Sleep Supports the Slow Abstraction of Gist from Visual Perceptual Memories

Nicolas D Lutz,Karsten Rauss,Patricia Hinse-Stern,Jan Born,Susanne Diekelmann

doi:10.1038/srep42950

Abstract

Sleep benefits the consolidation of individual episodic memories. In the long run, however, it may be more efficient to retain the abstract gist of single, related memories, which can be generalized to similar instances in the future. While episodic memory is enhanced after one night of sleep, effective gist abstraction is thought to require multiple nights. We tested this hypothesis using a visual Deese-Roediger-McDermott paradigm, examining gist abstraction and episodic-like memory consolidation after 20 min, after 10 hours, as well as after one year of retention. While after 10 hours, sleep enhanced episodic-like memory for single items, it did not affect gist abstraction. One year later, however, we found significant gist knowledge only if subjects had slept immediately after encoding, while there was no residual memory for individual items. These findings indicate that sleep after learning strengthens episodic-like memories in the short term and facilitates long-term gist abstraction.

Highlights

Sleep benefits consolidation of individual episodes and items in memory[1]
We analysed the effects of post-encoding sleep on the temporal evolvement of gist abstraction from individual item memories in a visual DRM paradigm
The number of high-confidence old shapes, reflecting veridical episodic-like memory, was enhanced after the 10-hour interval when sleep followed encoding, compared to wakefulness

Summary

Results

Separately conducted for the 20-min and 10-hour groups, showed that both high-confidence old shapes and prototype memory were significantly higher than the rate of ‘old’ responses for new shapes (main effect of Stimulus type in both groups (F(2,28) = 58.52, p < 0.001 and F(2,24) = 46.02, p < 0.001, respectively; all p < 0.001, for post-hoc comparisons; Supplementary Fig. S2). Control analyses confirmed that these differences in recognition performance were independent of the subjects’ original group assignment, as these contrasts were significant in both groups (t(8) = 6.25, p < 0.001, η2 = 0.83 and t(7) = 3.02, p = 0.019, η2 = 0.57, for Short- and Long-retention group, respectively) They were further not affected by the length of the individual retention interval between 20-min/10-hour and one-year recall (all p > 0.15 for correlations between the length of retention interval and memory of prototypes/old shapes). There were no other differences between retention groups or conditions (see Supplementary Table S2)

Discussion

Findings

Methods