Abstract
INTRODUCTION: Sleep plays a crucial role in memory consolidation, not only stabilizing newly encoded information but also potentially supporting forgetting. Yet it remains unclear how sleep prioritizes what is retained or discarded when multiple salience cues, such as emotional valence and top-down instructional goals, compete for consolidation. METHODS: In two studies, we examined how emotional content and intentional memory instruction interact to shape memory performance across a 12 h interval that included either nocturnal sleep or wakefulness. Participants completed a directed forgetting paradigm with neutral and negatively valenced words, followed by immediate recognition and delayed free recall. RESULTS: In both Study 1 (online) and Study 2 (in-lab), behavioral results showed that instruction to remember significantly enhanced recognition and recall, whereas emotion alone did not produce consistent benefits; however, sleep condition did not impact memory performance. In Study 2 (in-lab), which included overnight EEG monitoring, physiological markers of sleep revealed meaningful correlates of memory performance. Specifically, sleep spindle activity predicted recall for negative remember-cued words, while Slow Wave Sleep (SWS) and delta power were negatively correlated with total recall, suggesting a trade-off between deep sleep and memory accessibility. REM theta power was associated with increased false recall of emotionally negative foils, consistent with emotional memory generalization. DISCUSSION: Importantly, these findings extend prior nap-based research by demonstrating that full-night sleep physiology reflects selective consolidation mechanisms even in the absence of overt behavioral effects. Overall, results underscore the primacy of top-down instruction over emotional salience in shaping memory, and highlight the utility of sleep physiology for understanding selective memory consolidation.