Post-trial sleep sequences including transition sleep are involved in avoidance learning of adult rats

Sleep changes following intensive cognitive activity

Studies over the last 40 years have mainly investigated sleep structure changes as a result of wake duration, in the frame of the classical sleep regulation theories. However, wake intervals of the same duration can profoundly differ in their intensity, which actually reflects the degree of cognitive and physical activity. Data on how sleep can be modified by wake intensity changes (initially sparse and of little consistence) have become much more substantial, especially in the frame of the intense research debate on sleep-memory relationships. Our aim is to examine the vast repertoire of sleep modifications that depend on waking cognitive manipulations, highlighting the sleep features that appear most affected. By systematically addressing this issue, we want to set the basis for future research exploring both the specific nature of the mechanisms involved and the applicative psychosocial and clinical fall-outs, in terms of possible behavioural interventions for sleep quality improvement.

Sleep memory processing: the sequential hypothesis

According to the sequential hypothesis (SH) memories acquired during wakefulness are processed during sleep in two serial steps respectively occurring during slow wave sleep (SWS) and rapid eye movement (REM) sleep. During SWS memories to be retained are distinguished from irrelevant or competing traces that undergo downgrading or elimination. Processed memories are stored again during REM sleep which integrates them with preexisting memories. The hypothesis received support from a wealth of EEG, behavioral, and biochemical analyses of trained rats. Further evidence was provided by independent studies of human subjects. SH basic premises, data, and interpretations have been compared with corresponding viewpoints of the synaptic homeostatic hypothesis (SHY). Their similarities and differences are presented and discussed within the framework of sleep processing operations. SHY's emphasis on synaptic renormalization during SWS is acknowledged to underline a key sleep effect, but this cannot marginalize sleep's main role in selecting memories to be retained from downgrading traces, and in their integration with preexisting memories. In addition, SHY's synaptic renormalization raises an unsolved dilemma that clashes with the accepted memory storage mechanism exclusively based on modifications of synaptic strength. This difficulty may be bypassed by the assumption that SWS-processed memories are stored again by REM sleep in brain subnuclear quantum particles. Storing of memories in quantum particles may also occur in other vigilance states. Hints are provided on ways to subject the quantum hypothesis to experimental tests.

About sleep's role in memory

Over more than a century of research has established the fact that sleep benefits the retention of memory. In this review we aim to comprehensively cover the field of "sleep and memory" research by providing a historical perspective on concepts and a discussion of more recent key findings. Whereas initial theories posed a passive role for sleep enhancing memories by protecting them from interfering stimuli, current theories highlight an active role for sleep in which memories undergo a process of system consolidation during sleep. Whereas older research concentrated on the role of rapid-eye-movement (REM) sleep, recent work has revealed the importance of slow-wave sleep (SWS) for memory consolidation and also enlightened some of the underlying electrophysiological, neurochemical, and genetic mechanisms, as well as developmental aspects in these processes. Specifically, newer findings characterize sleep as a brain state optimizing memory consolidation, in opposition to the waking brain being optimized for encoding of memories. Consolidation originates from reactivation of recently encoded neuronal memory representations, which occur during SWS and transform respective representations for integration into long-term memory. Ensuing REM sleep may stabilize transformed memories. While elaborated with respect to hippocampus-dependent memories, the concept of an active redistribution of memory representations from networks serving as temporary store into long-term stores might hold also for non-hippocampus-dependent memory, and even for nonneuronal, i.e., immunological memories, giving rise to the idea that the offline consolidation of memory during sleep represents a principle of long-term memory formation established in quite different physiological systems.

Different types of avoidance behavior in rats produce dissociable post-training changes in sleep

Avoidance learning affects post-training sleep, and post-training sleep deprivation impairs performance. However, not all rats learn to make avoidance responses, and some rats fail to escape; a definitive behavior of learned helplessness, a model of depression. This study investigated the changes in sleep associated with different behaviors adopted following avoidance training. Rats (n=53) were trained for 100 trials over 2 days (50 trials/day), followed by 23-24 h of post-training polysomnography, then re-tested (25 trials). At re-test, rats were categorized into: 1) Active Avoiders (AA; n=22), 2), Non-learning (NL; n=21), or 3) Escape Failures (EF; n=10). AA rats increased avoidances over days, whereas the NL and EF groups did not. EF rats increased escape failures over days, whereas the NL and AA rats did not. EF rats had increased rapid eye movement (REM) sleep in the first 4h on training day 1. They also had increased non-REM sleep in the first 4h and last 4h on both training days. AA rats had increased REM sleep 13-20 h post-training. The type of behavioral strategy adopted throughout training is associated with a unique pattern of changes in post-training sleep. Training-dependent changes in post-acquisition sleep may reflect distinct processes involved in the consolidation of these different memory traces.

Increased periodic arousal fluctuations during non-REM sleep are associated to superior memory

Sleep has been implicated in the plastic cerebral changes that underlie learning and memory. The scientific investigation of people with exceptional memory has been relatively neglected. We report the results of a polysomnographic investigation of an individual with superior memory performance. The sleep structure, in terms of sleep induction and maintenance, as well as non-REM and REM sleep percentages, were normal. The main finding was an increased number of periodic arousal fluctuations during non-REM sleep (measured as cyclic alternating pattern, CAP) during two consecutive nights (7-8 S.D. units above that observed in age-matched controls). Since CAP rate reflects the structural organization of non-REM sleep, this observation supports the hypothesis of a link between non-REM sleep and declarative memory performance.

Waking EEG power spectra in the rat: correlations with training performance

Adult rats chronically implanted with supradural electrodes were telemetrically EEG recorded during a baseline session, a training session for a two-way active avoidance task, and a retention session. Rats were assigned to a fast learning (FL), slow learning (SL) and non learning (NL) group if they achieved criterion during the training session, the retention session, or in neither session. High-resolution EEG analyses indicated that intergroup differences were present in the low frequency range of waking baseline power spectra. Moreover, baseline delta emissions directly correlated with freezings, and inversely correlated with avoidances, while emissions at 7-10 Hz directly correlated with avoidances and inversely correlated with freezings. Interestingly, during the first training period, waking delta emission selectively increased in FL rats in concomitance with a marked performance improvement; instead, SL and NL rats displayed increments at 7-9 Hz. In addition, freezings scored during the first two training periods directly correlated with post-training waking emission at 2 Hz, and inversely correlated with emission at 7-10 Hz. Conversely, escapes and avoidances directly correlated with waking emission at 7-10 Hz. The data indicate that (i) waking baseline power spectra differ among behavioral groups, and correlate with behavioral performance the following day; (ii) selective modifications of waking power spectra occur in each behavioral group during training; and (iii) behavioral responses during training correlate with post-training waking power spectra. Notably, the delta increment selectively occurring in training FL rats is assumed to reflect online memory processing leading to better performance. The latter observation supports the primary involvement of delta waves in learning.

Cellular and molecular connections between sleep and synaptic plasticity

The hypothesis that sleep promotes learning and memory has long been a subject of active investigation. This hypothesis implies that sleep must facilitate synaptic plasticity in some way, and recent studies have provided evidence for such a function. Our knowledge of both the cellular neurophysiology of sleep states and of the cellular and molecular mechanisms underlying synaptic plasticity has expanded considerably in recent years. In this article, we review findings in these areas and discuss possible mechanisms whereby the neurophysiological processes characteristic of sleep states may serve to facilitate synaptic plasticity. We address this issue first on the cellular level, considering how activation of T-type Ca(2+) channels in nonREM sleep may promote either long-term depression or long-term potentiation, as well as how cellular events of REM sleep may influence these processes. We then consider how synchronization of neuronal activity in thalamocortical and hippocampal-neocortical networks in nonREM sleep and REM sleep could promote differential strengthening of synapses according to the degree to which activity in one neuron is synchronized with activity in other neurons in the network. Rather than advocating one specific cellular hypothesis, we have intentionally taken a broad approach, describing a range of possible mechanisms whereby sleep may facilitate synaptic plasticity on the cellular and/or network levels. We have also provided a general review of evidence for and against the hypothesis that sleep does indeed facilitate learning, memory, and synaptic plasticity.

Sleeping brain, learning brain. The role of sleep for memory systems

The hypothesis that sleep participates in the consolidation of recent memory traces has been investigated using four main paradigms: (1) effects of post-training sleep deprivation on memory consolidation, (2) effects of learning on post-training sleep, (3) effects of within sleep stimulation on the sleep pattern and on overnight memories, and (4) re-expression of behavior-specific neural patterns during post-training sleep. These studies convincingly support the idea that sleep is deeply involved in memory functions in humans and animals. However, the available data still remain too scarce to confirm or reject unequivocally the recently upheld hypothesis that consolidations of non-declarative and declarative memories are respectively dependent upon REM and NREM sleep processes.

Learning and sleep: the sequential hypothesis

During the last 30 years, paradoxical sleep (PS) has been generally considered as the only type of sleep involved in memory processing, mainly for the consistent increase of PS episodes in laboratory animals learning a relatively complex task, and for the retention deficits induced by post-training PS deprivation. The vicissitudes of this idea, examined in detail by several laboratories, have been critically presented in a number of review articles However, according to a more comprehensive unitary proposal (the sequential hypothesis), memory processing during sleep does require the initial participation of slow-wave sleep (SS) in addition to the subsequent involvement of PS. The evidence supporting this hypothesis, largely derived from experiments concerning rats trained for a two-way active avoidance task, is reviewed here in some detail. Recent studies of human sleep are in full agreement with this view. In the rat, the main effect of learning on post-training SS consists in the selective increment in the average duration of SS episodes initiating different types of sleep sequences. Notably, following training for a two-way active avoidance task, the occurrence of this effect in sleep sequences including transition sleep (TS), such as SS-->TS-->W and SS-->TS-->PS, appears related to the processing of memories of the novel avoidance response. Conversely, the occurrence of the same effect in sleep sequences lacking TS may reflect the processing of memories of innate responses (escapes and freezings). Memories of innate and novel responses are assumed to engage in a dynamic competitive interaction to attain control of waking behaviour. Interestingly, in baseline sleep, variables of SS-->TS-->W and SS-->TS-->PS sequences, such as the average duration of SS, TS, and PS episodes, have proved to be good indices of the capacity to learn, as shown by their strong correlations with the number of avoidances scored by rats the following day. Comparable correlations have not been displayed by variables of baseline SS-->W and SS-->PS sequences.

Trains of sleep sequences are indices of learning capacity in rats

In previous work dealing with the identification of four sleep sequences (SS-->W, SS-->PS, SS-->TS-->W and SS-->TS-->PS) in the baseline session of adult male Wistar rats [Mandile P, Vescia S, Montagnese P, Romano F, Giuditta A. Characterization of transition sleep episodes in baseline EEG recordings of adults rats, Physiol Behav 1996;60:1435-1439], we have shown that those containing an intervening episode of transition sleep (TS) strongly correlate with the number of avoidances scored the following day [Vescia S, Mandile P, Montagnese P, Romano F, Cataldo G, Cotugno M, Giuditta A. Baseline transition sleep and associated sleep episodes are related to the learning ability of rats, Physiol Behav 1996;60:1513-152]. More recently, clusters of sleep sequences (trains) separated by waking intervals longer than 60 s have been identified in the baseline session of the same rats [Piscopo S, Mandile P, Montagnese P, Cotugno M, Giuditta A, Vescia S. Identification of trains of sleep sequences in adult rats, Behav Brain Res, this volume], and distinguished in homogeneous or mixed trains according to the presence of a single sleep sequence or more than one sequence. Mixed trains have been further separated into trains containing the SS-->TS-->W sequence (+TSW trains) and trains lacking it (-TSW trains). Analysis of the distribution of variables of baseline trains (and of their sleep sequences and components) among fast learning (FL), slow learning (SL), or non-learning (NL) rats, indicates that variables of +TSW trains prevail in FL rats, while variables of -TSW trains prevail in NL rats. In addition, variables of +TSW trains correlate with the number of avoidances of the training session, while variables of -TSW trains do not significantly correlate, or show inverse correlations. Interestingly, sleep sequences such as SS-->W or SS-->TS-->W show direct or inverse correlations with avoidances depending on whether they are included in +TSW trains or in -TSW trains. The data are interpreted to suggest that the outcome of brain operations performed during a sleep sequence may selectively condition the appearance of later sequences within a time interval shorter than a given threshold. An analogous mechanism may be responsible for the aggregation of sleep components in sleep sequences.

Identification of trains of sleep sequences in adult rats

In previous studies based on high resolution EEG analyses of the 7 h baseline session of 18 adult male Wistar rats [6,14], we have identified four sleep sequences initiating with slow wave sleep (SS) and terminating with waking (W) or paradoxical sleep (PS). Two of these sequences contained an intervening episode of transition sleep (TS). Several variables of these sequences (SS-->W, SS-->TS-->W, SS-->TS-->PS, and SS-->PS) were selectively correlated with the capacity of rats to learn a two-way active avoidance task the following day, and were differently distributed in fast learning, slow learning and non learning rats [21]. The temporal organization of different sleep components in sequences suggested that a comparable temporal organization might concern the different sleep sequences, albeit on a longer time scale. We have now used waking periods longer than 60 s to separate clusters of baseline sleep sequences (trains) in the same rats. Trains containing the same sleep sequence (homogeneous trains) have been distinguished from trains containing different sleep sequences (mixed trains). In addition, mixed trains including the SS-->TS-->W sequence (+TSW trains) have been separated from mixed trains lacking that sequence (-TSW trains). Mixed trains of the +TSW type were longest and most numerous, while homogeneous trains were shortest and least abundant. Mixed trains of the -TSW type displayed intermediate values. Several variables of sleep sequences and sleep components differed within mixed trains and among mixed and homogeneous trains. The data indicate that baseline sleep sequences aggregate in relatively long strings in a non random fashion. The mechanism of this association is discussed.