The Benefits of Targeted Memory Reactivation for Consolidation in Sleep are Contingent on Memory Accuracy and Direct Cue-Memory Associations

Abstract

Study objectives: To investigate how the effects of targeted memory reactivation (TMR) are influenced by memory accuracy prior to sleep and the presence or absence of direct cue-memory associations.

Methods: 30 participants associated each of 50 pictures with an unrelated word and then with a screen location in two separate tasks. During picture-location training, each picture was also presented with a semantically related sound. The sounds were therefore directly associated with the picture locations but indirectly associated with the words. During a subsequent nap, half of the sounds were replayed in slow wave sleep (SWS). The effect of TMR on memory for the picture locations (direct cue-memory associations) and picture-word pairs (indirect cue-memory associations) was then examined.

Results: TMR reduced overall memory decay for recall of picture locations. Further analyses revealed a benefit of TMR for picture locations recalled with a low degree of accuracy prior to sleep, but not those recalled with a high degree of accuracy. The benefit of TMR for low accuracy memories was predicted by time spent in SWS. There was no benefit of TMR for memory of the picture-word pairs, irrespective of memory accuracy prior to sleep.

Conclusions: TMR provides the greatest benefit to memories recalled with a low degree of accuracy prior to sleep. The memory benefits of TMR may also be contingent on direct cue-memory associations.

Keywords: consolidation; memory; reactivation; slow wave sleep.

Figure 1

Experimental procedures (top) and tasks (bottom) used in this study. Targeted memory reactivation (TMR) occurred throughout slow wave sleep (SWS). The lightened image in Active Learning represents picture-location feedback. PSG, polysomnography.

Figure 2

The memory decay index (post-sleep picture-location error score – pre-sleep picture-location error score) was significantly lower for replayed picture locations than nonreplayed picture locations in the low pre-sleep accuracy (Low-Acc) category but not in the high pre-sleep accuracy (High-Acc) category (outlier-corrected data). Error bars represent within-subject standard error of the mean (*paired samples t-test; P < 0.05, **analysis of variance interaction; P < 0.05). ns, not significant; TMR, targeted memory reactivation.

Figure 3

In the low pre-sleep accuracy category, the percentage of total sleep time spent in slow wave sleep (SWS) predicted the memory decay index (post-sleep picture-location error score – pre-sleep picture-location error score) for nonreplayed picture locations but not replayed picture locations. The difference between these correlations was statistically significant (t(27) = 2.15; P = 0.041).