Reinstatement of conditioned responses in human differential fear conditioning

https://doi.org/10.1016/j.jbtep.2006.04.001Get rights and content

Abstract

The present study aimed at investigating reinstatement of conditioned responding in human classical conditioning using a differential fear conditioning paradigm. Reinstatement is defined as the return of extinguished conditioned responses due to the experience of one or more unexpected USs. As expected the reinstatement group showed reinstatement of US-expectancy while a similar return of conditioned responses was not present in the control group. In the fear ratings a similar pattern was observed. In addition, and in line with previous findings, we found that the more negative the CS+ remained after extinction, the more return of conditioned responding was observed. Clinical implications and suggestions for further research are discussed.

Introduction

During the last few decades exposure-based therapy has been demonstrated to be an effective treatment for anxiety disorders (e.g., Barlow, 1988; Rachman, 1990). Based on extinction theories, the general principle in this kind of therapy is to repeatedly expose patients to the object of their fear. However, despite the effectiveness of this treatment, clinical practice has demonstrated that not all patients remain symptom free at follow-up. A number of successfully treated patients show a return of anxiety symptoms after time has elapsed. For some, this is the basis for a complete restoration of the fear or phobia (i.e. relapse).

Recent conceptualizations of extinction contribute to a better understanding of this return of symptoms. While for a long time extinction was considered to be a process of ‘unlearning’ of a previously learned association (see Donegan, Gluck, & Thompson, 1989), there is now ample evidence to contradict this point of view. Specifically, four phenomena illustrate that extinction is better conceived of as a performance phenomenon that leaves the previously learned associations intact. These phenomena, named ‘spontaneous recovery’, ‘renewal’, ‘rapid reacquisition’ and ‘reinstatement’ (e.g., Bouton & Swartzentruber, 1991; Brooks, Hale, Nelson, & Bouton, 1995) demonstrate that conditioned responses can reappear after extinction, thus indicating that the previously learned association was not unlearned during extinction. Further understanding of these mechanisms and the conditions under which they appear in humans can provide us with important information on how to improve therapy in order to minimize the chances of relapse. In this respect there have been some interesting clinical studies on renewal (Mineka, Mystkowski, Hladek, & Rodriguez, 1999; Mystkowski, Craske, & Echiverri, 2002; Rodriguez, Craske, Mineka, & Hladek, 1999). Renewal is defined as the return of conditioned responses due to a context change after extinction. In these studies some evidence was found for return of fear (ROF) in spider phobics that were tested in a different context than they were treated in. In order to fill in the gap between animal laboratory research and clinical studies, Vansteenwegen et al. (2005) and Vansteenwegen et al. (2006) have been conducting human laboratory research on renewal. They found (ABA) renewal by using a differential fear conditioning paradigm in a student population with the darkness/illumination of the experimenter room as context manipulation (Vansteenwegen et al., 2005). In a next study (Vansteenwegen et al., 2006) they showed how providing extinction-cues during testing in the acquisition context, could attenuate these recovery effects.

The focus of this paper will be on reinstatement. Reinstatement is described as the return of conditioned responses due to the experience of one or more unexpected USs after extinction. Translated to a clinical situation this would mean the following: Imagine a student who develops elevator phobia after experiencing a panic attack in the elevator of the faculty building. After successful exposure treatment this student is again able to take the elevator without expecting a panic attack or being fearful. A few months after treatment, however, the student experiences a panic attack while waiting for an exam in the faculty building. This panic attack (analogous to the reinstating ‘US’ in an experiment) could lead to a re-emergence of the expectancy of a panic attack in the elevator.

Most of the research on reinstatement as a posttreatment source of fear return has been conducted in the animal laboratory (e.g., Bouton & Bolles, 1979; Bouton & Peck, 1989; Brooks et al., 1995; Rescorla & Heth, 1975). An experiment typically includes a number of subsequent phases. During acquisition, the animal learns the contingency between the CS and US, and conditioned responding is established. Next, an extinction phase follows in which the CS is presented alone, without the US. After conditioned responding is extinguished, one or more USs are administered without the CS being present. Following these USs the CS is tested under extinction and a return of conditioned responding is observed. In the animal laboratory there has been extended research on the conditions under which reinstatement can be observed. Besides important theoretical implications, all of these findings may have important consequences for clinical practice if they are shown to be generalisable to human fear conditioning and treatment of anxiety disorders. For example, research has shown that for reinstatement to occur the reinstating US does not need to be identical to the US used during acquisition (Rescorla & Heth, 1975). If this is true for human fears, this would amplify the chances of relapse occuring due to reinstatement.

In order to disentangle different theoretical explanations for reinstatement, several studies have focused on the context-dependency of reinstatement. The results of these studies are not always consistent. For example, while Bouton and Bolles (1979) claimed that reinstatement could only be found when after extinction the USs were presented in the test context, Richardson, Duffield, Bailey and Westbrook (1999) observed reinstatement in a test context different from the reinstatement context.

In contrast to numerous examples of animal research, demonstrations of reinstatement in humans are rare. Recently there have been some studies on reinstatement in the context of causal learning. Vila and Rosas (2001) demonstrated reinstatement of acquisition performance in a causal learning task and García-Gutiérrez and Rosas (2003) explored additivity of renewal and reinstatement in causal learning. However, this causal learning task differs strongly from fear-learning in that no emotionally significant stimulus is used as an outcome. A first study on reinstatement of fear in humans was conducted in our laboratory, using a human differential fear conditioning paradigm (Hermans, Dirikx, Vansteenwegen, Baeyens, & Eelen, 2005). In this study, participants first went through a conditioning phase in which they acquired the CS (picture) — US (electrocutaneous stimulus) association. Next, a series of CS-only presentations was administered in order to extinguish the conditioned responses as measured by verbal ratings and a reaction time task (see below). After extinction, the reinstatement group received two unexpected USs while the control group did not. Finally, the return of conditioned responding was tested under extinction. Thus, the procedure was very similar to that often used in animal research. To test whether acquisition, extinction and reinstatement took place, US-expectancy and fear of the CSs were assessed. Because these verbal reports are sensitive to demand effects, an indirect measure of the learning process was added. More specifically, a reaction time task was included. Dawson, Schell, Beers and Kelly (1982) were the first to use a reaction time task in human autonomic classical conditioning. They observed slower reaction times (RTs) on probes that were presented during the CS+ trials than during the CS− trials, both when these probes were presented early after stimulus onset and when they were presented late after stimulus onset. They concluded that the CS+ was associated with more allocation of processing resources than was the CS−. According to Dawson et al. the difference between the CS+ and CS− on early probes reflects the fact that participants are identifying the CS+ as the CS signalling the US. On late probes the CS+ is considered to allocate more processing resources because the participants are anticipating the US. Hermans and colleagues reasoned that if participants learned the CS+-US relationship and the CS+ became a significant predictor of the US, this would make the CS+ become a more salient stimulus that would require more allocation of processing resources. In this first study, evidence for reinstatement was found in the US-expectancy and fear ratings. Since it was the first time that reinstatement was established in human fear conditioning we consider this study to be an important addition to animal research and a step forward to a better understanding of posttreatment reinstatement of fear in humans. In addition to these reinstatement findings, Hermans et al. (2005) found a significant correlation between the affective meaning of the CS+ after extinction and the amount of return of conditioned responses in the reaction time task. The more negative the CS+ remained after extinction, the more return was observed, indicating that stimulus valence might play a role in return of fear.

Before taking further steps in investigating the circumstances that can attenuate reinstatement our primary goal in the present study was to replicate the findings of Hermans et al. (2005) and extend the reinstatement findings to our indirect measure of US-expectancy. We expected the reinstatement group to show a return of US-expectancy and fear for the CS+ after the reinstatement phase. Furthermore, we expected a differential effect of the reinstatement manipulation in the reaction times for the probes presented during the CS+ and the CS−. In the reinstatement group, a selective slowing of RTs on probes presented during the CS+ was expected from pre- to postreinstatement. No changes in RTs were expected for the control group.

Section snippets

Participants

Forty-five first year psychology students (40 women) participated in this study for course credit. Eleven participants (10 women) were excluded from the study because they were not aware of the CS+-US contingency after acquisition. The data of one participant were not included in the analyses because she had participated in a similar study a year before. Sixteen students were randomly assigned to the reinstatement group (15 women) and the other 17 participants were assigned to the control group

Data reduction and analyses

Because there were no crucial interactions with the condition variable, we combined the data from the reinstatement group and the control group for the analyses concerning acquisition and extinction. For one exceptional case where an important effect interacted with the condition-variable we have described this interaction. The data from verbal ratings were analysed using a 2×2 univariate analysis of variance (ANOVA) with CS-type (CS+/CS−a) and moment (combinations of (when available)

Acquisition phase

All relevant means concerning the acquisition phase are presented in Table 1. As depicted in Fig. 2, after acquisition the participants showed more US-expectancy with regard to the CS+ than concerning the CS−a, F (1, 32)=349.74, MSE=387, indicating that the CS+ had become a valid predictor of the US.

The fear ratings provided further evidence that conditioning had taken place. The CS+ and CS−a were equal in the amount of fear they evoked during the stimulus-selection phase, F<1. The

Discussion

In the present study conditioned responding after acquisition was revealed through US-expectancy and fear ratings and the reaction time task. A substantial decrement in these responses was accomplished in the extinction phase although there remained a significant difference between the CS+ and CS−a at the postextinction assessment for the US-expectancy and fear ratings. The main focus of the study, however, was the effect of unexpected USs on return of conditioned responding after extinction.

Acknowledgements

We would like to thank Jessica Zilski for her helpful comments on the manuscript.

References (26)

  • N.J. Vila et al.

    Reinstatement of acquisition performance by the presentation of the outcome after extinction in causality judgements

    Behavioural Processes

    (2001)
  • D.H. Barlow

    Anxiety and its disorders: The nature and treatment of anxiety and panic

    (1988)
  • M.E. Bouton

    Context, time and memory retrieval in the interference paradigms of Pavlovian learning

    Psychological Bulletin

    (1993)
  • Cited by (66)

    View all citing articles on Scopus
    View full text