Does splitting sleep improve long-term memory in chronically sleep deprived adolescents? · Cousins JN, van Rijn E, Ong JL et al.

Discussion

We assessed the incidental encoding of pictures and learning of educationally realistic factual knowledge in sleep restricted adolescents under two sleep schedules: split sleep and continuous sleep. We found that splitting sleep between a nocturnal period (5 h sleep opportunity) and a short daytime nap (1.5 h sleep opportunity) resulted in better recall of learned material compared to a purely nocturnal schedule with the same total sleep (6.5 h), specifically for information learned after the nap in the afternoon. Critically, the split sleep schedule did not impair memory for information learned in the morning, despite less nocturnal sleep and less total sleep across 24 h. These findings expand upon prior observations that a daytime nap benefits long-term memory,17–21 by showing that improvements remain under a habitual napping schedule when total available sleep time is matched between nap and no-nap conditions.

The optimization of sleep schedules has been proposed as a low cost way to improve educational outcomes.28 The afternoon learning improvement we observed for both memory tasks suggests that splitting sleep could be one such strategy to improve learning. The 90 min mid-afternoon sleep opportunity in the current study was chosen to maximize the probability of participants obtaining a full cycle of N2, SWS and REM sleep,29 but is toward the upper limits of nap durations that may be practical to implement in schools. Relatively short naps of 30–60 min have been shown to provide benefits to memory lasting up to a week,15,19,21 therefore future studies should examine whether splitting sleep with shorter nap opportunities provides similar benefits for cognition.

Most prior work has examined the benefit of naps in relation to the consolidation of material learned before sleep, but there are several indications that the nap advantage we observed is the result of enhanced encoding, rather than consolidation. The picture encoding task provides a relatively pure measure of encoding capacity, because the incidental nature of the task restricts the use of mnemonic strategies, while only those trials when participants are paying attention and make the correct building/no-building judgment were analyzed. The improved afternoon encoding we observed for the split sleep group on this task also indirectly suggests that improved afternoon learning for the factual knowledge task was related to enhanced encoding, rather than consolidation. This conclusion also seems likely given that the split sleep group obtained less nocturnal sleep in which to consolidate factual knowledge learned in the afternoon.

As splitting sleep resulted in less TST over 24 h, it was surprising that this group had an afternoon memory advantage without any negative impact on information learned in the morning. Although the mechanism underlying this observation cannot be directly explained with the current design—where learning was spread over several days and learning sessions—we posit that the synaptic homeostasis hypothesis2 provides a plausible mechanistic account. Synaptic connections are potentiated during wakefulness, and slow-wave activity is proposed to globally downscale net synaptic weight to avoid saturation in memory networks. This is proposed to facilitate new learning. Although the split sleep group obtained significantly less total SWS in the night prior to morning learning, most SWS takes place in the first 4 h of nocturnal sleep. Here, both groups obtained this early period of slow-wave rich sleep, which may be sufficient to restore encoding capacity. This hypothesis is supported by a pair of studies that show encoding to be comparable between those obtaining half a night (~4 h) and a full night of sleep (~8 h).30,31 It is well established that total25,32 or chronic sleep deprivation6 impairs encoding, but the relatively small difference between 5 and 6.5 h nocturnal sleep appears not to adversely affect encoding. The afternoon nap however, provides an additional period of downscaling for the split sleep group, and this may account for the difference in afternoon encoding capacity we observed.

It is important to note that while splitting sleep had no adverse effects for declarative memory, several other cognitive faculties linked to REM sleep may have been affected, including procedural memory,33,34 creative problem solving35, and emotional regulation.36 With regard to the latter, recently published data from the current study protocol showed that the split sleep schedule was associated with improved positive mood,37 despite reductions in total time spent in REM sleep. The split sleep schedule could be optimized to facilitate REM sleep, for example by slightly extending morning sleep, or taking naps earlier in the morning. Further studies are needed to optimize sleep schedules, establishing a balance between the cognitive benefits associated with these different aspects of sleep and the practicality of translating this behavior to schools.

Another factor to consider is the circadian variation in cognitive performance, particularly in relation to the afternoon ‘circadian dip’, which may have impacted on both memory tasks. The circadian dip is associated with reduced afternoon levels of alertness38 and perhaps also encoding capacity,17 therefore splitting sleep may have minimized the impairment associated with learning at this relatively disadvantageous time of day. This was hinted at by a non-significant reduction in memory performance across the day in the continuous sleep group.

The encoding of any stimulus relies on a complex interaction of factors that includes focusing and sustaining attention on a stimulus, as well as cognitive control processes that allocate resources to the goal of forming an enduring memory trace. These underlying cognitive processes deteriorate with sleep loss39 and are enhanced by daytime naps,40,41 therefore these factors may have contributed to the afternoon learning advantage for the split sleep group, particularly for the long and demanding learning blocks of the factual knowledge task. Consistent with this, measures of post-nap psychomotor vigilance, subjective alertness and subjective ability to focus during learning blocks of this task were significantly higher in the split sleep group, therefore it is likely that this contributed to their improved learning capabilities.

A potential limitation to the picture encoding task is that vigilance in the continuous sleep group prior to the retrieval session remained significantly lower than the split sleep group. Thus, impaired memory of the continuous sleep group may be due to decreased vigilance during retrieval, despite the two nights of intervening recovery sleep intended to equalize vigilance across groups. However, attentional lapses did not significantly correlate with memory performance. Additionally, the few studies that have assessed the effect of sleep loss specifically on retrieval have found no significant impairments after a night of total sleep deprivation,42–44 even though this level of sleep loss produces a consistent vigilance impairment.39 The group differences we observed in vigilance therefore seem unlikely to account for memory impairments, although we cannot entirely rule out this possibility. For the factual knowledge task, vigilance did not differ between groups prior to retrieval. Moreover, we observed an interaction between memory for information learned in the morning and afternoon: if vigilance during retrieval was an influential factor then memory for all information should be equally affected, regardless of when it was learned.

A significant limitation to broad generalization of our work may be that the sleep length in our split and continuous groups was below the 8–10 h recommended for adolescents.3 In the absence of a control group obtaining the recommended amount of sleep, we cannot determine the level of impairment associated with a 6.5 h TIB sleep schedule relative to the recommended sleep duration, or whether splitting sleep is also advantageous for individuals with habitual sleep length that is closer to the recommended amount. However, this sleep schedule provides an accurate reflection of sleep obtained by adolescents undergoing habitual chronic sleep restriction in East Asia,5,6 and around the world,45 therefore our findings suggest that facilitation of naps in schools would be beneficial for memory in a range of adolescent populations.

Another factor to consider for the potential translation of these findings to schools is the spacing of learning and how that relates to sleep. Spaced learning results in superior retention of memoranda although the underlying reasons are debated.46 Spaced learning was recently shown to limit the deleterious impact of multi-night sleep restriction on vocabulary learning.47 As factual knowledge learning in both conditions of the current study was spaced, further work will be needed to assess whether the interaction between spacing and sleep contribute to the observed effects.

To conclude, while adequate nocturnal sleep should continue to be strongly advised, under conditions of sleep restriction, a split sleep schedule appears to be preferable for harnessing the cognitive functions that underpin learning in the afternoon, without negatively impacting upon morning learning. These findings contribute towards growing experimental evidence that facilitating naps in schools may be an effective measure to improve learning outcomes20,21 in chronically sleep deprived adolescents.