Midlife decline in declarative memory consolidation is correlated with a decline in slow wave sleep

Abstract

Sleep architecture as well as memory function are strongly age dependent. Slow wave sleep (SWS), in particular, decreases dramatically with increasing age, starting already beyond the age of 30. SWS normally predominates during early nocturnal sleep and is implicated in declarative memory consolidation. However, the consequences of changes in sleep across the life span for sleep-associated memory consolidation have not been evaluated so far. Here, we compared declarative memory consolidation (for word-pair associates) during sleep in young and middle-aged healthy humans. The age groups (18-25 vs. 48-55 yr) did not differ with regard to learning performance before retention periods that covered, respectively, the first and second half of nocturnal sleep. However, after early retention sleep, where the younger subjects showed distinctly more SWS than the middle-aged (62.3 +/- 3.7 min vs. 18.4 +/- 7.2 min, P < 0.001), retrieval of the word pairs in the middle-aged was clearly worse than in the young (P < 0.001). In contrast, declarative memory retention did not differ between groups after late sleep, where retention was generally worse than after early sleep (P = 0.005). Retention of declarative memories was the same in both age groups when sleep periods containing equal amounts of SWS were compared, i.e., across late sleep in the young and across early sleep in the middle-aged. Our results indicate a decline in sleep-associated declarative memory consolidation that develops already during midlife and is associated with a decrease in early nocturnal SWS.

Figure 1.

Retention of word pairs (A), time in slow wave sleep (B), and performance in the digit span test (C) during the early and late sleep condition in groups of young and middle-aged subjects. In the early sleep condition, where the young subjects had more than a threefold higher amount of slow wave sleep compared with the middle-aged subjects, retention of word pairs was significantly better in the young than it was in the middle-aged subjects. With comparable amounts of slow wave sleep (late-sleep condition in the young subjects and early-sleep condition in the middle-aged subjects), groups did not differ in their retention of word pairs. Since the age groups performed equally well on the digit span test, differences in retrieval of declarative memory were not due to differences in working memory. P < 0.005.

Figure 2.

Retention of word pairs after early sleep plotted against the percentages of SWS (A) and REM sleep (B).