Sleep's Role in Human Spatial Learning

In this issue of SLEEP, Nguyen and colleagues¹ use a novel spatial navigation paradigm to probe the effect of sleep on hippocampus-based spatial learning in humans. The recent surge of research on the influence of sleep on learning and memory began with behavioral studies of perceptual learning^2–4 and motor skill learning⁵ in humans, and neurophysiological studies on rodents showing reactivation of hippocampal neural patterns during slow wave sleep.⁶ The study of Nguyen et al. provides a cross-species link between mechanism and behavior.

A number of studies in recent years have shown hippocampal contributions to human memory during sleep.^7–10 Some of these studies found a correlation between hippocampal activity in sleep and post-sleep performance but did not find a benefit of sleep-dependent memory reactivation.¹⁰ Studies either did not assess⁹ or failed to find⁸ a unique benefit of sleep on spatial memory consolidation. In a study where a unique benefit of sleep on spatial memory consolidation was found,⁷ factors such as differences in vigilance state between wakefulness and sleep or in the capacity to relearn the memory were not eliminated as alternative accounts. The study by Nguyen et al.¹ is a distinct departure in several key methodological aspects that go a long way in addressing these issues. First, it directly compares the effects of sleep following learning with the effects of wake using multiple measures of performance with fine-grained trial-by-trial pre- and post-assessments. Second, it simultaneously assesses vigilance state in participants. Thus, in terms of behavioral assessment, Nguyen et al. is arguably the most thorough to date. In addition, Nguyen et al. measure accuracy of spatial navigation within a virtual three-dimensional environment resembling a maze, an environment ideally suited to engage the hippocampus. Nguyen et al. report overnight benefits of sleep on the accuracy of navigation, providing support for the idea of sleep-dependent human spatial memory consolidation involving the hippocampus.

A closer look at the data illustrated in Figure 4 of Nguyen et al.¹ (bottom, and especially Figure S1 in supplementary materials) bears hints of a more nuanced interpretation. An argument can be made that Nguyen shows overnight “catch-up” by the participants undergoing sleep following training. Nguyen measured distance traveled and amount of backtracking during the time participants navigated through a virtual three-dimensional maze. They randomly assigned participants to either a sleep or a wake group. Each participant was subject to training and retest sessions. The sleep group, which slept after but not before the training, encoded the spatial map less well during training as compared to the wake group. In support of this, the sleep group traveled numerically farther to reach the goal during training. This observation is broadly in accord with a report of impaired ability to form new hippocampal memories without sleep¹¹ but could also be because of time-of-day differences. In and of itself, the observation may not mean much, as the effect of group on encoding did not reach significance (P > 0.1). However, note that the distance traveled by the sleep and wake groups at retest was indistinguishable (Figure S1; compare also light gray bars of Figure 4, top left showing comparable distance traveled at retest). Together with the fact that spatial encoding is somewhat weaker in the sleep compared with the wake group, this implies greater room for improvement in the sleep group at retest. That the sleep and wake groups end up at the same level of accuracy at retest indicates that sleeping after the training does not differentially (compared with wakefulness) enhance performance at the end of retest, but rather it helps performance of the sleep group reach par with the wake group.

Further analysis is consistent with other roles for sleep besides memory consolidation. In each training or retest session, participants navigated the maze three times. Figure S1 from Nguyen et al.¹ suggests spatial learning during the course of both training and retest, as illustrated by the lower values of distance traveled on the terminal versus the initial trial in each. In order to observe a beneficial effect of sleep on overnight spatial learning, one ought to look at the last training trial and the first retest trial following the sleep/wake manipulation. In case of retest trials, performance on subsequent retest trials is a combination of the effects of overnight sleep on both memory consolidation and one's capability to learn more efficiently, which can be thought of as offshoots of sleep's restorative effect on brain function.^12,13 Change in distance traveled (1st retest trial - 3rd training trial) show no between-group difference, which does not comport with the idea of sleep-dependent improvement in hippocampus-dependent spatial learning in humans (completion time also shows little difference in Figure 4).

In sum, Nguyen et al.¹ have conducted a thorough study to relate sleep with hippocampal based spatial learning. Their behavioral findings argue for a role for sleep in hippocampal spatial learning that goes above and beyond memory consolidation. Indeed, sleep is likely to have different effects on different kinds of memories, including passive shielding of fragile memories from sensory interference,^14,15 restitution of brain function that can increase the capacity to form certain new kinds of memories and unmask latent learning of certain kinds of memories formed earlier,^13,16,17 active memory reactivation,^18–20 and synaptic downscaling.²¹ Studies such as that of Nguyen et al.¹ will furnish valuable data for sifting through these and hitherto undiscovered roles for sleep in learning and provide data for a future unifying model of sleep's role and indispensability in different stages of the memory process.

CITATION

Sheth SR. Sleep's role in human spatial learning. SLEEP 2013;36(7):971-972.

DISCLOSURE STATEMENT

Dr. Sheth has indicated no financial conflicts of interest.