Reactivation fails to offer the improvement sleep does

Abstract

In a dynamic process that ultimately affords memories their persistence, memory reconsolidation can serve to strengthen associations following reactivation, particularly in sleep, where active processes may effect overnight enhancement. Reactivation can also occur in wake, where improvement would be unexpected. In an earlier study using performance on the Weather Prediction Task (WPT) as a measure of probabilistic category learning, we looked at the effect of sleep and found significant improvement after a daytime nap, where improvement correlated with the amount of REM sleep obtained. When we introduced interference training following sleep, this REM sleep benefit vanished: post-learning task memory was otherwise preserved. Here, we follow up on these results and test whether reactivation itself-wake reactivation-might be sufficient to induce the improvement found after REM sleep. Our results show that it is not: we saw no improvement on the WPT following reactivation in wake, suggesting sleep may be unique in supporting memory improvement. When we looked at interference effects, we saw unexplained differences between wake and sleep that suggest that while interference is uniformly destabilizing of WPT memories during wake, interference after REM show effects on the memory trace formed during initial learning that are distinctly different from its effects on the subsequently sleep-enhanced memory.

Keywords: Learning; Memory; Memory consolidation; Probabilistic learning; Prospective randomization; REM sleep; Reactivation; Retroactive interference; Sleep; Wake reactivation.

Figure 1.

Sleep vs. Wake in memory reactivation; Weather Prediction Task (WPT). Theoretical model. Changes to memory after reactivation can be constructive or destructive, yet only sleep shows evidence of supporting memory enhancement. Interference applied after reactivation can have differential effects after sleep or wake. Improvement estimates (+, -, or no change in adjusted score from initial test to testing following training) based on results from study data utilizing the WPT [19].

Figure 2.

Card sets and probability of Sun for each card.

Figure 3.

Experimental protocol. Participants underwent training and immediate testing on the WPT at 11AM and were retested at 4PM. Half of the participants had reactivation exposure at 1PM, while half of the + Reactivation participants and half of the non-Reactivation participants underwent interference training at 2PM.

Fig. 4.

Experimental protocol of our earlier nap study [19]. Subjects underwent training and immediate testing on the WPT at 11 AM and were retested at 4 PM. Half of the participants had a 90-min. nap opportunity, while half of the Wake participants and half of the Nap participants underwent interference training (without subsequent testing) at 2PM.

Figure 5.

Results of our earlier nap study [19]. Change in adjusted score from initial testing to retest. Left: Absolute change—a The Nap group performed significantly better at Retest compared to Initial Test (p = 0.005 for absolute and % change); and b performed significantly better than all other groups (all p‘s< 0.05); Right: Percent change.

Figure 6.

Results of the current wake study. Change in adjusted score from initial testing to retest. (a) Absolute change. The + Reactivation, + Interference (+ R,+I) group performed significantly worse at Retest compared to Initial Test (absolute change, p = 0.049; percent change, p = 0.026); (b) Percent test-retest change. The (+ R,+I) group performed significantly worse than all other groups (absolute change: p < 0.037; percent change: p < 0.022); Right: Percent change.