PT - JOURNAL ARTICLE AU - Matorina Nelly AU - Poppenk Jordan TI - Sleep promotes relational overlapping memories for long-term generalization AID - 10.1101/578492 DP - 2019 Jan 01 TA - bioRxiv PG - 578492 4099 - http://biorxiv.org/content/early/2019/03/16/578492.short 4100 - http://biorxiv.org/content/early/2019/03/16/578492.full AB - Human memory for recent events is believed to undergo reactivation during sleep. This process is thought to be relevant for the consolidation of both individual episodic memories and gist extraction, the formation of generalized memory representations from multiple, related memories. Which kinds of gist are actually enhanced, however, is the subject of less consensus. To address this question, we focused our design on four types of gist: inferential gist (relations extracted across non-contiguous events), statistical learning (regularities extracted from a series), summary gist (a theme abstracted from a temporally contiguous series of items), and category gist (characterization of a stimulus at a higher level in the semantic hierarchy) described in recent reviews. Sixty-nine participants completed memory encoding tasks addressing these types of gist and completed corresponding retrieval tasks the same evening, the morning after, and one week later. Inferential gist in a transitive inference task improved over the course of a week, whereas all other forms of memory were associated with decay. Higher proportions of REM and more spindles were associated with worse performance in a statistical learning task collapsed across time and after one week, respectively. We conclude that memory overlap is an insufficient condition for long-term generalization. Only structured relational memories, rather than associative memories (category and summary gist) or repeated co-occurrence (statistical learning) were enhanced over the course of a week. Furthermore, REM sleep may be involved in schema disintegration, which inhibits participants’ ability to separate a stream of shapes into discrete units.