Targeted Memory Reactivation During REM Sleep in Patients With Social Anxiety Disorder

Abstract

Background: Social anxiety disorder (SAD) is characterized by a significant amount of fear when confronted to social situations. Exposure therapy, which is based on fear extinction, does not often lead to full remission. Here, based on evidence showing that rapid eye movement (REM) sleep promotes the consolidation of extinction memory, we used targeted memory reactivation (TMR) during REM sleep to enhance extinction learning in SAD.

Methods: Forty-eight subjects with SAD were randomly assigned to two groups: control or TMR group. All patients had two successive exposure therapy sessions in a virtual reality (VR) environment, where they were asked to give a public talk in front of a virtual jury. At the end of each session, and only in the TMR group (N = 24), a sound was paired to the positive feedback phase of therapy (i.e., approval of their performance), which represented the memory to be strengthened during REM sleep. All participants slept at home with a wearable headband device which automatically identified sleep stages and administered the sound during REM sleep. Participants' anxiety level was assessed using measures of parasympathetic (root mean square of successive differences between normal heartbeats, RMSSD) and sympathetic (non-specific skin conductance responses, ns-SCRs) activity, and subjective measures (Subjective Units of Distress Scale, SUDS), during the preparation phase of their talks before (T1) and after (T2) one full-night's sleep and after 1 week at home (T3). Participants also filled in a dream diary.

Results: We observed an effect of time on subjective measures of anxiety (SUDS). We did not find any difference in the anxiety levels of the two groups after 1 week of TMR at home. Importantly, the longer the total duration of REM sleep and the more stimulations the TMR group had at home, the less anxious (increased RMSSD) these participants were. Finally, fear in dreams correlated positively with ns-SCRs and SUDS at T3 in the TMR group.

Conclusion: TMR during REM sleep did not significantly modulate the beneficial effect of therapy on subjective anxiety. Yet, our results support that REM sleep can contribute to extinction processes and substantiate strong links between emotions in dreams and waking stress levels in these patients.

Keywords: REM sleep; dreaming; exposure therapy; sleep; social anxiety; targeted memory reactivation.

Figure 1

Study design. Participants underwent a habituation night on Day 0. The following day (Day 1), they had one VR session of generalization task (T0) and two VR sessions of self-focused exposure therapy (T1a and T1b), before the experimental night. On Day 2, participants underwent another VR session of exposure therapy (T2). Then, they spent 1 week at home with a headband sleep device. On day 9, participants came for one last VR session of exposure therapy (T3) and one VR session of generalization task (T4). During the positive feedback phases of the VR, a sound was administered to the TMR group, while no sound was administered to the control group. Both groups were administered the sound during their REM sleep at the experimental night and during the nights at home with a sleep headband.

Figure 2

Mean values for the level of RMSSD (A), score of SUDS (B), and number of ns-SCRs (C) for the control and TMR groups during the preparation periods T1a, T2, and T3. N = 46. Error bars represent 95% CI.

Figure 3

(A) Correlation between REM duration (minutes) and RMSSD in T3 for the TMR Group. Across participants, RMSSD scores were positively correlated with the mean REM duration over 1 week in the TMR group (τ = 0.307, p = 0.047). (B) Correlation between the number of stimulations on average over a week and RMSSD in T3 for the TMR group. Across participants, RMSSD scores were positively correlated with the mean number of stimulations in the TMR group (τ = 0.359, p = 0.019). Each dot indicates a participant.

Figure 4

(A) Correlation between the change of fear in dreams (average of fear during the 2nd week with stimulations minus the 1st week without stimulations) and SUDS in T3 for the TMR group. Across participants, SUDS scores were positively correlated with the change of fear in the TMR group (r = 0.6513, p = 0.0046). (B) Correlation between the change of fear in dreams (average of fear during the 2nd week with stimulations minus the 1st week without stimulations) and ns-SCRs in T3 for the TMR group. Across participants, ns-SCRs scores were positively correlated with the change of fear in the TMR group (r = 0.6308, p = 0.0155). Each dot indicates a participant.