Counterconditioning reduces contextual renewal in a novel context but not in the acquisition context

Learning about cues that predict danger is adaptive for survival. It is also adaptive to learn when these cues no longer predict threats to prevent the unnecessary expenditure of energy resources. Secondary learning, where cues no longer signal danger (extinction), forms the basis for exposure therapy (Morrison & Ressler, 2014), and extinction-based techniques are a primary treatment for anxiety-related disorders (Abramowitz et al., 2019, Craske et al., 2014). However, extinguished threat behaviors often re-emerge, illustrating the transient nature of an extinction memory. One common scenario under which threat associations return is contextual renewal, whereby extinguished cues encountered outside the extinction context trigger retrieval of the threat association (Bouton & Bolles, 1979). A critical goal of extinction-based treatment for anxiety-related disorders is to override contextual renewal to allow extinction to generalize across contexts. Here, we used a causal associative learning task (the allergist task) (Wasserman, 1990), implemented online, to test the hypothesis that aversive-to-appetitive counterconditioning, whereby aversive outcomes are replaced with positive outcomes, is less susceptible to contextual renewal than standard extinction.

A variety of variations of response reduction protocols have been developed with the aim of enhancing the learning and retention of extinction learning in humans, often in the domain of Pavlovian fear conditioning (Craske et al., 2018, Dunsmoor et al., 2015). One long-standing procedure to retroactively interfere with the expression of defensive behavior involves replacing aversive outcomes with positive outcomes (Keller et al., 2020). This form of behavior modification is known as aversive-to-appetitive counterconditioning (CC). Emerging work on counterconditioning in humans indicates it is a more effective form of retroactive interference than standard extinction at diminishing the expression of conditioned threat over time, referred to as spontaneous recovery (Kang et al., 2018, Keller and Dunsmoor, 2020, Raes and De Raedt, 2012), although findings on the effectiveness of counterconditioning are mixed (Gatzounis et al., 2021, Meulders et al., 2015, van Dis et al., 2019).

Notably, counterconditioning might have different effects given the behavior under investigation, such as expectancy vs. evaluative learning. Expectancy learning arises when a contingency is established between a CS and the US and is eliminated when this contingency is violated. In evaluative learning, the association between the CS and US changes the previously neutral valence of the CS into something liked or disliked (Hermans et al., 2002, Vansteenwegen et al., 2006). While extinction successfully eliminates the expectation of an aversive US (Hermans et al., 2002), it fails to eradicate negative evaluative learning towards the CS, which has an effect on relapse (Dirikx et al., 2004, Huijding and De Jong, 2009). Nonetheless, counterconditioning, as compared to extinction, can successfully reverse the negative valence of a disliked conditioned stimulus (CS) by associating it with a pleasant unconditioned stimulus (US) (Baeyens et al., 1989, Engelhard et al., 2014, Kerkhof et al., 2011).

Whether counterconditioning is more effective than extinction at preventing contextual renewal in humans has received relatively less attention. One possibility is that the existence of a secondary positive outcome, compared to simply omitting the aversive US, reduces ambiguity in the meaning of the CS. Extinction learning operates through prediction error generated between the expected (US) and received (no US) outcome (McNally et al., 2011). A cognitive byproduct introduced by the sudden omission of the US is ambiguity in the signal value of the CS (Bouton, 2002). In other words, where once the CS reliably predicted threat, it may now be interpreted as predicting either threat or nothing significant. Resolving this ambiguity is determined by the context (Bouton, 1993); the CS predicts nothing significant in the extinction context but might continue to predict threat elsewhere. By comparison, counter-conditioning may enhance secondary learning and reduce ambiguity. First, a salient positive outcome is clearly a stronger discrepancy than detecting only the absence of the expected outcome. Second, counterconditioning may help reduce ambiguity by providing a significant alternative association for the CS, helping stabilize secondary learning more effectively than the “no US” association formed by extinction learning. In turn, this may render the counter-conditioned CS more resistant to the return of threat, including contextual renewal.

Direct experimental comparison of counterconditioning to extinction is crucial for informing translational efforts to improve conditioning-based therapies, as any improvement due to a novel technique must exceed that provided by standard extinction. As it pertains to contextual renewal, there is limited research on the effects of aversive-to-appetitive counterconditioning versus standard extinction in reducing renewal. For example, an early study in rats found that aversive-to-appetitive counterconditioning (as well as appetitive-to-aversive counterconditioning) was susceptible to contextual renewal when tested in the original acquisition context (the “A” context, referred to as an ABA renewal design) (Peck & Bouton, 1990), but this was not compared with standard extinction. Thomas et al. (2012) replicated this finding in rats, but critically also compared counterconditioning to standard extinction. However, they showed that a modified form of counterconditioning (in which rats received a reward through performing an instrumental response) effectively prevented contextual renewal, when compared to standard extinction. Holmes et al. (2016) found that in rats, counterconditioning produced increased contextual renewal when compared directly with standard extinction.

One explanation for stronger contextual renewal in counterconditioning concerns overlearning the contextual discrimination between the original context (“A”, wherein the CS predicted threat) and the secondary context (“B”, wherein the CS predicted an appetitive outcome). Consequently, the original meaning of the CS is more strongly preserved and thus more prone to retrieval in the original (A) and even in novel contexts (C). In human causal learning work, counterconditioning has been tested in relation to context manipulations and has been found to generally reduce renewal (e.g., Rosas et al., 2001). However, these designs have not included a direct comparison of counterconditioning to standard extinction.

To compare the effects of contextual renewal following counterconditioning versus extinction in humans, we employed a causal learning framework, the allergist task (Wasserman, 1990), implemented online. Here, CSs were food items initially paired with an allergic reaction (or no allergic reaction) during acquisition in a specific context (restaurant A). Subsequently, in a new context (restaurant B), participants learned that the same food items resulted in either no allergic response (i.e., extinction) or a positive outcome (i.e., counterconditioning). We designed extinction here to explicitly omit an outcome, as opposed to presenting a neutral outcome, to more closely align with laboratory animal and human extinction paradigms (Dunsmoor et al., 2015). Finally, in a between-subjects fashion, participants were tested for renewal of causal judgements through prediction allergy ratings and two-alternative forced choice responses (i.e., which stimulus is less likely to cause an allergic reaction) in different contexts (restaurant A, B, or C). Thus, we tested and compared the renewal of causal judgements in three separate groups: ABA, ABB, and ABC.

Given inconsistent findings from the animal literature regarding the effects of aversive-to-appetitive counterconditioning on contextual renewal (Holmes et al., 2016, Thomas et al., 2012) and limited research in humans, our primary aim was to test if counterconditioning, compared with extinction, survives a shift in context. Based on previous findings on the effectiveness of counterconditioning in reducing recovery of conditioned threat responses in humans (Keller & Dunsmoor, 2020), we predicted that counterconditioning would create a stronger and more salient learned association that would mitigate contextual renewal, as compared to standard extinction. To begin work on disentangling the mechanisms through which counterconditioning might mitigate contextual renewal, we also conduct exploratory analyses to determine if individual differences in perceived valence of the counterconditioned stimulus moderates the contextual renewal effect.

留言 (0)

沒有登入
gif