How Much Sleep is Needed?

Since we are not yet aware of the mechanism(s) by which sleep serves to consolidate memories, we cannot say why the entire gamut of sleep dependent plasticity events for learning and memory cannot occur within a few well-advantaged seconds of properly timed activity. Using the principles of spike-timing-dependent plasticity (STDP),² and theta-phase specific stimulation,^3,4 a single burst of activity timed appropriately can permanently alter a neural circuit. Why, then, is so much sleep needed?

Part of the answer as to why a certain packet length of time may be necessary for full expression of consolidation and why smaller increments could impair performance may lie in the process of interweaving novel memories into the fabric of already consolidated memories. As suggested by McClelland,⁵ consolidation of new memories into older circuits may require a restructuring of the broader memory network, thus preventing catastrophic interference that could arise from laying one memory on top of another. Integrating new information into an established neural circuitry might require reactivation of those memories affected by that new information in a coordinated fashion. Such could not be expected to happen all at once. The pattern of synaptic weights comprising a memory circuit is affected by the timing of firing of pre- and postsynaptic neurons, so individual network circuits must be reactivated carefully. Discrete reactivation of a neuronal set on each cycle of (for example) slow waves, spindle waves, or theta waves, would selectively affect only that circuit rather than all indiscriminately. The reformation of related memories could be primed with individual circuit reactivations at the peaks of EEG delta waves. Then novel memories, which were only laid on top of the existing neural network during waking, could be interwoven and the entire network tightened again through selective synaptic strengthening and weakening in the transition to REM and REM sleep, when bidirectional plasticity is probably permitted.^6–8

The evidence from nap studies suggests that, for some learning tasks, the length of sleep provided in a nap may serve the consolidation of that memory just as well as a whole night's sleep.⁹ More studies are needed to tell whether the total length of sleep necessary for efficient memory consolidation depends on the amount of network reorganization that the novel memory requires, and whether this is modulated by task complexity.

The Timing of Sleep: NonREM and REM Sleep Windows

In addition to the studies mentioned by Hagewoud et al.,¹ there are a few other reports that provide insight as to possible mnemonic mechanisms specific to sleep, including the fact that long term potentiation (LTP)—considered to be the building block for learning—is impaired after even short periods of appropriately timed REM sleep disruption. A number of studies have reported LTP induction deficits after REM sleep deprivation,^10–13 which would imply that it would be difficult to acquire a new associative memory after sleep deprivation. But Ishikawa et al.¹⁴ found that there were impairments in the maintenance of LTP when only 4 h of REM sleep deprivation was administered immediately after LTP induction. After learning, then, REM sleep within the first 4 hours would be necessary to maintain the memory. REM sleep deprivation may only impair part of the memory consolidation process. LTP maintenance requires both protein synthesis and gene transcription within a constrained period of time (∼4 h) post-induction to be converted from early LTP to late-LTP, the long lasting form of LTP.¹⁵ Protein synthesis is heightened during nonREM sleep¹⁶ and a protein synthesis inhibitor given during this time following learning impairs performance.¹⁷ Gene transcription associated with synaptic strengthening, however, may occur only in REM sleep.^18,19

How Relevant and Universal is the Sleep Timing and Amount Effect?

Because Hagewoud et al.¹ used a fear conditioning learning protocol, the results could have translational implications for human trauma conditions. The results are also of interest to all those learning something new who wish to schedule the timing of their sleep and learning periods for maximal effectiveness. The universal application of these findings to all learning that requires synaptic network reorganization has been recently confirmed. In addition to published work suggesting that both naps within a specific time period after learning, and a night's sleep improve subsequent performance⁹, recently it has been found that infants learning an artificial language need a nap within 4 h of the training session or they will show no sign of having been taught when tested 24 h later.²⁰

Critical Next Steps

Central to understanding why the timing and amount of sleep are critical to memory consolidation is the need to identify the precise processes and features of sleep that are important to its memory consolidation function. Are the critical processes dissociable from the sleep states, and if so, are there ways to maximize the consolidation function while minimizing sleep in chronically sleep restricted people? What therapeutic sleep interventions could be designed to maximize the attenuation of traumatic memories in those with posttraumatic stress disorder to minimize debilitating fear? Experiments designed to answer these questions will help settle debates that remain regarding the role of sleep deprivation vs. stress vs. attention deficits, etc. Better designed behavioral studies, like that of Hagewoud et al.,¹ will lead to better hypotheses and experimental designs that should deepen mechanistic understanding. Discerning the processes at work during sleep for learning will allow us to better assist those with learning impairments and help individuals avoid the hazards of sleep disorder related cognitive impairments.

DISCLOSURE STATEMENT

The authors have indicated no financial conflicts of interest.

REFERENCES