Editing unwanted memories during human sleep

Abstract

Controlling unwanted memories can be daunting, given that such control would require effortful cognitive control processes. Following trauma, people often avoid thinking of such unwanted memories, which are devoid of the opportunity for direct confrontation and control. It is desirable to modify unwanted memories during human sleep. This dissertation develops the idea that unwanted memories can be edited during human sleep. Chapter 1 introduced the challenges of unwanted memory editing and how sleep benefits memory consolidation. Moreover, chapter 1 introduced targeted memory reactivation (TMR) and its application in enhancing or weakening memory. Based on the literature review and critical analysis, chapter 1 proposed two ideas for editing newly acquired and older unwanted memories: 1) reactivation-based and 2) competition-based memory editing during sleep. In Chapter 2, I examined whether presenting positive emotional words (vs. neutral words) before aversive memory cues can update the unwanted emotion of aversive memories during human non-rapid-eye-movement (NREM) sleep. Results showed that such pairing during NREM sleep reduced negative affect judgments toward paired aversive memory cues during the post-sleep tests. Cue-elicited EEG analyses showed that theta power differences between the positive words and the paired aversive memory cues predicted affect changes across sleep. Particularly, if the positive words elicited larger theta powers than the subsequent memory cues, participants judged the cues less negatively. The item-level analysis further showed that slow oscillation upstate, a state characterized by cortical excitability during NREM sleep, was conducive to effective affect updating when the onset of the first positive words coincided with slow oscillation upstates, aFigure S1 1ffect updating was more likely to happen. Chapter 3 then examined whether presenting positive emotional words (vs. neutral words) after aversive memory cues could impair aversive memories during human NREM sleep. Results showed that positive words disrupted unwanted memory reactivation and impaired the gist of memories during pots-sleep testing, whereas neutral words did not affect the reactivation of unwanted memories. The results of chapters 2 and 3 offered novel possibilities for modifying recently acquired unwanted memories during human NREM sleep, without people being consciously confronted with such aversive memories. Chapters 4 and 5 examined how to edit older unwanted memories via reactivating competing memories during human NREM sleep. Chapter 4 investigated the neural mechanism of TMR in combating existing implicit social biases by reanalyzing the EEG data collected by Hu et al. (2015). Results show that TMR benefits in bias reduction were associated with (a) the timing of memory-related cue presentation relative to the 0.1-1.5 Hz slow-oscillation phase and (b) cue-elicited EEG power within the 1-4 Hz delta range. Chapter 5 examined whether the reactivation of competing positive memories of older unwanted memories can modify unwanted memories. Results showed that reactivating competing memories reduced their negative expression while retaining the negative memories from forgetting. Increased sigma and beta power after the cue may indicate memory integration and emotional competition during sleep. The results of chapters 4 and 5 demonstrated how reactivating competing memories could modulate older unwanted memories and their associated affective responses. Together, across four studies, this dissertation provided evidence that unwanted memories can be edited during human NREM sleep. This research bears implications for developing novel interventions in the treatment of psychiatric disorders characterized by troubling memories, such as Post Trauma Symptom Disorder, depression.