Avoidance Behavior Prevents Modification of Fear Memory During Reconsolidation

Introduction

Anxiety disorders and posttraumatic stress disorder (PTSD) are characterized by fear responses to objects and situations. The fear responses in patients with these disorders can be explained by the Pavlovian fear conditioning paradigm (Lang, Davis, & Ohman, 2000). Conditioned stimulus (CS) causes a conditioned response (CR) after CS is repeatedly paired with an aversive unconditioned stimulus (US). The fear memory trace is the physical representation of the CS–US association encoded in the subgroups of neurons in the lateral amygdala (LA) (Han et al., 2009; Johansen, Cain, Ostroff, & LeDoux, 2011). Activation of the fear memory trace in the LA is related to the expression of the fear response (Reijimers et al., 2007).

The procedure called extinction training involves repeatedly presenting the CS without the US after fear conditioning and thereby inhibits a CR (Dunsmoor, Niv, Daw, & Phelps, 2015; Pavlov, 2010). Standard extinction training does not modify the fear memory trace itself but establishes a new learning, which is called extinction learning (Myers & Davis, 2007). After standard extinction training, the conditioned fear response returns over a period of time (spontaneous recovery) when the context is changed (renewal) or after a US presentation (reinstatement) because the fear memory trace remains intact (Bouton, 2004). This phenomenon is called as fear recovery. Exposure therapy, which is based on the extinction learning paradigm, is used to treat anxiety disorders (Choy, Fyer, & Lipsitz, 2007; Powers, Halpern, Ferenschak, Gillihan, & Foa, 2010). Relapse of anxiety disorders following exposure therapy has also been reported (Boschen, Neumann, & Waters, 2009; Choy et al., 2007). Consequently, treatments that prevent the recovery of a fear response are needed.

Several studies have revealed that the fear recovery is prevented in rodents (Monfils, Cowansage, Klann, & LeDoux, 2009) and humans (Johnson & Casey, 2015; Oyarzún et al., 2012; Schiller, Monfils, Raio, Johnson, LeDoux, & Phelps, 2010; Steinfurth et al., 2014) when extinction training is conducted following the retrieval of fear memory. Standard and postretrieval extinction trainings are known to reflect different mechanisms. Postretrieval extinction training is related to the modification of memory during reconsolidation, rather than the establishment of extinction learning (Auber, Tedesco, Jones, Monfils, & Chiamulera, 2013). New protein synthesis in a group of neurons transforms labile short-term memory into a stable long-term memory, a process called memory consolidation (McGaugh, 2000). After a consolidated memory trace is reactivated, it is destabilized and stabilized again through reconsolidation (Nader, Schafe, & LeDoux, 2000). Providing new information during reconsolidation can modify or update the original memory (Schwabe, Nader, & Pruessner, 2014). Schiller et al. (2010) demonstrated that extinction training during reconsolidation prevents a spontaneous recovery and reinstatement of fear in humans. Oyarzún et al. (2012) replicated research using aversive auditory stimuli as USs. The role of extinction training conducted during reconsolidation is to integrate new information and update fear memory, rather than establish extinction learning (Tedesco, Roquet, DeMis, Chiamulera, & Monfils, 2014).

In addition, standard and postretrieval extinction trainings use different neural networks (Agren et al., 2012; Lee, Haberman, Roquet, & Monfils, 2016). Extinction training during reconsolidation in rodents has been shown to induce changes in their LA receptors (Clem & Huganir, 2010; Monfils et al., 2009), although standard extinction training strengthens the inhibitory neural circuits between the medial prefrontal cortex and LA (Sah & Westbrook, 2008). Postretrieval extinction training in humans reduces activity in the amygdala and involvement of the ventral medial prefrontal cortex (vmPFC) inhibitory circuits (Schiller, Kanen, Ledoux, Monfils, & Phelps, 2013), although the vmPFC is related to the retention of extinction learning (Phelps, Delgado, Nearing, & LeDoux, 2004).

The results of studies that examine the effectiveness of postretrieval exposure therapy for phobia are not consistent (Shiban, Brütting, Pauli, & Mühlberger, 2015; Telch, York, Lancaster, & Monfils, 2017). Shiban et al. (2015) indicated no difference in the long-term effect between standard exposure therapy and postretrieval exposure therapy as well as no reduction in fear on the physiological variable after both therapies. On the other hand, Telch et al. (2017) indicated rapid fear attenuation and low fear recovery at 1-month follow-up after postretrieval exposure therapy, which was relative to that after standard exposure therapy. Telch et al. (2017) discussed that the procedural differences in the fear reactivation manipulation, the duration of reactivation trial, the duration of waiting period, and explicit instructions may account for the discrepant findings. However, there is no clear evidence that these factors account for the inconsistent results as there were numerous methodological differences between these researches. Therefore, the relevant factors that influence the modification of memory during reconsolidation should be revealed to inform the appropriate use of reconsolidation interventions as treatments for anxiety and PTSD.

Avoidance behavior occurs in various forms of strategy to reduce unpleasant experiences, including safety-seeking behavior, escape, distraction, and suppression. Avoidance behavior is related to the onset and persistence of PTSD (Dunmore, Clark, & Ehlers, 1999). A meta-analytic review revealed that avoidance strategy was positively associated with the severity of anxiety disorders from a medium- to large-effect size (Aldao, Nolen-Hoeksema, & Schweizer, 2010). In addition, avoidance behavior can reportedly prevent the effect of exposure therapy for anxiety disorders (Salkovskis, Clark, Hackmann, Wells, & Gelder, 1999; Wells et al., 1995). Thus, it is believed that avoidance behavior maintains mental disorders related to fear conditioning, such as anxiety disorders and PTSD.

We propose that avoidance behavior is a relevant factor that prevents modification of fear memory trace during reconsolidation. Associative learning requires new information about the difference between a predicted event and an actual event (Rescorla & Wagner, 1972). Safety information is the information on the difference between a predicted event (CS is paired with US) and an actual event (CS is not paired with US) during extinction training, which establishes extinction learning (Auber et al., 2013; Myers & Davis, 2007). The presentation of a predictor of no US during extinction training prevents the establishment of extinction learning because there is no difference between the predicted event (CS is not paired with US) and the actual event (CS is not paired with US) (Lovibond, Davis, & O’Flaherty, 2000). Similarly, avoidance behavior prevents extinction learning and protects fear conditioning (Lovibond, Mitchell, Minard, Brady, & Menzies, 2009). It is thought that participants who are instructed to avoid a US attributed the absence of a US to their avoidance behavior, so they could not receive new safety information. Although the modification of fear memory by postretrieval extinction training occurs using different mechanisms from extinction learning, it also requires the reception of safety information during extinction training (Auber et al., 2013; Monfils et al., 2009). Hence, we predicted that avoidance behavior prevents a person from obtaining new safety information during postretrieval extinction training, resulting in the fear memory remaining.

Therefore, the aim of this study was to examine whether avoidance behavior during postretrieval extinction training prevents the modification of fear memory. We assumed that postretrieval extinction training without avoidance behavior diminishes the fear response and prevents the spontaneous recovery of fear as described in previous studies (Oyarzún et al., 2012; Schiller et al., 2010). We also hypothesized that the fear response does not decrease and remains intact after postretrieval extinction training with avoidance behavior because avoidance can prevent a person from obtaining new safety information during reconsolidation and the fear memory trace is not modified.

Method

Procedure

A within-subject design was used, consisting of a three-day experiment (see Figure 1): on Days 1, 2 (24 h after Day 1), and 3 (one week after Day 2). We used 38% partial reinforcement schedule and pseudorandomized orders because they have been frequently adopted in previous research (Oyarzún et al., 2012; Schiller et al., 2010, 2013). Participants were instructed to pay attention to the relations between squares and sounds before acquisition, extinction training, and re-extinction training phase.

OPEN IN VIEWER

Figure 1.

Experimental design and timeline. The design consisted of a three-day experiment: Days 1, 2 (24 h after Day 1), and 3 (one week after Day 2). ACS+ was paired with AUS, and CCS+ was paired with CUS on Day 1. ACS+ and CCS+ were presented to reactivate CS–US associations on Day 2. Instructions on how to avoid AUS were given 10 min after retrieval. The participants underwent extinction training after instructions were given. Participants underwent reextinction training on Day 3. ACS+, CCS+, and CS− were presented without US during extinction and re-extinction training. CS: conditioned stimulus; US: unconditioned stimulus; ACS: avoidance conditioned stimulus; AUS; avoidance unconditioned stimulus; CCS: control conditioned stimulus; CUS: control unconditioned stimulus.

On Day 1, ACS+ was paired with AUS and CCS+ was paired with CUS on a 38% partial reinforcement schedule. The CS− was always presented without aversive sounds. Eight ACS+ without AUS, eight CCS+ without CUS, eight CS−, five ACS+ with AUS, and five CCS+ with CUS were presented in pseudorandomized order. Namely, two different CS–US associations, such as ACS+–AUS and CCS+–CUS, were the learning foci. On Day 2, one ACS+ and one CCS+ were presented at the retrieval phase to reactivate the CS–US associations. We set the interval between the retrieval phase and the extinction training phase to 10 minutes, which is in line with that reported in some previous studies (Oyarzún et al., 2012; Schiller et al., 2010, 2013). Participants were requested to play the Othello game for 10 minutes immediately after retrieval phase. Instructions on how to avoid AUS were given 10 minutes after retrieval phase. We created the instruction on avoidance based on the instruction specified in the experiment of Lovibond et al. (2009). Participants were instructed to press the enter key on the keyboard to avoid the sound presentation immediately after ACS+ was presented during extinction training. In addition, they were also instructed that pressing the enter key was not related to avoiding CCS+ sound presentation and not to press the enter key. The participants underwent extinction training after being given the instructions. Ten ACS+, 10 CCS+, and 11 CS− were presented without a US in a pseudorandomized order. On Day 3, the participants underwent reextinction training to test spontaneous recovery of the fear response. Blockade of fear recovery has been reported to persist for one year after postretrieval extinction training (Schiller et al., 2010). We have set the test of spontaneous recovery to one week after extinction training based on Haarker, Lonsdorf, and Kalich (2015). Eleven ACS+, 11 CCS+, and 11 CS− were presented without a US in a pseudo-randomized order.

Skin potential response (SPR) was measured to assess the fear response to CSs using three Ag–AgCl electrodes placed on the ventral and dorsal surfaces of the left hand and the ventral surface of the left lower arm.

Results

SPRs to all CSs at acquisition (the mean of two last trials), at extinction training (the last trial), and at re-extinction training (the first trial) were used as the dependent variables in the analysis. A two-way analysis of variance with within-subject factors of stimulus (ACS+, CCS+ and CS−) and time (acquisition, extinction training, and reextinction training) was conducted. The following results were revealed (Figure 2): (a) significant main effect for stimulus (F(2, 18) = 8.211, p = 0.003, η_p² = 0.477) and time (F(2, 18) = 3.663, p = 0.046, η_p² = 0.289), (b) significant stimulus × time interaction (F(4, 36) = 4.779, p = 0.003, η_p² = 0.347). The simple effect tests revealed: (c) significant simple main effect for time at CCS+ (F(2, 18) = 9.063, p = 0.002, η_p² = 0.502) and no significant simple effect for time at ACS+ (F(2, 18) = 1.598, p = 0.230, η_p² = 0.151) and CS− (F(2, 18) = 0.607, p = 0.556, η_p² = 0.063), (d) significant simple main effects for stimulus at acquisition (F(2, 18) = 8.562, p = 0.002, η_p² = 0.488), extinction training (F(2, 18) = 4.246, p = 0.031, η_p² = 0.321), and at reextinction training (F(2, 18) = 5.214, p = 0.016, η_p² = 0.367). It is argued that the Shaffer’s approach is robust when the assumption of sphericity is rejected in repeated measure design (Keselman, Algina, & Kowalchuk, 2001). In this study, Mendoza’s sphericity test rejected the assumption of sphericity (p = 0.012) Then, the following Shaffer’s post hoc tests were conducted to reveal the processes of postretrieval extinction training with or without avoidance behavior.

OPEN IN VIEWER

Figure 2.

Mean SPRs to ACS+, CCS+, and CS− at acquisition (mean of two last trials), extinction training (last trial), and reextinction training (first trial). ACS+ and CCS+ were equivalently conditioned. SPR to CCS+ was reduced by extinction training and was not recovered at reextinction training. SPR to ACS+ was significantly higher than CCS+ and CS− at extinction training. SPR to ACS+ was significantly higher than CS− at reextinction training. *p <.05, **p <.01. Error bars represent standard error. CS: conditioned stimulus; ACS: avoidance conditioned stimulus; CCS: control conditioned stimulus; SPR: skin potential response.

Fear acquisition was assessed by the post hoc test. Significant differences were detected between the fear response to ACS+ and CS− (t(9) = 3.345, p = 0.009, g = 1.077) and between CCS+ and CS− (t(9) = 3.362, p = 0.008, g = 1.045) at acquisition. No difference was observed between ACS+ and CCS+ (t(9) = 0.522, p = 0.614, g = 0.101). These results indicate that ACS+ and CCS+ were equivalently conditioned.

The decrease in the fear response to CCS+ from acquisition to extinction training was assessed. A significant difference was observed between acquisition and extinction training (t(9) = 3.235, p = 0.010, g = 1.058) at CCS+, indicating that the fear response to CCS+ was diminished by postretrieval extinction training. The decrease in the fear response to ACS+ was not assessed because no significant simple effect for time was detected at ACS+.

The prevention of a spontaneous recovery to CCS+ at reextinction training was assessed. A significant difference was observed between acquisition and reextinction training (t(9) = 3.052, p = 0.014, g = 0.970) at CCS+, showing that postretrieval extinction training without avoidance behavior prevented the spontaneous recovery of fear.

The effect of avoidance behavior at extinction training was assessed. Significant differences were detected between the fear responses to ACS+ and CCS+ (t(9) = 3.504, p = 0.007, g = 0.967) and between ACS+ and CS− (t(9) = 2.382, p = 0.041, g = 0.824) at extinction training. No difference was observed between CCS+ and CS− at extinction training (t(9) = 0.083, p = 0.935, g = 0.037). It is likely that postretrieval extinction training did not diminish the fear response to ACS+, although it diminished the fear response to CCS+.

The long-term effect of avoidance behavior was assessed. A significant difference was observed between ACS+ and CS− (t(9) = 2.563, p = 0.031, g = 0.885) at reextinction training. There were no differences between CCS+ and CS− (t(9) = 2.103, p = 0.065, g = 0.591), and between ACS+ and CCS+ (t(9) = 1.822, p = 0.102, g = 0.415) at reextinction training. It is possible that the differences between the fear responses to ACS+ and CS− remained until one week after postretrieval extinction training due to avoidance behavior.

Discussion

The aim of this study was to examine whether avoidance behavior during postretrieval extinction training prevents the modification of fear memory. Postretrieval extinction training without avoidance behavior reduced the fear response to CS and prevented the spontaneous recovery of fear in the study, which corresponded with previous studies (Schiller et al., 2010; Oyarzún et al., 2012). Under the condition of postretrieval extinction training with avoidance behavior, the fear response was not reduced as much as it was in the condition without avoidance. Although the difference between the fear responses under both conditions at the last trial of extinction training was diminished at the first trial of re-extinction training, the difference between the fear responses to the CS under the condition with avoidance and to the neutral stimulus was remained. It is possible that avoidance behavior prevents the modification of fear memory during reconsolidation, and the fear response to the CS remains. The current study findings revealed that avoidance behavior is one of the factors that prevent modification of a fear memory induced by postretrieval extinction training.

Postretrieval extinction training provides new safety information and this is integrated into fear memory traces during reconsolidation (Auber et al., 2013; Monfils et al., 2009). Participants who were instructed that avoidance prevents a US presentation could not receive new safety information because they attributed the absence of a US to their avoidance, and there was no difference between a predicted event (avoidance prevented US presentation) and an actual event (no US presentation).

The results indicated that the differences between the fear responses under the conditions with and without avoidance were less one week after postretrieval extinction training. Presenting a distractor during reconsolidation can disrupt the fear memory (Crestani et al., 2015). It is possible that postretrieval extinction training with avoidance behavior functions as a distractor and slightly disrupts the fear memory during reconsolidation. However, the maintenance of differences between fear responses to a CS under the condition with avoidance and to a neutral stimulus at re-extinction training indicates that avoidance prevents the reception of new safety information and the fear memory was not completely disrupted.

Blocking avoidance behavior during a reconsolidation intervention may be important to update a fear memory in clinical settings. Future studies should investigate the effect of avoidance behavior on other forms of reconsolidation interventions in a clinical setting (e.g., postretrieval exposure therapy).

Alexander and French (1946) defined “corrective emotional experience” as reexposure of patients to emotional situations that the patient could not handle in the past under more favorable circumstances. Lane, Lee, Nadel, and Greenberg (2015) proposed that the essential ingredients to reduce the influence of traumatic experiences included reactivating traumatic memories (e.g., ‘important people abandoned me in the past’) and introducing corrective emotional experiences that provide new safety information (e.g., ‘therapist actually did not abandon me, rather provided me with a consistent emotional support’) to be incorporated into the memory traces via reconsolidation. In fact, reactivating traumatic memories and providing corrective emotional experiences are believed to be important components of the therapeutic process in some types of psychotherapy (e.g., schema therapy, emotion-focused therapy, and psychodynamic therapy). For example, schema therapists reactivate traumatic memories using imagery work and provide corrective emotional experiences in the imagery by fulfilling a patient’s emotional needs that were not adequately met in the past (Young, Klosko, & Weishaar, 2003). If avoidance behavior prevents the incorporation of new safety information into a memory trace, blocking avoidance behavior during the reconsolidation process may be an important intervention during these existing psychotherapies.

This study had some limitations that should be discussed. Approximately half of the participants were excluded from the analysis because they did not acquire fear conditioning. We used the same exclusion criteria for acquisition as in a previous study (Oyarzún et al., 2012). However, we used different aversive sounds from the sounds used in that study. We used male screaming and female screaming from IADS, whereas Oyarzún et al. (2012) used girl screaming and pig squealing. It is possible that the sounds used in the current study were less aversive than those used in the previous study, and therefore, some participants did not experience fear in response to these sounds. It is known that individual differences, including past learning experience and cognitive features, affect the acquisition of fear conditioning (Lonsdorf & Merz, 2017). We cannot discuss the effect of these individual differences on the exclusion of the large number of participants because we did not use questionnaires to assess the past learning experience and cognitive features in this study. It is also known that instructions before acquisition influences the strength of fear acquisition. Providing instructions on CS-US contingency (e.g., “a scream will present with a square”) prior to fear conditioning facilitates contingency awareness and strengthens the acquisition of CS-US association (Lonsdorf et al., 2017). We instructed participants to pay attention to the relations between CSs and USs before the fear acquisition phase but did not provide instructions on CS-US contingencies. Future studies should consider using instructions on CS-US contingency to facilitate contingency awareness and strengthen fear acquisition.

The fear responses at later trials of acquisition were used as the indicator of fear conditioning in this study in line with previous studies (Oyarzún et al., 2012; Schiller et al., 2010). The fear responses at retrieval trial or at the first trial of extinction training might be more adequate indicators of long-term fear memory. However, we used a similar experimental design to previous studies that did not present all the CSs at the retrieval trial; therefore, we cannot compare the fear response at the retrieval trial and use it as an indicator of long-term fear memory. In addition, the fear response at the first trial of extinction training in this study was not a suitable indicator of long-term fear memory because it was, in all likelihood, affected by the instruction given about avoidance behavior. Future studies should present all CSs and use the fear response at the retrieval trial as an indicator of long-term fear memory to strengthen the claim that avoidance prevents the modification of memory during reconsolidation.