The Role of Sleep in Memory Consolidation and Brain Plasticity - Dream or Reality

The idea that sleep may facilitate learning and memoryconsolidation has attracted considerable attention inrecent years, both within the sleep research communityand among the general public. Learning-related functionsof sleep have been more actively investigated over the pastdecade than any other class of sleep functions. Yet there isby no means universal acceptance within the sleepresearch community that sleep facilitates learning andmemory to any considerable degree. The hypothesis hasstrong proponents (Smith 1995; Peigneux and others2001; Rauchs and others 2005; Stickgold 2005; Stickgoldand Walker 2005), including most of the researchersactively investigating in this area, as well as no less deter-mined and principled detractors (Vertes and Eastman2000; Siegel 2001; Vertes and Siegel 2005).

Classic studies of sleep and learning were conductedlargely at the behavioral and electroencephalographic lev-els of inquiry, and such approaches continue to yield inter-esting findings today. But increasingly, this question isbeing addressed also in terms of the response properties ofsingle neurons in networks and the regulation of gene andprotein expression. Sleep researchers are drawing fromthe rapidly growing body of work concerning cellular andmolecular mechanisms of synaptic plasticity, in search ofspecific mechanisms that might link the neuronal activitycharacteristic of sleep states with modulation of synapticconnections in the brain (Benington and Frank 2003). The

establishment of such links holds the greatest promise fordemonstrating conclusively that facilitating memory con-solidation is an important function of sleep.

In this article, we will explore what findings havealready been reported, linking sleep with memory con-solidation and synaptic plasticity. We will first review thebehaviorally based literature and discuss how it has beenanalyzed by both proponents and detractors of learning-related functions of sleep. We will then review findingslinking sleep with changes in neurons and consider whatcellular and molecular mechanisms could explain theseobservations. In closing, we will discuss what work canbe done in the immediate future to test specific hypothe-ses relating to a functional association between sleep andsynaptic plasticity.

Sleep and Learning

Behavioral evidence for a link between sleep and memoryconsolidation falls into 3 broad categories: Sleep depriva-tion has been reported to impair memory consolidation,exposure to new environments or cognitive tasks has beenreported to alter subsequent sleep, and correlations havebeen reported between sleep variables and next-dayimprovements in learning tasks. In the following sections,we will briefly review the findings of these 3 classes ofstudies and consider their implications for establishing alink between sleep and memory consolidation. Becausethis literature has already been the subject of a number ofmore comprehensive reviews (Smith 1995; Peigneux andothers 2001; Benington and Frank 2003; Rauchs and oth-ers 2005; Stickgold 2005), we will herein cite specific find-ings only by way of providing examples of broader trends.

Sleep Deprivation and Memory Consolidation

Experimental tests of the effects of sleep deprivation onmemory consolidation have used a remarkable diversity

The Role of Sleep in Memory Consolidation andBrain Plasticity: Dream or Reality?MARCOS G. FRANK and JOEL BENINGTON

The notion that a good night of sleep improves memory is widely accepted by the general public. Amongsleep scientists, however, the idea has been hotly debated for decades. In this review, the authors considercurrent evidence for and against the hypothesis that sleep facilitates memory consolidation and promotesplastic changes in the brain. They find that despite a steady accumulation of positive findings over the past decade, the precise role of sleep in memory and brain plasticity remains elusive. This impasse maybe resolved by more integrated approaches that combine behavioral and neurophysiological measurementsin well-described in vivo models of synaptic plasticity. NEUROSCIENTIST 12(6):1–12, 2006. DOI:10.1177/1073858406293552

KEY WORDS Sleep, Memory, Learning, Synaptic plasticity

THENEUROSCIENTIST 1

From the Department of Neuroscience, School of Medicine, Universityof Pennsylvania, Philadelphia (MGF), and the Department of Biology,St. Bonaventure University, St. Bonaventure, New York (JB).

This research was supported by the National Institutes of Health (5-R01MH067568).

Address correspondence to: Marcos G. Frank, Department ofNeuroscience, School of Medicine, University of Pennsylvania, 215Stemmler Hall, 35th and Hamilton Walk, Philadelphia, PA 19104-6074(e-mail: [email protected]).

REVIEW

Volume 12, Number 6, 2006Copyright © 2006 Sage PublicationsISSN 1073-8584

of experimental protocols. Both humans and laboratoryanimals have been used as subjects. Various types ofsleep deprivation have been used, including selectiverapid eye movement (REM) sleep deprivation, slow-wave sleep deprivation, and total sleep deprivation foreither the first or second half of the rest period. Theduration of sleep deprivation has also varied betweenstudies as well as the intervals between learning andsleep deprivation and between sleep deprivation andtesting. Studies have also used a wide range of differentlearning paradigms and criteria for learned improve-ment. In humans, effects on both declarative and procedural memories have been measured. Studies ofdeclarative memory have tested for effects on episodicmemory, semantic memory, and priming. Studies of pro-cedural memory have tested for effects on fine motorcontrol, perceptual skills, and cognitive operations. Andeven within any one of those broad categories, studiesconducted by different laboratories have most oftenemployed different specific learning protocols. A simi-larly wide variety of experimental paradigms has beenused in animal (chiefly rodent) studies, in which a divi-sion has typically been made between hippocampal(explicit) and presumed nonhippocampal (implicit)forms of memory.

Unfortunately, no consistent picture has emergedfrom this impressive body of experimental work. Untilabout 10 years ago, there was roughly the same numberof studies yielding positive versus negative findings,although positive findings have predominated since then(Stickgold and Walker 2005). In many cases, both posi-tive and negative findings have been reported in thesame article, depending on the type of sleep deprivationor learning paradigm employed. For example, REMsleep deprivation has been reported to impair memoryconsolidation for complex tasks but not simple ones(Peigneux and others 2001) and after animals haveachieved a certain mastery of a new task but not shortlyafter being introduced to it (Dujardin and others 1990).In one study in humans, memory consolidation in a pro-cedural task was impaired after total sleep deprivationfor the second half of the posttraining night, when REMsleep predominates, but not after REM sleep deprivationfor the entire night (Smith and MacNeill 1994). Inanother study, subjects’ reactions improved following aperiod of normal sleep in an explicit learning paradigmbut not an implicit one—when color cues were used to alert subjects to which particular trials includedrepeated, learnable task sequences (Robertson and oth-ers 2004). Inconsistency in experimental results is a notan uncommon problem during the formative stages of ascientific area of investigation (Collins and Pinch 1993).In such cases, inconsistent findings can be attributed toany of the following five causes.

1. Positive findings could result from something otherthan the intended treatment (in this case, sleep depri-vation) that is inadvertently associated with treatmenttrials, whereas negative findings result from studiesin which such an experimental confound is avoided.

2. Negative findings could result from inconsistencies inexperimental procedures that increase within-groupvariability to the point at which statistically significantbetween-group findings do not appear.

3. The effect could be real but subtle enough to reach thelevel of statistical significance in only a fraction ofstudies, even if experimental design and execution areunchanged.

4. The analyses reported in the research literature couldreach statistical significance and thus appear to bepositive findings, whereas a greater number of otheranalyses of the same data sets may not have reachedstatistical significance and thus remained unreported.Analogously, studies for which apparently positivefindings were achieved may have been published,whereas studies resulting in negative findings mayhave remained unpublished.

5. The experimental effects could be genuinely sensitiveto differences in the experimental protocols (e.g., typesof sleep deprivation or learning paradigms) used instudies yielding positive versus negative findings.

In offering these five possible explanations, we do notmean to imply that any one of them is especially likelyin the field of sleep and memory consolidation; theymerely represent the range of possible explanations forany such body of inconsistent findings in the researchliterature. At the moment, it is not at all clear which oneor more of these causes may be responsible for theinconsistent findings in this field, although argumentshave been advanced for and against one or another ofthem. Not surprisingly, proponents of the idea of sleep-dependent memory consolidation tend to prefer causes 2and 5, whereas opponents tend to prefer causes 1 and 4.

In support of cause 1, opponents of the idea of sleep-dependent memory consolidation have suggested thatimpairment of memory consolidation is in fact causedby the stress associated with either loss of sleep in gen-eral or the specific paradigms used to deprive subjects ofsleep (Horne and McGrath 1984). Indeed, sleep depriva-tion has been shown to increase serum levels of gluco-corticoids, which have in turn been shown to affectanimal and human cognition (Plihal and others 1996).There are, however, a small number of studies in whichstress can be largely ruled out as a factor (Hairston andothers 2005; Ruskin and others 2005), so this criticismmay wane as experimental designs continue to improve.Lingering performance deficits associated with inade-quate sleep could also be responsible for poor scoringduring retest in sleep-deprived conditions (Harrison andHorne 2000), although this is unlikely to apply to proto-cols involving very mild sleep deprivation or a longerinterval between sleep deprivation and retest. The searchfor circumscribed time windows to which an effect ofsleep deprivation on memory consolidation is restrictedhas likewise been geared toward undermining argumentsfor a performance-deficit confound (Smith 1985; Gravesand others 2003). One argument that has been raisedagainst causes 3 and 4 is the fact that, in some studies,between-group differences far exceed minimally accepted

2 THE NEUROSCIENTIST Sleep and Plasticity

levels of statistical significance, indicating that thosecases at least are highly unlikely to represent mere sta-tistical anomalies (Stickgold and Walker 2005).

The preponderance of positive findings in the recentresearch literature may indicate that researchers in thefield have identified the causes of the earlier inconsistencyand corrected them. The standard of expertise in adminis-tering learning tests and/or performing sleep deprivationsmay now be enabling researchers to get results that areconsistent enough to achieve statistical significance inbetween-group comparison, or researchers may now beconcentrating on the types of experimental protocol thatyield positive findings. But in comparing the earlier andmore recent experimental literature, it is not at all clearwhat distinguishes the more successful recent findingsfrom the more ambiguous earlier ones. One such sugges-tion that has been advanced is that sleep deprivation inter-feres with consolidation of procedural memory in humansbut not declarative memory (Smith 1995). But althoughprocedural memory protocols have indeed predominatedin the recent literature, there have also been recent posi-tive findings concerning declarative memory (Plihal andBorn 1999).

The great diversity of experimental methodologiesused by researchers in the sleep and learning communityis partly responsible for this impasse. What the field des-perately needs is more repetition of specific experimentalprotocols in different laboratories. If a given protocol con-sistently produces positive findings in support of an effectof sleep on that type of learning, causes 2, 3, and 4 couldbe ruled out for that case. That well-established findingcould then be used as a baseline for further explorationsof the range of learning and sleep paradigms that producesimilarly positive findings and in tests of candidate exper-imental confounds, by way of ruling out or confirmingcause 1. It would obviously be too expensive and time-consuming to do this for every experimental protocol thathas so far been used in the field, so researchers will haveto agree on a smaller number of model systems for thismore intensive approach. In the meantime, supporters of alink between sleep and memory consolidation should becautious in drawing theoretical conclusions from varia-tions in experimental findings with different experimentalprotocols, as though cause 5 were alone responsible forthe inconsistencies.

Learning and Subsequent Sleep

Studies of the effects of learning on sleep have likewiseproduced a great diversity of findings. Learning has beenreported to increase REM sleep duration, number of REMsleep episodes, REM density and ponto-geniculo-occipital(PGO) wave density in REM sleep, non-REM sleepduration, and localized increases in electroencephalo-graph (EEG) slow-wave activity in non-REM sleep(Smith 1995; Peigneux and others 2001; Rauchs andothers 2005). This literature includes studies in humansand experimental animals following specific learningtasks, as well as studies in which experimental animalsare exposed to an enriched environment. As in the case

of the effects of sleep deprivation on learning describedabove, the cause of this diversity of findings is unclear.As each study has sampled only a fraction of the vari-ables showing changes, it is not even clear whether thisrepresents real inconsistency or merely the differingfocus of different studies. Once again, a number of thereported findings await replication in other laboratories.

Effects of learning or an enriched environment on sub-sequent sleep are consistent with a role for sleep in mem-ory consolidation, but they are equally consistent withother proposed functions of sleep by means of whichnovel experiments and cognitive challenges could indi-rectly alter subsequent sleep. For example, if sleep func-tioned primarily in the restoration of a metabolic deficitaccumulated during waking (Benington and Heller 1995),then learning and enriched environments would beexpected to enhance subsequent sleep insofar as theyplace greater metabolic demands on the brain during wak-ing. Taken alone, therefore, such effects have little impli-cation for the role of sleep in learning and memory,although they can further strengthen the case for such arole if taken in combination with other types of evidence.

Correlations between Sleep and Memory Consolidation

It is a truism in the natural sciences that correlation neednot imply causation and that traditional experimentaldesigns provide stronger evidence than correlative find-ings do. Thus, one might expect that correlationsbetween specific sleep variables and improved perform-ance in learning tasks following a night of sleep wouldprovide the weakest class of evidence for a link betweensleep and memory consolidation. But one could arguethat correlative studies are free of some of the potentialconfounds of sleep deprivation experiments and enableone to test simultaneously the effects of various aspectsof sleep on memory consolidation. Perhaps for these rea-sons, correlative data have been reported more fre-quently in the recent experimental literature.

Following training in a visual discrimination task inhumans, overnight improvement was most strongly corre-lated with duration of slow-wave sleep (stages 3 and 4 innon-REM sleep) in the first quarter of the night and dura-tion of REM sleep in the last quarter (Stickgold, Whifbee,and others 2000). The correlation coefficient between acombination of these two sleep variables and performanceimprovement was an impressive 0.89, accounting foralmost 80% of the variability in performance improve-ment. Following training in patterned finger tapping,overnight improvement was most strongly correlated with amount of stage 2 non-REM sleep (r = 0.66), partic-ularly in the last quarter of the night (r = 0.72; Walker and others 2002). In another finger-tapping protocol,overnight improvement was most strongly correlated withamount of non-REM sleep (r = 0.73), but only when sub-jects were cued as to which patterns were being repeatedand thus learned (Robertson and others 2004). Followingtraining in a procedural learning task involving hand-to-eye coordination, overnight improvement was correlated

Volume 12, Number 6, 2006 THE NEUROSCIENTIST 3

(r = 0.86) with localized increases in EEG slow-waveactivity in non-REM sleep, in specific parts of the parietalcortex that are activated during training and reactivatedduring posttraining non-REM sleep (Huber and others2004). In contrast to the above non-REM sleep–relatedcorrelations, overnight improvement in the classic Towerof Hanoi procedural learning task was correlated withREM density in REM sleep (r = 0.48–0.62; Smith andothers 2004). In rats trained in an active avoidance task,improvement was highly correlated with increased PGOwave density in posttraining REM sleep (r = 0.95; Dattaand others 2000).

As in the case of sleep deprivation studies, these find-ings are inconsistent as to whether overnight improvementis correlated with deep non-REM sleep, stage 2 non-REMsleep, or REM sleep. Each of the above studies uses a dif-ferent procedural learning task, involving visual discrimi-nation, motor actions, or a combination of visual, motor,and/or planning skills. As a result, it is impossible to deter-mine at this point whether the inconsistencies reflect realdifferences in the roles of different sleep stages in differenttypes of memory consolidation.

Memory Consolidation versus Performance Deficits

The most obvious weakness in the above-described stud-ies is that impaired memory consolidation following sleepdeprivation and correlations between memory consolida-tion and sleep variables could merely reflect sleep-relatedperformance deficits during the retest period. A consider-able experimental literature demonstrates that sleep lossimpairs cognition on a number of levels (Harrison andHorne 2000). This explanation is most persuasive whenthe sleep-deprivation protocol is such as would cause sub-jective increases in sleepiness during the retest period, butit cannot be ruled out even when the effects of sleep lossare more subtle. In the visual discrimination and finger-tapping tasks described above (Stickgold, Whifbee, andothers 2000; Walker and others 2002; Robertson and oth-ers 2004), sleep-dependent improvements in performanceare measured in tens of milliseconds, between the displayof a stimulus and a visual mask or between steps in a finger-tapping pattern. It is possible that such time-intensivecognitive operations are ideally suited to revealing per-formance deficits associated with relatively mild sleeploss, thus producing the appearance of sleep-dependentmemory consolidation.

One approach that has been used to obviate this inter-pretation is to compare equal 12-hour intervals of normaldaytime waking with normal nighttime sleep betweentraining and retest, on the assumption that sleepiness following normal daytime waking should not interferewith performance (Stickgold, Whifbee, and others 2000; Walker and others 2002; Huber and others 2004;Robertson and others 2004). But studies in humans whosesleep-wake and circadian cycles are systematically desyn-chronized have demonstrated that sleep propensityincreases substantially during 16 hours of waking but is masked in the late active period by a wake-inducingoutput of the circadian pacemaker (Dijk and Czeisler

1995). Given that performance deficits in humans havebeen observed following sleep deprivation, even whensleepiness is combated by caffeine administration(Harrison and Horne 2000), it is entirely possible thatabsence of performance improvement following 12 hoursof waking reflects merely the increased sleep propensityduring the retest period, in comparison with the morningfollowing a good night’s sleep.

The obvious answer to this potential experimental con-found is to demonstrate differences in performance inequally well-rested subjects, depending on whether theyslept during an earlier interval, shortly after training. Forexample, performance in a visual discrimination task hasbeen reported to improve 3 days after initial training, after2 nights of sleep, but not when subjects were sleepdeprived on the first night and allowed to sleep normallyon the subsequent 2 (Stickgold, LaTanya, and others2000). Although a few such longer term findings havebeen published, they represent isolated reports in compar-ison with the much larger body of experimental findingsinvolving retesting during periods of abiding sleep deficit.

Sleep and Synaptic Plasticity

It is unlikely that the debate over sleep’s role in learningwill ever be resolved by studies that rely solely onbehavioral measurements, no matter how sophisticatedthey become. Although behavioral assays have oftenanticipated the identification of neural processes bymore direct means, such measurements alone cannotreveal the cellular underpinnings of the observed behav-ior. Only a convincing demonstration that sleep directlyinfluences the underlying mechanisms of memory willsatisfy the more ardent critics. In the following sections,we consider cellular and molecular findings that supporta role for sleep in promoting synaptic plasticity. Thismaterial has been extensively reviewed elsewhere(Benington and Frank 2003; Frank 2006), and wherespecific citations are not provided below, the reader mayconsult these reviews for more information.

As is true for behavioral studies of sleep and learning,studies of sleep and plasticity can be broadly dividedinto 3 categories. First are studies that have identifiedpatterns of activity during sleep that might promotesynaptic remodeling. Second are studies that have exam-ined changes in artificially induced forms of plasticity orplasticity-related molecules after sleep or sleep depriva-tion. A third category includes studies that have exam-ined the effects of sleep or sleep deprivation on relatedprocesses that occur in development.

Neuronal Replay in Sleep

A number of investigators have reported reactivation orreplay of wake-active neurons during sleep in rodents,primates, and even birds (Wilson and McNaughton1994; Dave and Margoliash 2000; Hoffman andMcNaughton 2002). Similarly, increased neuronal syn-chronization and metabolic activity in specific brainareas have been reported in humans following learningtasks (Maquet 2001; Maquet and others 2003; Peigneux


and others 2003; Huber and others 2004; Peigneux andothers 2004). A few studies in rodents have also shownsmall but significant correlations between hippocampalevents when replay is reported and thalamocortical spin-dles and delta waves during sleep (Benington and Frank2003; Frank 2006).

There is strong evidence of a rich exchange betweenthe thalamus and the cortex during sleep, which, undercertain conditions, might promote plasticity (Steriade andTimofeev 2003). For example, in vivo recordings in catshave shown that anesthesia-induced spindles produceaugmenting responses in cortical neurons that persist forseveral minutes. Similar spindle stimulus trains candepress or potentiate cortical postsynaptic responses invivo and in vitro (Crochet and others 2005; Rosanova andUlrich 2005). On the other hand, stimulation that mimicsslower EEG rhythms does not reliably depress synapsesin vitro (Perrett and others 2001), contrary to the predic-tions of recent proposals positing a role for non-REMsleep in Hebbian and non-Hebbian synaptic weakening(Benington and Frank 2003; Tononi and Cirelli 2003).

The significance of these findings remains unclear. Inmost rodent studies, reactivation is observed only afterextensive training on familiar tasks, rapidly dissipates,and makes up a small proportion of total recorded activ-ity in sleep. It also appears that sleep is not necessary forreplay as replay is also detected in quiet wakefulness(Kudrimoti and others 1999; Peigneux and others 2006).The correlations between hippocampal and corticalactivity during sleep are likewise interesting but may notreflect an actual transmission of information betweenthese structures (Pelletier and others 2004). Most impor-tant, with the exception of some correlational findings inhumans (Huber and others 2004; Peigneux and others2004), there is no evidence that reactivation of wakingneural activity or spontaneous sleep rhythms promotefunctionally important changes in circuits.

Sleep, LTP, Long-term Depression, and Plasticity Molecules

LTP and long-term depression (LTD) refer to use-dependent,persistent alterations in synaptic weights that strengthen(LTP) or weaken (LTD) specific synapses (Malenka andBear 2004). Although these effects were originally identi-fied in vitro and involved what were at the time considerednonphysiological stimulus protocols, LTP and LTD can beinduced and may occur naturally in vivo (Malenka andBear 2004). These and related forms of synaptic plasticityare now widely considered to be cellular correlates ofmemory. To what extent, then, does sleep influence LTPand LTD?

Beginning in the late 1980s, several investigatorsshowed that sleep states influence LTP in vivo. Overall,it appears that hippocampal LTP can be induced duringREM sleep, whereas similar stimulus protocols duringnon-REM sleep have no effect or produce LTD. A largenumber of studies have also shown that sleep deprivationinhibits the induction or maintenance of LTP in vivo andin vitro (in brain slices obtained from sleep-deprivedrats; Benington and Frank 2003; Frank 2006).

Sleep and sleep deprivation are also reported to influ-ence the synthesis of mRNAs and proteins involved insynaptic plasticity. For example, cortical mRNA tran-scripts for 2 genes important for LTD (i.e., calcineurin andcamKIV) are specifically up-regulated by sleep (Cirelli2005), and neuronal expression of the LTP-related genezif-268 is increased in REM sleep following exposure toenriched environments or LTP in vivo (Ribeiro and others1999; Ribeiro and others 2002; Ribeiro and others 2004).Sleep deprivation has also been shown to decrease hippocampal neurogenesis and neuronal concentrationsof several proteins involved in synaptic remodeling(Hairston and others 2005; Davis and others 2006; Frank2006). In a few instances, these sleep deprivation–inducedchanges are associated with deficits in hippocampal LTPin vitro and learning in rodents (Wright and others 2002;Davis and others 2003; Guan and others 2004; Hairstonand others 2005; Chen and others 2006).

Overall, these findings suggest that sleep and sleeploss can modify synaptic plasticity and the expression of plasticity-related molecules, but a few caveats shouldbe considered. First, only a few of these studies have controlled for circadian effects or stress. Second, thefunctional significance of these cellular/molecular andelectrophysiological changes is not always demonstrated.For example, the fact that experimental stimuli caninduce LTP in REM sleep does not mean that similarprocesses normally occur during REM sleep. Likewise,few of the cellular/molecular changes reported in sleep orafter sleep deprivation have been conclusively linked tochanges in learning or synaptic plasticity.

Sleep and Neural Development

Sleep amounts are higher in the neonatal period than atany other time of life, and this is also a period of rapidbrain development and synaptic plasticity (Roffwarg and others 1966; Frank 2006). Therefore, if sleep con-tributes to synaptic plasticity, one would expect this tobe especially true in developing animals. This possibil-ity has been primarily investigated in the visual system.Development in this sensory system can be divided into2 stages: an early experience-independent stage, inwhich intercellular molecular signals and endogenousneural activity form nascent circuitry, and a subsequentcritical period, in which circuits are sculpted by externalvisual input (Berardi 2000). The critical period has his-torically been investigated by blocking vision in one eye(monocular deprivation [MD]), which causes a massiverewiring of cortical circuits in favor of the open eye, aprocess known as ocular dominance plasticity (Hubeland Wiesel 1970; see Fig. 1). This in vivo plasticity isparalleled by a type of LTP (investigated in vitro) thatcan be induced only in developing animals.

A role for REM sleep in brain maturation has beenexamined by studying the effects of REM sleep depriva-tion (using mechanical techniques), or the elimination ofREM sleep PGO waves, on subsequent visual systemdevelopment. For example, bilateral electrolytic lesions ofthe PGO wave generator in the rostral pontine tegmentumin the neonatal cat cause lateral geniculate nucleus (LGN)


neurons to be physically smaller and to exhibit immatureresponse properties. These morphological and functionalchanges in LGN neurons are consistent with delayeddevelopment in the LGN and suggest that REM sleepneuronal activity may be necessary for normal LGNdevelopment (Davenne and Adrien 1984; Davenne andothers 1989).

Total or selective REM sleep deprivation, or the depri-vation of REM sleep PGO waves, is reported to augmentthe effects of MD on LGN neuronal morphology duringthe critical period. For example, Oksenberg and others(1996) found that 1 week of REM sleep deprivation in kit-tens combined with MD produced smaller LGN neuronsinnervated by the deprived eye (compared to MD alone).A similar increase in LGN neuron size disparity wasfound when MD was combined with brain stem lesionsthat eliminated PGO waves (Shaffery and others 1999).

REM sleep deprivation also appears to augment adevelopmentally regulated form of LTP in the visual cor-tex (Shaffery and others 2002). In this type of LTP, whitematter stimulation in cortical slices prepared from criti-cal period rats produces synaptic potentiation in cortical layers 2/3; however, this is not observed in cortical

slices from older rats. Interestingly, a week of REMsleep deprivation appears to extend the age at which thisform of LTP can be elicited (Shaffery and others 2002).The extension of the critical period by REM sleep deprivation was similar to effects produced by dark rear-ing, which also prolongs the period of induction of thisform of LTP (Shaffery and others 2002). Subsequentstudies from these investigators showed that this plastic-ity could be extended also if REM sleep deprivation was administered near (or overlapping) the end of thecritical period (Shaffery and Roffwarg 2003; Shafferyand others 2005).

Frank and others (2001) demonstrated a role for sleepin ocular dominance plasticity by combining MD withperiods of sleep or sleep deprivation (Fig. 2). Opticalimaging of intrinsic cortical signals and extracellularunit recording showed that sleep nearly doubled thesynaptic remodeling in the visual cortex induced by MD,whereas wakefulness in complete darkness tended toerase the effects of the preceding monocular visual expe-rience (Figs. 3 and 4). This latter finding was particu-larly revealing because it showed that the effects of sleepwere not simply due to a reduction of interfering signals


Fig. 1. Ocular dominance plasticity as revealed by intrinsic signal optical imaging of the visual cortex. The figure showspolar maps generated by stimulating a nondeprived eye (left panel) and an eye deprived of patterned vision (right panel)during the critical period (monocular deprivation [MD]). Each point in the polar map is colored according to the visual stim-ulus orientation that best activated it, and the brightness of each pixel reflects the strength of the response. Radiating con-tour lines are overlaid to highlight pinwheel centers in the maps. Dark regions are areas where responses are greatlyreduced. Note the massive loss of response in the deprived eye map. This type of plasticity can be triggered by as little as6 hours of MD and is enhanced by sleep through unknown mechanisms. However, reversible inactivation of the sleepingvisual cortex inhibits this plasticity, suggesting that the activity of the sleeping brain is a critical component of this process.Each panel represent approximately 4 × 3 mm of cortical tissue. Adapted with permission from the Society for Neuroscience,© 2005 (Jha and others 2005).

(i.e., passive effects of sleep) and instead resulted froman active process that targeted remodeling circuits.

Interestingly, no brain state other than sleep is knownto have such augmenting effects on ocular dominanceplasticity because anesthetic states and cortical inactiva-tion suppress ocular dominance plasticity (Frank 2006).The precise contribution of REM and non-REM sleep tothis process is still unknown, but the enhancement ofcortical plasticity was highly correlated with non-REMsleep time, suggesting an important role for non-REMsleep in the rapid cortical synaptic remodeling elicitedby MD (Frank and others 2001).

It has more recently been shown that this form ofsleep-dependent plasticity requires cortical neuronalactivity (Jha and others 2005; Frank and others 2006).Using a modified version of the experimental designdescribed above, these investigators found that reversibleinactivation of neuronal activity in cortical area V1 during

sleep largely inhibited the remodeling normally observedafter sleep (Fig. 4). This blockade of sleep-dependentplasticity was not due to disruption of ongoing sleep orgross alteration of visual processing, demonstrating thatthe loss of plasticity was specifically caused by the tran-sient silencing of neurons during sleep. Interestingly,additional sleep (with cortical activity restored) did notrescue ocular dominance plasticity. This indicated thatsleep immediately following waking experience was crit-ical for the consolidation of this type of plasticity (Figs.5 and 6). Subsequent findings showed that the effects oflidocaine were likely mediated by changes in postsynap-tic activity. Cortical infusion of the GABAA receptor agonist muscimol during the post-MD sleep period completely blocked sleep-dependent plasticity (Frankand others 2006). Because muscimol predominantlyreduces postsynaptic activity while sparing presynap-tic input, these findings indicate that the underlying


Fig. 2. Experimental design used to test the role of sleep in ocular dominance plasticity. A, Time course of electroen-cephalograph (EEG) and electromyograph (EMG) signals from three representative experiments. EEG slow-wave activ-ity (SWA) and integrated EMG signals expressed as a percentage of their respective maximum values are shown foreach experiment. Top, MD6 (cats assayed for changes in ocular dominance immediately after 6 hours of monoculardeprivation). Middle, MDS (6 hours of monocular deprivation [MD] followed by 6 hours of ad lib sleep in the dark).Bottom, MDSD (6 hours of MD followed by 6 hours of sleep deprivation in the dark). Arrows mark regions of small SWAincreases during wake. B, Expansion of 1 hour of the MDSD experiment with a hypnogram superimposed on the EMGtrace in the bottom panel. Wake (W) is characterized by intermediate to high EMG signals coupled with low EEG SWAvalues. Non-REM sleep (N) is characterized by low to intermediate EMG signals coupled with intermediate to high EEGSWA values. Very low EMG signals and very low EEG SWA values characterize REM sleep (R). C, Polygraphic EEG andEMG signals from selected 10-second epochs of wake, non-REM sleep, and REM sleep in an expanded hypnogram.Adapted with permission from Elsevier, © 2001 (Frank and others 2001).

mechanisms, though still unknown, require postsynapticaction potentials.

To summarize, there are several intriguing findings thatsupport a role for sleep in brain development. Althoughthe interpretation of some of these studies is complicatedby the use of prolonged sleep deprivation or secondaryeffects of electrolytic lesions, they suggest that sleepinfluences the early formation of neural circuitry and itssubsequent sculpting by experience. Unfortunately, veryfew of these results have been replicated by independentlaboratories, and only a handful of scientists are currentlyengaged in this line of research, even though a role forsleep in brain development was first hypothesized 40years ago. In 1966, when Roffwarg and others first pro-posed such a role for sleep, much less was known abouthow endogenous neural activity, intercellular molecular

signals, and experience shape developing circuitry. Thesituation has changed, and the proposal is no less interest-ing and important today than it was decades ago. As willbe discussed further, a rapprochement between sleep anddevelopmental neurobiology is not only timely but alsomay ultimately reveal precisely how sleep influencesbrain plasticity.

Summary of Current Findings

There is now behavioral, electrophysiological, cellular,and molecular evidence to support the hypothesis thatsleep facilitates memory consolidation and brain plastic-ity. These myriad findings, however, have not been inte-grated in a manner that reveals by what mechanism sleepmay achieve these functions. Part of the problem resides


Fig. 3. Ocular dominance plasticity in cats deprived of vision in one eye with or without a subsequent sleep period. Dataare from cats shown in Figure 2. Ocular dominance maps represent regions of the cortex preferentially responsive toeither the deprived eye (dark regions with red borders) or the nondeprived eye (white/gray areas). The top row shows ocu-lar dominance maps of the hemisphere ipsilateral to the deprived eye; the bottom row shows ocular dominance ratiomaps of the hemisphere contralateral to the deprived eye. In the MDS condition (6 hours of monocular deprivation [MD]followed by 6 hours of ad lib sleep in the dark), the deprived eye regions in both hemispheres are smaller than in the othertwo conditions. The bottom panels show corresponding ocular dominance histograms from randomly placed microelec-trode penetrations in the two hemispheres of the individual animals illustrated above. The recordings are ranked on ascale ranging from 1 to 7 developed by Hubel and Weisel that shows the distribution of responses biased toward thenondeprived eye (open symbol, ranks >4) or deprived eye (closed symbol, ranks <4). The shift index (SI) shown for opti-cal maps and electrophysiological recordings is a scalar measure of ocular dominance across both hemispheres andvaries from –1 (complete dominance by the deprived eye) to 1 (complete dominance by the nondeprived eye). The SI value for cats with normal binocular vision is approximately 0. Note that SI values are largest in animals allowed tosleep after MD and smallest in animals kept awake in the dark (for details, see Frank and others 2001). Adapted with permission from Elsevier, © 2001 (Frank and others 2001).

in the wide variety of experimental paradigms used in bothhumans and animals. A more salient issue is that each par-ticular approach in isolation has limited utility. For exam-ple, even if the inconsistencies in experimental findingsare eventually resolved, sleep and learning studies alone

cannot reveal what happens in the sleeping brain thatpromotes memory consolidation. Likewise, althoughphysiological studies have identified interesting changesin gene expression or neuronal activity in sleep or fol-lowing sleep deprivation, most have not shown that thesechanges are functionally important for plasticity orlearning. At present, only a handful of studies haveattempted a more integrated approach, and these havebeen primarily limited to human learning studies (inwhich cellular mechanisms can only be grossly investi-gated) or rodent LTP (Frank 2006; but see below). Thus,the current situation is at once tantalizing and frustrat-ing; it appears that many single threads of evidence existin support of the hypothesis, yet they have not beenwoven into a seamless whole that irrefutably demon-strates a role for sleep in brain plasticity.

Dreaming of Integrative Models of Sleep-Dependent Plasticity

A truly integrative approach to the problem shouldincorporate the following criteria. First, it should focus


Fig. 4. Sleep enhances ocular dominance plasticity.Data represent pooled neuronal recordings from bothvisual cortices in cats immediately assayed after 6 hoursof monocular deprivation (MD) only (A, MD6, n = 349recordings), after 6 hours of MD + 6 hours of ad lib sleep(B, MDS, n = 387 recordings), or after 6 hours of MD + 6hours of sleep deprivation in complete darkness (C,MDSD, n = 408 recordings). Histograms are ranked suchthat 1 represents cells driven exclusively by the nonde-prived eye, 7 represents cells driven exclusively by thedeprived eye, and 4 represents cells that are drivenequally by the two eyes. Histograms for each hemisphereare ranked according to the traditional 7-point scale ofHubel and Wiesel (1970). Note that more neurons are biased toward the nondeprived eye after sleep (MDSvs. MD6) and that this distribution is partially reversed whensleep is prevented (MDSD vs. MDS). Adapted with permis-sion from Elsevier, © 2001 (Frank and others 2001).

Fig. 5. Reversible inactivation of the visual cortex duringpost–monocular deprivation (MD) sleep. To determine ifthe effects of sleep on ocular dominance plasticity wereactivity dependent, the design shown in Figure 2 wasmodified so that the sleeping visual cortex was reversiblyinactivated with the cation blocker lidocaine. The top pan-els represent rapid eye movement (REM) sleep and non-REM (NREM) sleep electroencephalograms (EEGs) fromanterior positions far from the infusion site (right hemi-sphere anterior position), within 1 mm of the infusing can-nula in the visual cortex (right hemisphere infusion) and thenuchal electromyogram in a control cat infused bilaterallywith artificial cerebrospinal fluid (vehicle). Bottom panelsrepresent equivalent measurements made from an experi-mental cat infused bilaterally with lidocaine. Note the pro-nounced suppression of EEG activity in V1 in the lidocainecat during the infusion period in both REM and NREMsleep. The infusion of lidocaine had no significant effectson ongoing sleep behavior and only modest effects oncortical EEGs distant from the infusion site. Each panelrepresents 30 seconds. Adapted with permission from theSociety for Neuroscience, © 2005 (Jha and others 2005).

on a type of plasticity that occurs in the intact brainbecause this would potentially allow one to measurechanges in neural circuitry as they occur in natural sleep.Second, the type of plasticity should be physiologicallymeaningful, with known behavioral outputs that haveadaptive significance. For example, several studies haveshown that sleep and sleep loss modify hippocampalLTP induced by high-frequency stimulation of postsy-naptic neurons, yet the adaptive significance of thesefindings is unclear. Naturally occurring LTP and LTDmay be driven by more subtle forms of activity, and insome cases in vitro LTP/LTD is unrelated to plasticity invivo (Malenka and Bear 2004). Third, the type of plas-ticity examined should be reasonably well understoodon a cellular level so that one can examine the contribu-tion of specific and well-characterized cellular signalingpathways in sleep-dependent synaptic remodeling.Fourth, the plasticity should be rapid; that is, it can betriggered over a relatively short period of time, thusallowing one to use very modest manipulations of sleepand wake in an experimental design. Last, the resultsshould reveal basic principles by which sleep affectsbrain plasticity more generally. That is, the mechanismsrevealed in this system should provide insights into howsleep influences plasticity in other parts of the brain andat different stages of life.

We believe that the study of ocular dominance plasticityin the developing visual system is especially promising

because it satisfies the first four of these criteria and in timemay satisfy the last. Ocular dominance plasticity occurs inthe intact, unanesthetized brain and is triggered by naturalforms of stimuli (i.e., changes in vision as opposed to arti-ficial trains of action potentials). It is associated withbehavioral outputs that have adaptive significance for theanimal (stereoscopic vision and acuity) and is welldescribed on a cellular level. Changes in ocular dominanceoccur within hours, which allows for minimal manipula-tions of sleep and wake in the experimental design. Last,the study of ocular dominance plasticity has providedimportant insights into synaptic plasticity in other sensorysystems and has recently been shown to persist beyond theclassic critical period of development (albeit in slightly dif-ferent forms; He and others 2006). Therefore, by demon-strating a role for sleep in this system, one may eventuallyunlock the deeper mystery of how sleep interacts with spe-cific cellular and molecular plasticity mechanisms acrossthe life span and in many parts of the brain. A similarapproach may be feasible in other sensory systems thatalso exhibit well-described forms of plasticity in vivo,such as somatosensory barrel cortex in rodents or auditorycortex (Berardi and others 2000).

Concluding Remarks

We cannot emphasize strongly enough the importance ofresolving the current controversy regarding sleep’s pur-ported role in memory consolidation and synaptic plas-ticity. Determining the function of sleep is one of thebiggest unsolved problems in neuroscience today, andlearning-related functions of sleep are among the mostactively investigated candidate functions. If proponentsof these functions are on the right track, then we stand togain considerable insight into the nature of the verte-brate nervous system by determining how neural activityin sleep facilitates the remodeling of neural circuits. Butto do so, we need to establish broadly applicable modelsystems that produce consistent experimental results andthat link sleep-dependent changes in behavior with iden-tifiable physiological processes on the cellular andmolecular level. It is only through such an approach thatwe will move past an increasingly sterile debate andtoward hard proof that supports or refutes this mostintriguing hypothesis.

References

Benington JH, Frank MG. 2003. Cellular and molecular connectionsbetween sleep and synaptic plasticity. Prog Neurobiol 69:77–101.

Benington J, Heller HC. 1995. Restoration of brain energy metabolismas the function of sleep. Prog Neurobiol 45:347–60.

Berardi N, Pizzorusso T, Maffei L. 2000. Critical periods during sen-sory development. Curr Opin Neurobiol 10:138–45.

Chen C, Hardy M, Zhang J, LaHoste GJ, Bazan NG. 2006. AlteredNMDA receptor trafficking contributes to sleep deprivation-inducedhippocampal synaptic and cognitive impairments. Biochem BiophysRes Comm 340:435–40.

Cirelli C. 2005. A molecular window on sleep: changes in gene expression between sleep and wakefulness. NEUROSCIENTIST11:63–74.

Collins H, Pinch T. 1993. The golem: what everyone should knowabout science. Cambridge (UK): Cambridge University Press.


Fig. 6. Reversible inactivation of the sleeping visual cor-tex inhibits ocular dominance plasticity. Ocular dominancehistograms compiled from pooled unit recordings in bothhemispheres from (A) cats whose visual cortices werereversibly inactivated during sleep (VEH, n = 1628 record-ings), (B) cats infused with vehicle only (LIDO, n = 1558recordings), (C) cats infused with lidocaine into the cere-brospinal fluid (SHAM, n = 1203 recordings), and (D) catstreated identically as the LIDO cats but provided an addi-tional 6-hour ad lib sleep period without cortical inactiva-tion (RESCUE, n = 532 recordings). Note that the normalshift in responses toward the nondeprived eye is reducedin cortices reversibly silenced during post–monocular dep-rivation sleep (VEH, SHAM vs. LIDO) and that additionalsleep after this inactivation does not restore plasticity(RESCUE vs. VEH, SHAM). Histograms are ranked as inFigure 4. Adapted with permission from the Society forNeuroscience, © 2005 (Jha and others 2005).

Crochet S, Fuentealba P, Cisse Y, Timofeev I, Steriade M. 2005. Synapticplasticity in local cortical network in vivo and its modulation by thelevel of neuronal activity. Cereb Cortex 16:618–31.

Datta S. 2000. Avoidance task training potentiates phasic pontine-wavedensity in the rat: a mechanism for sleep-dependent plasticity. J Neurosci 20:8607–13.

Dave AS, Margoliash D. 2000. Song replay during sleep and compu-tational rules of sensorimotor vocal learning. Science 290:812–6.

Davenne D, Adrien J. 1984. Suppression of PGO waves in the kitten:anatomical effects on the lateral geniculate nucleus. Neurosci Lett45:33–8.

Davenne D, Fregnac Y, Imbert M, Adrien J. 1989. Lesion of the PGOpathways in the kitten. II. Impairment of physiological and mor-phological maturation of the lateral geniculate nucleus. Brain Res485:267–77.

Davis CJ, Harding JW, Wright JW. 2003. REM sleep deprivation-induced deficits in the latency-to-peak induction and maintenanceof long-term potentiation within the CA1 region of the hippocam-pus. Brain Res 973:293–97.

Davis CJ, Meighan PC, Taishi P, Krueger JM, Harding JW, Wright JW.2006. REM sleep deprivation attenuates actin-binding protein cor-tactin: a link between sleep and hippocampal plasticity. NeurosciLett. 400:191–6.

Dijk DJ, Czeisler CA. 1995. Contribution of the circadian pacemakerand the sleep homeostat to sleep propensity, sleep structure, elec-troencephalographic slow waves, and sleep spindle activity inhumans. J Neurosci 15:3526–38.

Dujardin K, Guerrien A, Leconte P. 1990. Sleep, brain activation andcognition. Physiol Behav 47:1271–8.

Frank MG. 2006. The mystery of sleep function: current perspectivesand future directions. Rev Neurosci. Forthcoming.

Frank MG, Issa NP, Stryker MP. 2001. Sleep enhances plasticity in thedeveloping visual cortex. Neuron 30:275–87.

Frank MG, Jha SK, Coleman T. 2006. Blockade of post-synaptic activ-ity in sleep inhibits developmental plasticity in visual cortex.Neuroreport. Forthcoming.

Graves LA, Heller EA, Pack AI, Abel T. 2003. Sleep deprivation selec-tively impairs memory consolidation for contextual fear condition-ing. Learn Mem 10:168–76.

Guan Z, Peng X, Fang J. 2004. Sleep deprivation impairs spatial mem-ory and decreases extracellular signal-regulated kinase phosphory-lation in the hippocampus. Brain Res 1018:38–47.

Hairston IS, Little MTM, Scanlon MD, Barakat MT, Palmer TD,Sapolsky RM, and others. 2005. Sleep restriction suppresses neuro-genesis induced by hippocampus-dependent learning. J Neurophysiol94:4224–33.

Harrison Y, Horne JA. 2000. The impact of sleep deprivation on deci-sion making: a review. J Exp Psychol Appl 6:236–49.

He H-Y, Hodos W, Quinlan EM. 2006. Visual deprivation reactivatesrapid ocular dominance plasticity in adult visual cortex. J Neurosci26:2951–5.

Hoffman KL, McNaughton BL. 2002. Coordinated reactivation of dis-tributed memory traces in primate cortex. Science 297:2070–3.

Horne JA, McGrath MJ.1984. The consolidation hypothesis for REMsleep function: stress and other confounding factors—a review.Biol Psychol 18:165–84.

Hubel DH, Wiesel TN. 1970. The period of susceptibility to the phys-iological effects of unilateral eye closure in kittens. J Physiol206:419–36.

Huber R, Ghilardi MF, Massimini M, Tononi G. 2004. Local sleep andlearning. Nature 430:78–81.

Jha SK, Jones BE, Coleman T, Steinmetz N, Law C, Griffin G, andothers. 2005. Sleep-dependent plasticity requires cortical activity. J Neurosci 25:9266–74.

Kudrimoti HS, Barnes CA, McNaughton BL. 1999. Reactivation ofhippocampal cell assemblies: effects of behavioral state, experi-ence and EEG dynamics. J Neurosci 19:4090–101.

Malenka RC, Bear MF. 2004. LTP and LTD: an embarrassment ofriches. Neuron 44:5–21.

Maquet P. 2001. The role of sleep in learning and memory. Science294:1048–51.

Maquet P, Schwartz S, Passingham R, Frith C. 2003. Sleep-related con-solidation of a visuomotor skill: brain mechanisms as assessed byfunctional magnetic resonance imaging. J Neurosci 23:1432–40.

Oksenberg A, Shaffery JP, Marks GA, Speciale SG, Mihailoff G,Roffwarg HP. 1996. Rapid eye movement sleep deprivation in kit-tens amplifies LGN cell-size disparity induced by monocular dep-rivation. Dev Brain Res 97:51–61.

Peigneux P, Laureys S, Delbeuck X, Maquet P. 2001. Sleeping brain,learning brain: the role of sleep for memory systems. Neuroreport12:A111–24.

Peigneux P, Laureys S, Fuchs S, Collette F, Perrin F, Reggers J, andothers. 2004. Are spatial memories strengthened in the human hip-pocampus during slow wave sleep? Neuron 44:535–45.

Peigneux P, Laureys S, Fuchs S, Destrebecqz A, Collette F, DelbeuckX, and others. 2003. Learned material content and acquisition levelmodulate cerebral reactivation during posttraining rapid-eye-movements sleep. Neuroimage 20:125–34.

Peigneux P, Orban P, Balteau E, Degueldre C, Luxen A, Laureys S, andothers. 2006. Offline persistence of memory-related cerebral activ-ity during active wakefulness. PLoS Biol 4:e100.

Pelletier JG, Apergis J, Pare D. 2004. Low-probability transmission ofneocortical and entorhinal impulses through the perirhinal cortex.J Neurophysiol 91:2079–89.

Perrett SP, Dudek SM, Eagleman D, Montague PR, Friedlander MJ.2001. LTD induction in adult visual cortex: role of stimulus timingand inhibition. J Neurosci 21:2308–19.

Plihal W, Born J. 1999. Effects of early and late nocturnal sleep onpriming and spatial memory. Psychophysiology 36:571–82.

Plihal W, Krug R, Pietrowsky R, Fehm HL, Born J. 1996.Corticosteroid receptor mediated effects on mood in humans.Psychoneuroendocrinology 21:515–23.

Rauchs G, Desgranges B, Foret J, Eustace F. 2005. The relationshipsbetween memory systems and sleep stages. J Sleep Res 14:123–40.

Ribeiro S, Gervasoni D, Soares ES, Zhou Y, Lin SC, Pantoja J, and oth-ers. 2004. Long-lasting novelty-induced neuronal reverberation dur-ing slow-wave sleep in multiple forebrain areas. PLoS Biol 2:E24.

Ribeiro S, Goyal V, Mello CV, Pavlides C. 1999. Brain gene expres-sion during REM sleep depends on prior waking experience. LearnMem 6:500–8.

Ribeiro S, Mello CV, Velho T, Gardner TJ, Jarvis ED, Pavlides C. 2002.Induction of hippocampal long-term potentiation during wakingleads to increased extrahippocampal zif-268 expression during ensu-ing rapid-eye-movement sleep. J Neurosci 22:10914–23.

Robertson EM, Pascual-Leone A, Press DZ. 2004. Awareness modifiesthe skill-learning benefits of sleep. Curr Biol 14:208–12.

Roffwarg HP, Muzio JN, Dement WC. 1966. Ontogenetic develop-ment of the human sleep-dream cycle. Science 152:604–19.

Rosanova M, Ulrich D. 2005. Pattern-specific associative long-termpotentiation induced by a sleep spindle-related spike train. J Neurosci 25:9398–405.

Ruskin D, Dunn K, Billiot I, Bazan N, Lahoste G. 2005. Eliminatingthe adrenal stress response does not affect sleep deprivation-induced acquisition deficits in the water maze. Life Sci 78:2833–8.

Shaffery JP, Lopez J, Bissette G, Roffwarg HP. 2005. Rapid eye move-ment sleep deprivation revives a form of developmentally regu-lated synaptic plasticity in the visual cortex of post-critical periodrats. Neurosci Lett 391:131–5.

Shaffery JP, Roffwarg HP. 2003. Rapid eye-movement sleep depriva-tion does not “rescue” developmentally regulated long-term poten-tiation in visual cortex of mature rats. Neurosci Lett 342:196–200.

Shaffery JP, Roffwarg HP, Speciale SG, Marks GA. 1999. Ponto-geniculo-occipital wave suppression amplifies lateral geniculatenucleus cell-size changes in monocularly deprived kittens. DevBrain Res 114:109–19.

Shaffery JP, Sinton CM, Bissette G, Roffwarg HP, Marks GA. 2002.Rapid eye movement sleep deprivation modifies expression of long-term potentiation in visual cortex of immature rats. Neuroscience110:431–43.

Siegel JM. 2001. The REM sleep-memory consolidation hypothesis.Science 294:1058–63.

Smith C. 1985. Sleep states and learning: a review of the animal liter-ature. Neurosci Biobehav Rev 9:157–68.

Smith C. 1995. Sleep states and memory processes. Behav Brain Res69:137–45.

Smith C, MacNeill C. 1994. Impaired motor memory for a pursuitrotor task following stage 2 sleep loss in college students. J SleepRes 3:206–13.


Smith CT, Nixon MR, Nader RS. 2004. Posttraining increases in REMsleep intensity implicate REM sleep in memory processing andprovide a biological marker of learning potential. Learn Mem11:714–9.

Steriade M, Timofeev I. 2003. Neuronal plasticity in thalamocorticalnetworks during sleep and waking oscillations. Neuron 37:563–76.

Stickgold R. 2005. Sleep-dependent memory consolidation. 437:1272–8.

Stickgold R, LaTanya J, Hobson JA. 2000. Visual discrimination learn-ing requires sleep after training. Nat Neurosci 3:1237–8.

Stickgold R, Walker MP. 2005. Sleep and memory: the ongoingdebate. Sleep 28:1225–7.

Stickgold R, Whifbee D, Schirmer B, Patel V, Hobson JA. 2000. Visualdiscrimination task improvement: a multi-step process occurringduring sleep. J Cogn Neurosci 12:246–54.

Tononi G, Cirelli C. 2003. Sleep and synaptic homeostasis: a hypoth-esis. Brain Res Bull 62:143–50.

Vertes RP, Eastman KE. 2000. The case against memory consolidationin REM sleep. Behav Brain Sci 23:867–6.

Vertes RP, Siegel JM. 2005. Time for the sleep community to take a crit-ical look at the purported role of sleep in memory processing. Sleep28:1228–9.

Walker MP, Brakefield T, Morgan A, Hobson JA, Stickgold R. 2002.Practice with sleep makes perfect: sleep-dependent motor skilllearning. Neuron 35:205–11.

Wilson MA, McNaughton BL. 1994. Reactivation of hippocampalensemble memories during sleep. Science 265:676–82.

Wright JW, Reichert JR, Davis CJ, Harding JW. 2002. Neural plastic-ity and the brain renin-angiotensin system. Neurosci Biobehav Rev26:529–52.


Documents

The Role of Sleep in Memory Consolidation and Brain Plasticity - Dream or Reality