Abstract

1. Introduction

A major challenge to behavioral therapies for anxiety disorders, including post-traumatic stress disorder, is the relapse of fear that occurs outside of the clinical context. One account of relapse builds on the observation that the extinction of Pavlovian conditioned responses (CRs), including fear responses, is context-specific (Bouton & Bolles, 1979; Bouton, Mineka, & Barlow, 2001; Craske et al., 2008). For example, if a conditional stimulus (CS) that once predicted an aversive footshock unconditional stimulus (US) is presented alone several times in a distinct context, fear to the CS will be lost—but only in the extinction context (Bouton & Bolles, 1979). That is, the fear CR will return or renew’ if the CS is encountered outside of the extinction context, including the conditioning context (‘ABA’ renewal, where the letters denote the contexts used for conditioning, extinction, and testing) or an equally familiar context in which the CS has never been experienced (‘AAB’ or ABC renewal’, which is often found to be weaker than ABA renewal). This reveals that extinction does not erase the CS-US association but encourages the formation of a new inhibitory CS- no US’ memory (Konorski, 1967; Pavlov, 1927). It has been posited that these inhibitory memories are particularly context-bound, and that animals use contexts as retrieval cues to determine how to respond to a CS (Bouton, 1993; Bouton, Westbrook, Corcoran, & Maren, 2006). By this view, inhibition acquired during extinction (e.g., the formation of a CS- no US’ association) is linked to the extinction context, and is not retrieved outside of that context. In other contexts, fear is supported by the context-independent expression of the CS-US association—as a result, conditioned fear is broadly generalized, whereas extinction is not.

Another possibility is that the inhibition acquired during extinction fails to generalize to other contexts because the circumstances of the test situation produce “inhibition of an inhibition” (Pavlov, 1927). Pavlov observed that novel or unexpected events (including his own presence during a colleague’s experiment) caused a spontaneous return of an extinguished salivary response, a phenomenon he termed external disinhibition’(Pavlov, 1927). The loss of inhibition that occurs when novel stimuli accompany an extinguished CS might account for other forms of CR recovery that are accompanied by unexpected events, including renewal. In this case, extinguished CSs are unexpectedly encountered in contexts in which they had never been presented (AAB, ABC), or had not been encountered recently (ABA). Hence, renewal may be a form of disinhibition produced by associative novelty when testing occurs in a familiar context. This account has found recent support in a computational model of extinction that includes attentional mechanisms and novelty in determining conditioned responding (Larrauri & Schmajuk, 2008).

Interestingly, there is considerable evidence in both rats and humans that the hippocampus plays an important role in novelty detection, particularly in the detection of associative novelty (Honey & Good, 2000; Knight, 1996; Kumaran & Maguire, 2006; Vinogradova, 2001). For example, hippocampal lesions in rats reduce the orienting response to the novel pairing of habituated stimuli, a phenomenon that requires the detection of an associative mismatch among familiar cues (Honey, Watt, & Good, 1998). In addition, human neuroimaging experiments have revealed that hippocampal activity is elevated under conditions in which unexpected sequences of familiar stimuli are encountered (Kumaran & Maguire, 2006; Wessel, Danielmeier, Morton, & Ullsperger, 2012). Moreover, it is now apparent that the hippocampus and its afferents are critical for the renewal of extinguished fear that occurs when an extinguished is unexpectedly encountered outside the extinction context (Corcoran, 2004; Corcoran & Maren, 2001; Corcoran, Desmond, Frey, & Maren, 2005; Hobin, Ji, & Maren, 2006; Ji & Maren, 2005; 2008a; 2008b; Maren & Hobin, 2007; Zelikowsky, Pham, & Fanselow, 2011). While most authors have interpreted these hippocampal lesion effects in terms of contextual retrieval processes and occasion setting (Maren, 2005; Maren & Holt, 2000), a more parsimonious account appeals to the role of the hippocampus in novelty detection. That is, the hippocampus may mediate the renewal effect, because it is detects associative novelty during the retrieval test that leads to the disinhibition of extinguished fear. If true, hippocampal inactivation should limit the return of extinguished fear that accompanies exposure to a novel stimulus (external disinhibition) or presentation of the extinguished CS in a novel context (renewal). To examine this question, we examined the effect of hippocampal inactivation with the GABA_A agonist muscimol on the expression of fear to an extinguished CS presented in either a novel context (Experiment 1) or after delivery of a novel stimulus (white noise) in the extinction context (Experiment 2). Insofar as the hippocampus is required for detecting novel or unexpected events, we hypothesized that hippocampal inactivation would impair the return of fear driven by the presence of novel contexts or cues.

2. Materials and methods

3. Results

3.1 Histology

A representative cannula placement in the dorsal hippocampus is illustrated in Figure 1. Placements in both Experiments 1 and 2 were similar to those reported in our previously published work using this methodology. None of the rats in Experiment 1were excluded from analysis based on their cannula placements, and two rats were excluded in Experiment 2 because their cannula were placed ventral to the target.

Open in a separate window

Figure 1

Photomicrograph showing a thionin-stained coronal section from the brain of a rat with representative cannula placements in the dorsal hippocampus.

3.2 Experiment 1

Conditioned freezing during the conditioning, extinction, and test sessions is shown in Figure 2. On the conditioning day, freezing behavior in the 1-min block after each trial increased significantly across the five conditioning trials (F _(7,420) = 312.4, p < 0.0001); there were no differences between the groups. For the extinction sessions, freezing was averaged across all 30 post-CS blocks (1-min blocks) and plotted for each of the three extinction sessions, which were conducted over three days. Conditioned freezing declined significantly across the extinction sessions (F _(2,120) = 133.2, p < 0.0001). Although the terminal level of fear suppression was similar in all groups, the context in which rats were extinguished produced significantly different levels of fear (main effect of context, F _(1,60) = 23.5, p < 0.0001) and influenced extinction rate (context x day interaction, F _(2,120) = 6.2, p < 0.005). Animals that were extinguished in the conditioning context (here designated as Novel’ groups, because of their ultimate test context in theAAB procedure) exhibited higher levels of freezing than those extinguished in a novel context (here designated as Same’ groups’, because of their ultimate test context in the ABB procedure). This difference in the contextual specificity of conditioning expressed during the first extinction session was unexpected, as we have previously observed similar levels of conditioned freezing in the two contexts (Corcoran, 2004; Corcoran et al., 2005; Corcoran & Maren, 2001). Nonetheless, all animals reached a similar level of freezing at the end of the extinction phase.

Open in a separate window

Figure 2

Hippocampal inactivation disrupts renewal of fear in a novel context (Experiment 1). (Left panel) Mean (±SEM) percentage of freezing on the conditioning day. Data are 1-min averages for a 3-min baseline period (BL) and after each of the five tone-shock conditioning trials. (Middle panel) Freezing behavior during the extinction sessions. Freezing was averaged across the 30-trials extinction session. Extinction was conducted in the conditioning context for the ‘Novel’ groups (AAB design) and a novel context for the ‘Same’ groups (ABB design). Hence, the group labels refer to the treatment conditions that would later be imposed during retrieval testing (not during conditioning or extinction). (Right panel) Freezing behavior during the retrieval test. Data are 1-min averages for a 2-min baseline period (BL) and after each of the eight CS-alone trials. Rats were tested in either the extinction context (SAME) or in a novel context (NOVEL).

After extinction, rats were infused with muscimol or saline and tested either in the extinction context (‘Same’ groups) or in a novel context (‘Novel’ groups). As shown in the last panel of Figure 2, saline-treated rats exhibited a robust renewal of fear to the extinguished CS when tested in the novel context. Interestingly, this effect was eliminated in rats receiving muscimol infusions into the DH. A comparison of CS-evoked freezing (an average of freezing during the 8 post-CS intervals minus the pre-CS baseline) supported this conclusion (Figure 3). An ANOVA with variables of context and drug revealed significant main effects of context (F _(1,60) = 12.4, p < 0.005), drug (F_(1,60) = 4.3, p < 0.05), and a significant interaction of the two variables (F _(1,60) = 10.7, p < 0.002). The significant interaction reveals that saline-treated rats exhibited renewal of fear in the novel context, whereas muscimol-treated rats did not. It should be noted that in this experiment, muscimol elevated baseline (pre-CS) freezing. We have not observed this in other published work (Corcoran, 2004; Corcoran et al., 2005; Corcoran & Maren, 2001), including Experiment 2. We do not believe, however, that the shifted baseline mitigated the renewal effect insofar as there was ample room for muscimol-treated rats to increase freezing behavior to the CS. It should also be noted that this design confounds context familiarity with context change; that is, animals in the NOVEL condition experience both a novel context and a change in context in between the extinction and retention testing. However, we have previously observed that context change alone is not sufficient for fear renewal in within-subjects designs in which animals maintain extinguished responding despite context shifts (Hobin, Goosens, & Maren, 2003).

Open in a separate window

Figure 3

Hippocampal inactivation disrupts renewal of fear in a novel context (Experiment 1). Mean (±SEM) percentage of freezing during the retrieval test. Data are averages of eight post-CS blocks minus the average of the pre-CS baseline (as shown in the last panel of Figure 2). Rats were tested in either the extinction context (SAME; open bars) or in a novel context (NOVEL; gray bars). Dorsal hippocampal inactivation with muscimol (MUS) impaired the renewal of conditioned freezing to the CS in a novel context compared to saline-treated (SAL) rats.

3.3 Experiment 2

The procedures for Experiment 2 were similar to those of Experiment 1, except that all animals were conditioned, extinguished and, tested using an ABB procedure. We used this procedure reduce the contribution of contextual fear to our assessment of cue-evoked fear during the test. Conditioned freezing during the conditioning, extinction, and test sessions is shown in Figure 4. On the conditioning day, freezing behavior in the 1-min block after each trial increased significantly across the five conditioning trials (F_(7,175) = 153.8, p < 0.0001); there were no differences between the groups. For the extinction sessions, freezing was averaged across all 30 post-CS blocks (1-min interstimulus intervals after each CS) and plotted for each of the three extinction sessions, which were conducted over three days. Conditioned freezing declined significantly across the extinction sessions (F _(2,50) = 72.5, p < 0.0001). All animals reached a similar level of freezing at the end of the extinction phase.

Open in a separate window

Figure 4

Hippocampal inactivation disrupts renewal of fear by a novel white noise (Experiment 2). (Left panel) Mean (±SEM) percentage of freezing on the conditioning day. Data are 1-min averages for a 3-min baseline period (BL) and after each of the five tone-shock conditioning trials. (Middle panel) Freezing behavior during the extinction sessions. Freezing was averaged across all 30 trials of each extinction session. Extinction was conducted in a novel context for the all groups (ABB design). The group labels refer to the treatment conditions that would later be imposed during retrieval testing (not during conditioning or extinction). (Right panel) Freezing behavior during the retrieval test. Data are 1-min averages for the 2-min baseline period (BL) and after each of the eight CS-alone trials. Rats were tested in the extinction context and the CS was either preceded on each trial by a novel white noise (N+CS) or was presented alone (CS).

After extinction, rats were infused with muscimol or saline and tested in the extinction context. In half of the animals, each CS test trial was immediately preceded by a novel white noise (N+CS), whereas the other half of the animals received the CS alone (CS). As shown in the last panel of Figure 4, presentation of the novel white noise greatly increased fear to the CS in saline-treated rats as indicated by increased freezing in the post-CS blocks (1-min blocks) in CS+N relative to CS groups. Interestingly, this effect was also apparent in rats receiving muscimol infusions into the DH. A comparison of CS-evoked freezing (an average of freezing during the 8 post-CS intervals minus the pre-CS baseline) supported this conclusion (Figure 5A). An ANOVA with variables of cue and drug revealed a significant main effect of cue (F _(1,25) = 27.3, p < 0.0001, a non-significant main effect of drug and a non-significant drug x cue interaction (Fs < 1.6). This indicates that over the duration of the test, the novel white noise augmented conditioned freezing to the CS relative to the CS alone.

Open in a separate window

Figure 5

Hippocampal inactivation disrupts renewal of fear in a novel context (Experiment 1). (A) Mean (±SEM) percentage of freezing during the retrieval test. Data are averages of eight post-CS blocks minus the average of the pre-CS baseline (as shown in the right panel of Figure 4). Rats were tested in the extinction context and the CS was either preceded on each trial by a novel white noise (N+CS) or was presented alone (CS). Dorsal hippocampal inactivation with muscimol (MUS) did not impair the renewal of conditioned freezing by a novel stimulus. (B) Mean (±SEM) percentage of freezing during first test trial. Freezing data are 10-sec averages for the pre-trial period, the white noise stimulus, and the tone CS (an equivalent stimulus-free period is shown for CS rats that did not receive white noise presentations). Hippocampal muscimol infusions attenuated the renewal of fear elicited by the novel white noise on the first test trial.

However, close inspection of the test data in Figure 4 reveals that muscimol-treated rats exhibited a transient impairment in the noise-induced augmentation of freezing behavior during the first post-CS period. To understand this effect more completely, we performed an analysis of freezing behavior during 10-sec intervals immediately before the trial (baseline), during the white noise stimulus in the subsequent 10-sec interval, and during the first CS. For animals in the CS groups that did not receive white noise, we used the 10-sec stimulus free periods that corresponded to the interval sampled during the noise in the N+CS groups. These data are shown in Figure 5B. Saline-treated rats exhibited low levels of freezing before the first trial, and responded to the novel noise and consequent CS with robust freezing behavior. Freezing to the CS was substantially greater than that to the noise, although whether this freezing reflects fear to the CS, the preceding noise, or both cannot be determined with this design. Moreover, it is possible that generalized fear to the noise augments freezing to the CS, an effect that might occur along with disinhibition. Notably, muscimol-treated rats exhibited a much smaller increase in freezing after presentation of the novel white noise, and fear did not increment when the subsequent CS was delivered. Animals that did not receive white noise stimuli exhibited low levels of freezing during each of these intervals. These observations were confirmed in repeated measures ANOVA with between-subjects variables of drug and cue, and a within-subjects variable of interval. This analysis revealed a main effect of cue (F _(1,25)= 28.5, p < 0.0001), a significant drug x cue interaction (F _(1,25) = 7.0, p < 0.02), and a significant three-way interaction (F _{(2, 50)} = 4.1, p < 0.05). Hence, muscimol inactivation of the dorsal hippocampus impaired the disnhibition of freezing to an extinguished CS by a novel white noise, although this effect was transient, lasting only a single trial. This pattern of deficit is not entirely unexpected, particularly because renewal phenomena can be short-lived and therefore may be most sensitive to disruption early in retrieval testing (but see (Zelikowsky et al., 2011). It is also conceivable that muscimol inactivation of the hippocampus disrupted generalized fear to the noise that may have promoted consequent CS freezing.

4. Discussion

The present results reveal that inactivation of the dorsal hippocampus attenuates the return of fear to an extinguished CS in a novel context (Experiment 1) and when the CS is accompanied by a novel stimulus in the extinction context (Experiment 2). Interestingly, the hippocampal impairment differed in severity in the two experiments, and was relatively transient when a novel stimulus renewed fear. Together, these results add to the growing list of studies pointing to an important role of the hippocampus in regulating the expression of fear extinction. Moreover, the present results suggest that the role of the hippocampus in regulating fear after extinction might be understood in terms of mnemonic functions that are used in predicting future events based upon past experience (Buckner, 2010; Cohen & Eichenbaum, 1995; Honey & Good, 2000). Here we propose that the hippocampus is essential for detecting unexpected events, and unexpected events, at least in the present case, restore conditioned fear to an extinguished CS.

The first idea in this proposition—that the hippocampus is essential for detecting unexpected events—was first proposed by Vinogradova, who pioneered studies of hippocampal single-unit responses during orienting responses to novel stimuli (Vinogradova, 2001). Orienting responses are driven by the absence of information about a stimulus in memory (Sokolov, 1963), and Vinogradova argued that detecting novel stimuli rests on recognition memory processes mediated by the hippocampus. In addition to detecting stimulus novelty, the hippocampus is also widely believed to detect novel arrangements of familiar stimuli, so-called “associative novelty” as well as familiar stimuli in novel places (“spatial novelty”) (Gabriel, Sparenborg, & Stolar, 1987; Honey et al., 1998; Jeewajee, Lever, Burton, O’Keefe, & Burgess, 2008; Kaye & Pearce, 1984; Kumaran & Maguire, 2006). For example, Honey and colleagues have shown that hippocampal lesions in rats prevented orienting responses to familiar and physically identical visual cues whose relationship to a familiar auditory signal had changed (Honey et al., 1998). Interestingly, hippocampal damage did not affect orienting responses to stimulus novelty per se, or mismatches in which the spatial and temporal properties of the stimulus changed; hippocampal lesions appeared to have a selective role in associative novelty.

In the behavioral experiments reported here, a familiar, extinguished CS is presented in a novel place (Experiment 1) or after a novel cue in a familiar place (Experiment 2). We found that control rats exhibited a return of conditioned responding to the CS in both cases. In the first case, the return of fear might be understood as an example of “renewal”, in which fear returns outside of the extinction context. Current theories of renewal (Bouton, 1993; Pearce & Bouton, 2001) posit that the extinction context possesses occasion setting properties that modulate the response to the extinguished CS presumably by activating an inhibitory CS- no US’ association learned anew in that context (Konorski, 1967; Pavlov, 1927). Outside of that context, conditioned responding to the CS returns because the CS-US memory is not context bound, although the conditioning memory may become more context-bound after extinction (Harris, Jones, Bailey, & Westbrook, 2000). In essence, this view assumes that the renewal of fear results from the failure of the current context (whether novel or familiar) to retrieve the inhibitory extinction memory. Lesions or inactivation of the hippocampus or its afferents prevent renewal of conditioned fear (Corcoran, 2004; Corcoran et al., 2005; Corcoran & Maren, 2001; Hobin et al., 2006; Ji & Maren, 2005; 2008a; Orsini, Kim, Knapska, & Maren, 2011), an effect we have attributed to a loss of contextual control of the inhibitory extinction memory (Maren, 2005; 2011; Maren & Quirk, 2004; Orsini & Maren, 2012). In other words, performance in animals with hippocampal damage appears to reflect the sum of the associative strengths of the excitatory CS-US and inhibitory CS- no US’ memories, which might be anticipated by theories positing an important role for the hippocampus in indexing associations to the contexts in which they occur (Hirsh, 1974; Maren & Holt, 2000). Consistent with this perspective, there is evidence that negative occasion setting requires the hippocampus (Holland, Lamoureux, Han, & Gallagher, 1999), although not all forms of occasion setting are hippocampal-dependent (Holland & Bouton, 1999; Yoon, Graham, & Kim, 2011).

Despite the substantial evidence for the involvement of occasion setting in extinction, a more parsimonious account for renewal effects and their hippocampal dependence may be found in Pavlov’s original account of extinction as a loss of CS processing that can be recovered by novel events (Pavlov, 1927; Robbins, 1990). As Pavlov first described, novel stimuli result in an external disinhibition of extinguished responding, which he attributed to an “inhibition of an inhibition” (Pavlov, 1927). By this view, extinction memories are context-specific not because the extinction context has a special role in retrieving the inhibitory CS- no US’ association, but because those associations are easily disrupted by unexpected events including unexpected events in the extinction context).

The possibility that extinction involves attentional decrements to the CS that could limit conditioned responding has received considerable support in the literature (Delamater, 2004; Kaye & Pearce, 1984; 1987; Kehoe & White, 2002; Pavlov, 1927; Robbins, 1990). For example, Robbins (1990) found pronounced declines in attention to extinguished stimuli that could be dissociated from their occasion setting properties, and dissipation of the attentional effects over time accounted for spontaneous recovery (Robbins, 1990). It is conceivable then that other factors that restore attention to an extinguished CS, including stimulus or associative novelty, might promote conditioned responding (Larrauri & Schmajuk, 2008). Insofar as loss of attention to an extinguished CS is driven by an habituation process (Chang & Maren, 2009; Kamprath & Wotjak, 2004; McSweeney & Swindell, 2002; Plendl & Wotjak, 2010)(e.g., habituation of the orienting response), novel stimuli or contexts may lead to dishabituation and restore conditioned responding. Bouton and Ricker (1994) examined this issue and considered whether dishabituation might account for renewal in a conditioned suppression procedure in which orienting responses to the CS have the opportunity to compete with lever pressing and masquerade as renewed conditioned suppression (Bouton & Ricker, 1994). They reasoned that if dishabituation of the orienting response accounted for lever press suppression, that it should have the opposite effect (impairing renewal) in an appetitive procedure in which an orienting response would prevent (not mimic) the CR. However, they found that a context-shift that would be expected to dishabituate orienting responses to the CS still produced robust renewal in an appetitive procedure. That said, the response systems mediating conditioned suppression generate behavior that far outlasts the orienting response. A contextual change or novel stimulus that increments attention to the CS may therefore increase conditioned responding to that stimulus by increasing attention to the CS independent of response competition or compatibility (Larrauri & Schmajuk, 2008; Schmajuk, Gray, & Lam, 1996).

The neural mechanism by which novelty might inhibit the expression or retrieval of extinction memories is not understood. However, if the hippocampus is essential for detecting unexpected events, then hippocampal inactivation would be expected to eliminate the return of fear experienced by presenting an extinguished CS in a novel context (Experiment 1), presenting a novel stimulus along with an extinguished CS in the extinction context (Experiment 2), presenting an extinguished CS in a familiar context in which it has never been encountered (AAB or ABC renewal) (Corcoran, 2004; Corcoran & Maren, 2001), or even presenting the extinguished CS in the original conditioning context (Ji & Maren, 2005). All of these outcomes have been confirmed. Interestingly, hippocampal deficits in fear renewal are most pronounced when an extinguished CS is presented in a novel context or in another context in which it has not been presented, both of which engender associative novelty. Consistent with earlier work, this implies that the hippocampus has a weaker role in detecting (or acting upon) stimulus novelty per se (Honey et al., 1998). One might question whether experiencing an extinguished CS in the conditioning context (as in ABA renewal) should be regarded as an unexpected event’ given that the animal has experienced conditioning in that context. However, proper ABA renewal designs extinguish fear of the conditioning context prior to testing to allow for an unambiguous interpretation of CS renewal in that context. Hence, the reoccurrence of the CS in the conditioning context after extensive exposure to that context in the absence of the CS would surely be unexpected. One can extend the same logic to experiencing the extinguished CS in the conditioning context after a long delay, which, by this view, is an unexpected event that yields spontaneous recovery.

Associative learning models (Rescorla & Wagner, 1972) (Mackintosh, 1975; Pearce & Hall, 1980) have considerable explanatory power in accounting for a variety of associative learning phenomena, yet, in their original form, they fall short in fully accounting for the context-dependence of extinction. Elaborations of these models that introduce context-CS associations or configural representations of elemental and configural stimuli begin to capture some elements of the context-dependence of extinction memory (Pearce, 1994; Wagner, Spear, & Miller, 1981), but ultimately resort to those configurations guiding performance through direct associations with the US. In contrast, it has been argued that direct context-US associations or configural (context-CS)-US associations are not essential for the contextual control of extinguished CSs (Bouton, 2004). Recently, there has been interest in understanding conditioning processes from a non-associative perspective that considers how animals make inferences about the likelihood of events (or the likelihood of an outcome given a particular response) given their past experience. This approach has roots in Krechevsky’s work on hypothesis testing’ (Krechevsky, 1932), and has been formalized in temporal accounts of conditioning (Gallistel & Gibbon, 2000) and so-called latent cause theory (Courville, Daw, & Touretzky, 2006; Gershman & Niv, 2012). In the latter case, computational models have been developed that attempt to capture how animals act upon latent causes’ that may be hidden from observation (Gershman, Blei, & Niv, 2010). From this perspective, animals gather statistical information about the probability of events in their world, and then use this information to make inferences in novel situations (e.g., when a CS occurs in a context in which it has never been presented, such as ABC renewal). This approach has had some success in explaining the context-dependence of extinction (Gershman et al., 2010).

From a functional behavioristic perspective, natural selection may have favored psychological and cognitive systems in which aversive experiences establish latent causes for an event (e.g., the environmental context, the cue, the experimenter, transport cues, the time of day), and this myriad of causes promotes fear that broadly generalizes to other cues and contexts. Extinction may represent a narrowing of these latent causes, establishing a unique observation in which a CS is known to be safe in a particular context. Unexpected events, such as the presence of a novel stimulus or the occurrence of an extinguished CS in an unfamiliar context or indeed the unexpected occurrence of an extinguished CS in a familiar context, challenge these narrowed causes. The immediate response in the face of this expectancy violation may be to engage defensive behaviors; at least until additional extinction trials occur—indeed, once bitten, twice shy.

The present data suggest that the hippocampus is important for novelty-induced renewal of fear, which is consistent with its long-standing role in memory processes (Cohen & Eichenbaum, 1995; Fanselow, 2010) and novelty detection (Honey & Good, 2000; Vinogradova, 2001). There is emerging evidence that the hippocampus mediates renewal effects via connections with the amygdala and prefrontal cortex (Herry et al., 2008; Knapska & Maren, 2009; Knapska et al., 2012; Maren, 2011; Orsini & Maren, 2012). To restore conditioned fear after extinction, the hippocampus might either inhibit the expression of the extinction memory (“inhibition of an inhibition”) or promote attention to the CS or expression of the fear memory, allowing the return of fear. Increases in Fos expression in brain areas associated with fear expression, including the prelimbic cortex and lateral and central amygdaloid nuclei, are consistent with the latter possibility (Knapska et al., 2012; Knapska & Maren, 2009). Yet it remains possible that hippocampal inhibition of infralimbic cortical circuits involved in the expression of extinction are also involved in renewal, and further work is required to determine how prefrontal cortical circuits are regulated by associative novelty.

In sum, novel events challenge animals’ expectations about what is intended to happen in a given situation, and when expectations are violated behavior changes to anticipate alternative outcomes—freezing behavior in the present case. Violations in expectancy produced by novel stimuli, associative novelty, or prediction errors (e.g., unexpected reinforcers) may not only support new learning (Huh et al., 2009; Kamin, 1967; Mcnally, Johansen, & Blair, 2011), but also influence the expression of what has already been learned. In the present work, we now show that the hippocampus is critical for either detecting or acting upon novel information that restores extinguished conditioned responding. This has important implications for how fear recovery phenomena and their neural substrates are understood.

Acknowledgments

Footnotes

References