The Sleeping Brain's Influence on Memory

DOES SLEEP INFLUENCE MEMORY? THIS IMPORTANT QUESTION DESERVES A STRAIGHTFORWARD RESPONSE. BUT FEW PROBLEMS IN SCIENCE CAN be answered with unequivocal monosyllables (i.e. “yes” or “no”). Whether sleep benefits memory is no exception. After all, trying to unlock the biological determinants of memory, or to understand the functions of sleep, are enormous challenges in their own right. Blend them together (e.g. does sleep benefit memory?) and you have a web of relationships that are as fascinating as they are difficult to study.

For one, there are several memory systems in the brain. The hippocampus, for example, is the key brain structure responsible for forming declarative (explicit) memories: memories for facts (like learning that the password for my email account is ‘boston1’) and episodic events in time (e.g. knowing that I created that password yesterday). Nondeclarative (implicit) memories, however, employ different neural systems. For example, the hippocampus is not necessary for learning the visual-motor skills of mirror tracing (i.e. drawing a figure while looking in the mirror). Thus, when examining memory, one must be careful to specify what type of memory is being evaluated.

Adding more complexity, we know that memories are not static. They weaken, strengthen or even morph into something new, depending on a variety of influences. The training conditions, for example, can influence the depth of learning and, by extension, the stability of memory for long-term storage and retrieval.

Like memory, sleep, too, is a complex neurobiological process. We use the binary terms “wake” and “sleep,” and further classify sleep into a few discrete categories (stages). Yet we know that the biology of our conscious states is an intricate mesh of continuous and fluctuating variables, rather than a set of discrete stages. To study sleep, we analyze EEG patterns, such as slow-wave sleep (SWS) or sleep spindles. We assume that the measured electrophysiology reflects biological phenomenon that can be understood, and perhaps even manipulated. But more work needs to be done to understand the neurobiological determinants of EEG activity.

So when asking whether sleep influences memory, the challenges seem enormous. Sleep and memory are both derived from complex systems in the brain; they change over time; are different among people; change with age; and are affected by the methods of experimental manipulation. So where does one begin to study their potential relationship? In this issue of SLEEP, Tucker¹ and Tamaki² advance our understanding of how sleep might influence memory by focusing on nuances of learning and specific attributes of sleep electrophysiology.

Tucker examined three hippocampus-dependent memory tasks, one verbal and two spatial. Participants in the study learned the task and then had a one-hour nap opportunity comprised solely of NREM sleep. (If any REM sleep was seen, those participants were excluded from the study.) Participants were tested on all three tasks after sleep, and comparisons were made to a waking control group. Results showed that only a subset of the participants in the verbal and spatial tasks obtained a benefit from sleep: among those who performed well during the training session, performance was better after sleep compared to a waking control group.

It is interesting, and somewhat surprising, that only those who learned the tasks particularly well go on to gain a boost from sleep. Why might sleep specifically enhance memories that are most deeply learned? Perhaps there is something biologically different about a well-encoded memory that renders it available for sleep-dependent improvement. Or, perhaps one hour is not a sufficient quantity of sleep to equally enhance all memories (i.e. if sleep enhances memory, would more sleep provide more enhancement?). Questions remain. But Tucker's results open new avenues for investigation, pointing, importantly, to subtleties of the learning session and how they might influence the effects of sleep on memory.

Another question that this study leaves open is whether there is something special about a nap with just NREM sleep, as the authors speculate. Had they included the participants with REM sleep in their study, would those participants not show the benefit of a nap? The correlation seen between SWS and performance in the verbal task would have been even more intriguing if all of sleep architecture were present, rather than just NREM. While there is strong evidence pointing to aspects of NREM sleep as important for declarative memory consolidation, other models include REM, or even an interaction between NREM and REM (for a review, see reference 3).

In contrast to Tucker's study, Tamaki focused on very different aspects of sleep and memory (see table for differences). Tamaki et al. studied EEG signals, looking for insights into how sleep might benefit hippocampus-independent memory. They hypothesized that properties of fast spindles (13-16 Hz) would correlate with performance improvement overnight on a visual-motor skill. Their results showed that mirror-tracing performance improved overnight, hypothesizing that the improvement resulted from sleep (although they did not have a waking control group for comparison). They also found that this improvement correlated with the amplitude, duration and density (but not number) of fast spindle activity during sleep. No correlations were seen between traditional sleep stages and overnight performance improvement.

Table 1.

Comparison of the Tucker and Tamaki Sleep Studies

Study	Sleep	EEG feature examined	Memory system examined
Tucker et al1	Nap (one hour)	Slow-wave sleep	Encoding of hippocampus-mediated (declarative) memory tasks
Tamaki et al2	Entire night of sleep	Sleep spindles	Hippocampus-independent task (mirror tracing)

Shown in this table are three of the many essential elements one should consider when evaluating a paper on sleep and memory. Even though both studies examined the relationship of sleep and memory, these two studies are remarkably different. The advantage of these differences is that they add different elements to the topic. The disadvantage (through no fault of the researchers) is that the differences make it difficult to make generalizations about the principles discovered.

Tamaki's findings provoke new questions about how fast-spindle activity might facilitate visual-motor (nondeclarative) memory. The authors speculate that thalamocortical networks thought to generate fast-spindle activity are responsible for enhancing motor learning by inducing long-term potentiation.

Oddly, there were more fast-frequency spindles in portions of the control night, compared to the night after learning, and they correlated with performance improvement on the learning night. The authors further speculate that fast-frequency spindles might be a marker of aptitude for motor-skill acquisition.

Does sleep influence memory? These articles by Tucker and Tamaki add to a large body of evidence supporting the idea that sleep can benefit memory. They do so by, respectively, focusing on the subtleties of learning (encoding) and the electrophysiological signatures of sleep. The next steps in understanding the influences of sleep on memory will be (1) to examine how different aspects of memory (e.g. encoding and retrieval) interact with sleep in other sleep-dependent cognitive tasks; and (2) to develop mathematical and neurobiological models of EEG, in order to help us better understand the biology that generates these signals. Achieving these two goals will help establish the empirical limits of the fascinating relationship between sleep and memory, both by understanding which aspects of memory are, in fact, modulated by sleep, and what anatomical or physiological aspects of sleep are responsible for that modulation.