (i.e., AAA, Bouton & Bolles). The third form of renewal is termed AAB renewal, in which acquisition and extinction treatments are both provided in Context A, with subsequent behavioral testing in Context B, relative to an AAA control condition ( Bouton & Ricker, 1994). Enhanced conditioned responding to the test CS in Context B demonstrates this form of the renewal effect. One difference between these procedures is the strength of the return of conditioned responding observed at test (as evidenced by cross-experiment comparisons). Specifically, ABA renewal tends to be more robust than either ABC or AAB renewal ( Bouton, 1991; Bouton & Bolles, 1979; Bouton & King, 1983; Bouton & Swartzentruber, 1989; Tamai & Nakajima, 2000). The mechanism underlying the renewal effect has been a focus of research for several years. One possible explanation of the renewal effect is that the extinction context becomes a conditioned inhibitor as a consequence of the nonreinforced presentations of the excitatory CS in the context (i.e., a form of Pavlovian inhibition). However, numerous studies have failed to find that the extinction context can pass summation and retardation tests for conditioned inhibition (e.g., Bouton & King, 1983; Bouton & Swartzentruber, 1986). An alternative interpretation of the renewal effect is that the extinguished CS gains inhibitory strength during extinction. Supportive of such a view are recent studies that have found that massive extinction allows an extinguished CS to pass both summation and retardation tests for conditioned inhibition (e.g., Denniston & Miller, in press; Hart, Bourne, & Schachtman, 1995; but see Rauhut et al., 2001, for an exception). Toward explaining the context specificity of extinction (inhibition), Bouton and colleagues ( Bouton, 1994; Bouton & Nelson, 1994; Bouton & Swartzentruber, 1986) have suggested that the CS becomes ambiguous following extinction treatment as a consequence of having signaled both reinforcement and nonreinforcement, and that the context acts as an occasion setter which disambiguates the meaning of the CS. That is, acquisition training establishes an excitatory CS-unconditioned stimulus (US) association, whereas extinction treatment establishes an inhibitory CS–US association (i.e., CS–noUS). This latter association is modulated by the context in which it was acquired, whereas the former excitatory association is not typically modulated by context (see Nelson, 2002, who found that the association that is learned second is relatively context specific). In other words, the context functions as an “AND” gate, in which the joint presence of the CS and the extinction context are necessary for the CS–noUS association to be expressed. When testing is conducted outside of the extinction treatment context, the absence of the context results in decreased activation of the CS–noUS association, thereby allowing for full expression of the excitatory CS–US association (i.e., renewal). The observation that extinction is relatively context specific is consistent with the view that the extinction context functions as a negative occasion setter for the CS–US association, rather than as a conditioned inhibitor. Recent research has identified circumstances that limit the context specificity of experimental extinction. For example, Gunther et al. (1998) found that, following conditioning in Context A and extinction treatment in Contexts B, C, and D, renewal of conditioned responding was attenuated when testing was conducted in associatively neutral Context E (note that this is effectively an ABC design). Similar results using a conditioned taste aversion preparation were reported by Chelonis et al. (1999) who found that following taste aversion training in Context A and extinction in Contexts B, C, and D, renewal was attenuated when testing was conducted in Context A (this was effectively an ABA design). According to the contextual occasion setting account, behavior indicative of extinction would be expected to be restricted to the extinction context. In contrast, weak conditioned responding (indicative of generalization of extinction) was observed by both Gunther et al., who tested in an associatively neutral context (i.e., Context E) and Chelonis et al. who provided testing in the context in which acquisition training had been provided (i.e., Context A). Gunther et al. suggested that generalization of elements from the three extinction contexts (B, C, and D) to the test context might result in generalization of the tendency to not respond to the CS, as the test context is likely to have more stimulus elements in common with the three extinction contexts than with the single acquisition context. Such a view is consistent with the contextual occasion setting interpretation provided by Bouton and his colleagues (e.g., Bouton, 1993; Bouton & Nelson, 1994; Bouton & Swartzentruber, 1986). However, Chelonis et al. explained their results in terms of encoding variability, in which interference by the CS–noUS association is enhanced following extinction in multiple contexts relative to extinction in a single context, thereby facilitating retrieval of the CS–noUS association outside of the extinction contexts. In a separate line of research, Denniston and Miller (in press) investigated the informational content of inhibitory associations produced through experimental extinction. Their studies found that a massively extinguished CS passed traditional summation and retardation tests for conditioned inhibition and that the passage of these tests depended upon an equivalency between the temporal relationship between the CS and the US during acquisition training and the temporal relationship between the transfer excitor and the US. They claimed that the potential of an extinguished CS to pass summation and retardation tests for conditioned inhibition depended (in part) upon the retrievability of the inhibitory CS–US association. That is, extinction treatment might favor retrieval of the inhibitory CS–US association at the expense of the excitatory CS–US association. When the temporal information content of the inhibitory association matched the information content of the excitatory transfer CS, maximal negative summation was observed. These results suggest that massive extinction treatment might provide another means for favoring retrieval of the inhibitory CS–US association established during experimental extinction. The present series of experiments investigated whether massive extinction treatment can attenuate the renewal effect. Tamai and Nakajima (2000) found that massive, but not moderate, extinction treatment attenuated AAB, but not ABA renewal. In their experiment, Tamai and Nakajima provided rats with various magnitudes of extinction treatments (ranging from 32 to 112 extinction trials) following 24 CS–US acquisition trials, and observed both ABA and AAB renewal after 72 extinction trials, but only ABA renewal following 112 extinction trials. Thus, AAB renewal was attenuated following prolonged extinction treatment. They interpreted their results as being consistent with Bouton and Ricker’s (1994) view that the extinction context functions as a negative occasion setter for the inhibitory CS–US association and that conditioned responding is more likely to be “released” when testing is conducted outside of the extinction context. Tamai and Nakajima hypothesized that the contextual gating might weaken following prolonged extinction, thereby providing an explanation for the attenuation of AAB renewal. Toward explaining the preservation of ABA renewal following prolonged extinction treatment, they suggested that the acquisition context (A) can function as a “retrieval cue” for the excitatory CS– US association. This retrieval cue was present during testing for subjects in their ABA renewal group, but not the AAB renewal group, which can potentially explain why renewal was observed in the former group even after prolonged extinction treatment. The present series of experiments was conceptually similar to that of Tamai and Nakajima’s (2000) experiment. However, we investigated whether massive extinction treatment could attenuate ABA and ABC renewal, rather than ABA and AAB renewal. We anticipated that massive extinction treatment provided in a single context should favor retrieval of the inhibitory CS–US association outside of the extinction context. These anticipated results stand in contrast to the view that the context specificity of extinction is the consequence of occasion setting by context (i.e., Bouton, 1993; Bouton & Nelson, 1994; Bouton & Swartzentruber, 1986) and facilitated retrieval of excitation by the acquisition context (Tamai & Nakajima). Indeed, the occasion setting by context view might anticipate that massive extinction treatment should enhance the potential of the context to serve as a negative occasion setter (provided that the potential of a stimulus to serve as a negative occasion setter is monotonically related to number of training trials), thereby enhancing the context specificity of extinction, a result opposite of what we expected (see Tamai & Nakajima for a discussion of this possibility). Notably, other researchers have failed to observe an attenuation of renewal with increased extinction experience (e.g., Bouton & Swartzentruber, 1989; Rauhut et al., 2001; Tamai & Nakajima, 2000). However, Bouton and Swartzentruber, who provided 84 nonreinforced exposures to a 1-min CS, did not include a group that received fewer extinction trials (notably, the effect of magnitude of extinction treatment was not the focus of their research). Rauhut et al., used an ABA renewal procedure and did not equate exposure to the extinction and testing contexts, leaving open the possibility that their failure to observe an attenuation of the renewal effect following 100 2-min exposures to the CS was the result of residual fear of the test (acquisition) context. Although preCS response rates did not differ, subthreshold fear might have influenced their results. Additionally, Rauhut et al.’s assessment of renewal consisted of a comparison of responding to the extinguished CS during the test session in Context A to that observed during the final extinction session (two days earlier) in Context B, which might have allowed for some spontaneous recovery to occur. Although Rauhut et al. argued against the potential of spontaneous recovery to undermine their results, this possibility combined with the failure to equate exposure to the extinction and test contexts calls for a more systematic analysis of whether massive extinction might attenuate renewal. The present series of experiments sought to investigate whether truly massive extinction treatment (nearly eight times the amount provided by Tamai & Nakajima) can attenuate the renewal effect while avoiding some of the limitations cited above. EXPERIMENT 1: ABC RENEWAL Experiment 1 investigated whether massive extinction could prevent the return of conditioned responding observed following Pavlovian training in one context (Context A), experimental extinction in a second context (Context B), and behavioral testing in a third associatively neutral context (Context C; i.e., ABC renewal). Four groups of rats received training intended to condition fear to a CS, X, in Context A, followed by either massive, moderate, or no extinction experience in Context B (see Table 1). Specifically, following fear conditioning in Context A, Group Ext-Many received 800 nonreinforced presentations of CS X in Context B, Groups ExtMod and Ext-Mod-B received 160 nonreinforced presentations of CS X in Context B, and Group NoExt merely received equivalent exposure to Context B. Following extinction treatment, the potential of CS X to disrupt baseline drinking rates was assessed in an associatively neutral context for Groups NoExt, Ext-Mod, and Ext-Many, or in Context B for Group Ext-Mod-B. If massive extinction enhances the retrievability of CS–noUS associations, rather than enhancing occasion setting of CS X by the extinction context, then massive extinction should attenuate the renewal effect in Group Ext-Many, relative to Group Ext-Mod. The purpose of including Group Ext-Mod-B was to demonstrate that the magnitude of extinction provided in Group Ext-Mod was sufficient to attenuate conditioned responding when testing is conducted in the extinction context (Context B). Table 1. Design Summary for Experiment 1 Note. CS X was a 10-s white noise; US was a 0.5-s, 1.0-mA footshock. Subscript letters (A, B, and C) refer to contexts. During Phase 2, subjects received two sessions per day, one in Context B and the other in Context C, counterbalanced for order. () represents context exposure with no nominal stimulus presentations. Acclimation (not shown) and Reacclimation sessions in Contexts B and C were provided on separate days prior to both Phase 1 and Testing. Subjects in Group Ext-Mod-B were tested for conditioned responding to CS X in Context B. CR = strong responding expected; CR = weak responding expected.