Manipulating memory processing during sleep to explore the critical duration of reactivation events

Abstract

Newly encoded memories are reactivated and consolidated during sleep. However, how the reactivation of a specific memory unfolds over time is poorly understood. What are the temporal dynamics of a single reactivation event within a period of sleep? Does extending a single reactivation opportunity translate to stronger memory benefits? We explored these dynamics by utilizing targeted memory reactivation (TMR), a technique that biases the consolidation of memories via the unobtrusive presentation of memory-associated cues during sleep. Participants learned the on-screen positions of sixty objects, each linked with a unique sound (e.g., cat - meow). Some sounds were then presented during non-REM sleep, with the duration allotted for reactivation causally controlled by varying the timing of the interstimulus interval. TMR did not lead to uniform improvement in memory, and no differences were observed between objects allotted short (2.5 s) and long (7.5 s) reactivation windows. However, memory for objects allotted short windows was impacted by TMR in an encoding-strength-dependent manner, with poorly encoded objects benefiting the most. Classification models trained on EEG data revealed memory reactivation that was time-locked to sound onset during sleep, and this measure of reactivation was linked with memory gains one week later. We did not find evidence for reactivation that extended beyond the time window immediately after sound onset (<2 s). Although our results are not entirely conclusive, they suggest that the critical processes supporting memory consolidation conclude within <2 s after reactivation onset and that extended reactivation windows do not confer additional benefits.