ARTICLE

Sleep Intensity and the Evolution of HumanCognitionDAVID R. SAMSON AND CHARLES L. NUNN

Over the past four decades, scientists have made substantial progress inunderstanding the evolution of sleep patterns across the Tree of Life.1,2 Remark-ably, the specifics of sleep along the human lineage have been slow to emerge.This is surprising, given our unique mental and behavioral capacity and theimportance of sleep for individual cognitive performance.3–5 One view is that ourspecies’ sleep architecture is in accord with patterns documented in other mam-mals.6 We promote an alternative view, that human sleep is highly derived rela-tive to that of other primates. Based on new and existing evidence, wespecifically propose that humans are more efficient in their sleep patterns thanare other primates, and that human sleep is shorter, deeper, and exhibits ahigher proportion of REM than expected. Thus, we propose the sleep intensityhypothesis: Early humans experienced selective pressure to fulfill sleep needs inthe shortest time possible. Several factors likely served as selective pressuresfor more efficient sleep, including increased predation risk in terrestrial environ-ments, threats from intergroup conflict, and benefits arising from increasedsocial interaction. Less sleep would enable longer active periods in which toacquire and transmit new skills and knowledge, while deeper sleep may be criti-cal for the consolidation of those skills, leading to enhanced cognitive abilities inearly humans.

Sleep occupies approximately one-third of a typical human life span.Deviations from this standard arelinked to cognitive impairment andnegative health consequences. For

example, sleep is critical for workingmemory, attention, decision-making,and visual-motor performance.7 Sleeploss, driven by access to artificiallighting, shift-work, and increased

international travel, has societalcosts, ranging from decreases inworkplace productivity to fatal acci-dents. Occupations that demandhigh-level cognitive function duringshift work or that restrict sleep areparticularly relevant.8 Humans in thedeveloped world sleep in vastly differ-ent ways than our hominin ancestorsslept.3 These differences may haveimportant consequences for globalhealth and treatment of sleepdisorders.4,9

Understanding human sleep alsohas implications for understandinghuman evolution,5 which is the focusof our paper. To synthesize currentunderstanding of human sleep ecol-ogy and evolution, we turn to theethnographic and historical litera-tures and recent studies to synthe-size findings related to sleep insmall-scale, subtropical, noncontra-ceptive human populations that lackready access to artificial light (hence-forth called traditional populations).In addition, using new phylogeneticcomparative methods, we investigateprimate sleep and identify uniqueaspects of human sleep. Building onrecent ideas concerning the impor-tance of the tree-to-ground transitionin hominin sleep and cognition,5,10

we argue that the transition to obli-gate terrestrial environments mayhave been a consequence of allomet-ric scaling. This transition mayexplain resulting physiological andtechno-cultural adaptations, such asbeds, shelters, controlled use of fire,variation in chronotypes (see glos-sary), and large social groups thatgave early members of the genusHomo the advantage of deep, effi-cient sleep. We suggest that changes

David R. Samson is a Postdoctoral Associate in the Department of Evolutionary Anthropol-ogy at Duke University. He has conducted field research on wild apes in Uganda, captiveapes, and small-scale human societies in Madagascar. His main research focuses on theecology and behavior of primates, with a specific interest in the evolution of sleep andcognition in the human lineage. Email: david.samson@duke.edu

Charles L. Nunn is a Professor of Evolutionary Anthropology and Global Health at DukeUniversity, and Director of the Triangle Center for Evolutionary Medicine in the ResearchTriangle region of North Carolina. His research focuses on evolutionary approaches tounderstand and improve human health, including research on the ecology and evolution ofinfectious disease, evolution of sleep architecture, and the links between ecology, evolu-tion, and global health.

Key words: comparative study; human uniqueness; human evolution; allometry; evolutionary mismatch

VC 2015 Wiley Periodicals, Inc.DOI: 10.1002/evan.21464Published online in Wiley Online Library(wileyonlinelibrary.com).

Evolutionary Anthropology 24:225–237 (2015)

in sleep were central to the fabric ofhuman evolution even though, todate, the study of sleep in traditionalsocieties and nonhuman primateshas received remarkably littleattention.3,11–13

INVESTIGATING UNIQUE ASPECTSOF HUMAN SLEEP

Sleep can be viewed as a brainstate, a process, and a behavior;14,15

it is an emergent property of thebrain that serves several purposes,including energy restoration, immu-nocompetence, brain metabolichomeostasis, neural ontogenesis, andcognitive and emotional process-ing.16–18 Sleep is regulated byhomeostatic19 and circadian mecha-nisms.20 Consequently, the more wego without it the more we need it(homeostatic drive). Similarly, asnight falls, physiological mechanismssuch as melatonin release are acti-vated in diurnal animals to facilitatesleep (the circadian drive).

When asleep, the brain shiftsbetween qualitatively and quantita-tively different states, nonrapid eye

movement (NREM) and rapid eyemovement (REM) sleep.21 NREMsleep is subdivided into two impor-tant stages. Light N2 (NREM stages1-2) is accompanied by sleep spin-dles and K-complexes and is associ-ated with the lowest arousalthreshold; that is, a sleeper is mosteasily awakened from this stage. Thesecond stage of NREM sleep is deepN3 slow-wave activity (SWA) (NREMstage 3), characterized by deltarhythms and slow, global corticaloscillations. Deep N3 sleep, as com-pared to light sleep, is associatedwith a high arousal threshold, mak-ing it is more difficult to awaken asleeper from this stage.22,23

REM sleep, in contrast, has anelectroencephalographic (EEG) pat-tern that, despite its association withcomplete behavioral paralysis, indi-cates brain activity that is similar toan awake state. In general, EEGsduring REM sleep show faster thetarhythms, which arise from bidirec-tional subcortical, hippocampal, andcortical network interactions.24,25

REM sleep can be subclassified intotwo modes, tonic and phasic. Tonic

REM sleep refers to the state ofwidespread theta rhythms, whereasphasic REM is characterized byactual rapid eye movements (REMs)associated with ponto-geniculo-occipital (PGO) waves.26 Impor-tantly, as noted by Ermis and col-leagues27 external stimuli are heavilyinhibited during phasic REM, whichthey describe as a state, “with maxi-mal environmental shielding (discon-nection from the external world),[and] hence a most vulnerable phaseof sleep.” In summary, modernhuman sleep research has revealedthree discrete sleep stages: Light N2sleep, deep N3 slow wave activity,and REM (tonic and phasic)sleep.14,27

Is Human Sleep Flexible?

Perspectives promoted by sleephygiene (that is, behaviors that areconducive to habitually sleepingwell) largely focus on the importanceof consolidated sleep at consistentintervals from one 24-hour period tothe next.28 Yet we all have experi-ence with staying up too late, and

GLOSSARY

Actigraphy — a noninvasivemethod that measures gross motoractivity and, using a small actimetrysensor, monitors rest-activity cycles.

Chronotype — a behavioral pro-pensity to sleep during a particularphase during a circadian period,often described as “owls” or“eveningness” (delayed sleep period)versus larks or “morningness”(advanced sleep period).

Delta-rhythm — a high-amplitudebrain wave with a frequency oscilla-tion between 0–4 hertz and associ-ated with deep, slow-wave sleep.

K-complexes — an electroenceph-alographic waveform that occursduring stage 2 NREM and serve tosuppress cortical arousal and aidsleep-based memory consolidation.

Phasic REM — a state of REMsleep characterized by greaterarousal threshold compared to

tonic REM and associated with dis-tinct oculomotor activity (that is,rapid eye movements) and cardio-respiratory irregularities.

Polyphasic sleep — a behaviorof multi-phase sleep periods, usu-ally two (biphasic sleep).

Sleep efficiency — the total timespent asleep divided by the total timespent in a sleeping environment.

Sleep fragmentation — thenumber of awakenings greater orequal to two minutes per hour ofsleep time.

Sleep intensity — a homeostaticmechanism that regulates sleepdepth. Greater sleep intensity, andtherefore sleep depth, varies in thecourse of sleep and usually occursshortly after sleep onset.

Sleep motor activity — Thenumber of motor activity move-ments per hour of sleep time.

Sleep spindles — burst of oscil-latory brain activity occurring dur-ing stage 2 sleep; it is visible on anEEG consisting of 12–14 hertzwaves.

Theta rhythms — Electroen-cephalographic oscillations associ-ated with the hippocampus in the4-8 hertz frequency range.

Tonic REM — A state of wide-spread, low-voltage, fast electro-cortical activity associated with alower arousal threshold than pha-sic REM and hippocampal theta, adecrease in neck and chin electro-myogram amplitude, and braintemperature elevation.

226 Samson and Nunn ARTICLE

many people have sleep disordersthat affect the quality of sleep andnext-day performance.29 Are thesepatterns of late nights and disruptedsleep typical only of modern humansor are they characteristic of allhumans and throughout human evo-lution? In other words, are today’ssleep patterns a case of evolutionarymismatch to ancestral environmentsor actually characteristic of the selec-tive pressures on human sleep? Thestudy of small-scale modern humanforager populations living in adap-tively relevant habitats is essentialfor addressing these questions.

Information on human sleep insettings without artificial (that is,electrically produced) light is accessi-ble from historical and ethnographicrecords,4,30,31 along with a handfulof more recent studies that moredirectly quantify sleep patterns.13,32

Worthman and Melby3 argued that,compared to their postmodernindustrial counterparts, traditionalsocieties are characterized by strik-ingly different sleep ecology andbehavior. As summarized in Table 1,equatorial hunter-gatherer sleepenvironments are characterized by apattern that is more similar to that

of other primates: co-sleeping, inwhich individuals sleep within closeenough proximity to monitor eachother using two or more senses;33

daytime napping, often during thedaily extremes in temperature; audioconditions that are acousticallydynamic and often noisy; and light-ing conditions that are generally dimor dark. Security from large huntingpredators and smaller blood-suckingarthropods is achieved by sleepingsocially and through use of fire.

Arguably, the most significantbehavioral facilitator of sleep qualityis the environment where the individ-ual sleeps; sleep sites encompassextremes in temperatures, noise, andsleepers’ familiarity with their sur-roundings.34,35 Modern equatorialhunter-gatherer sleeping platformsinclude organic substrates and coversto facilitate thermoregulation and,provide a better quality of sleep.36

These substrates may take the formof piles of vegetation constructedfrom branches, lianas, leaves, andgrasses, which are sometimes inter-woven and usually accompanied byanimal hides.37 In addition to fire,hunter-gatherers display a variety ofanti-predation defenses, including

semi-permanent shelter structuresand earthworks dug into the groundto form concealed concavities inwhich to sleep.3,38,39 Although earlystudies showed Australian aboriginalpopulations to be characterized bygreater metabolic tolerance to bodycooling,40 it still is not known howthese adaptations help cope with thestresses of dynamic terrestrial sleepenvironments.

Hunter-gatherer sexual partnersgenerally co-sleep. Accordingly, themale may gain increased opportunityfor care of nearby offspring andreduce opportunities for infidelity byhis mate; in turn, females may gainincreased paternal care, includingprotection of offspring.41,42

In addition, evidence exists of agenetic underpinning to variation inhuman chronotypes.43 Relative toother mammals, human sleep timingis highly variable and these extremesare measured by the timing of sleeponset, which informs chronotype(colloquially dubbed morning “larks”versus evening “owls”).44 Naturalsentinels could also increase thegroup survival of terrestrial sleepers.The elderly exhibit less slow-wavesleep, lower thresholds for

TABLE 1. Hominoid Sleep Ecology: Overview of Ape, Human Hunter-gatherer, and Post-industrial Sleep Environmentsa

Great ape Hunter-gatherer Post-industrial

Chronology 18-14 mya to present 1.8 mya to present Nineteenth century to presentSleeping platform Arboreal sleeping platforms

made of foliageFoliage, animal hide Padded bed, profuse sleeping

accoutrementsSleep group sizea 5 26 1-2Diurnal inactivity Fluid Fluid RigidFire Absent Present AbsentSleep onset Rigid (sunset) Fluid ScheduledWake onset Rigid (sunrise) Rigid (sunrise) ScheduledLux Dark/dim (source: moonlight) Fire, moonlight Artificially controlledAcoustics Dynamic (fauna, conspecifics) Fire ambient noise;

dynamic (fauna,group members)

Silent, environmentally buffered

Security Arboreal platforms, group size,insect repellant/odor maskingproperties of nests

Fire, group size,defensive structures,sentinels, males in proneposition closest topotential threats,mother-infant co-sleeping

Environmentally bufferedthrough complex domicileconstruction

Thermoregulation Sleep platform complexity,foliage; mother-infantco-sleeping

Fire, shelter, mother-infantco-sleeping, group sleepduring temperature nadir

Closed domicile, temperatureregulation via blankets andmodification of ambienttemperature.

aAveraged great ape group size71 and hunter-gatherer group size109

ARTICLE Sleep Intensity and the Evolution of Human Cognition 227

awakening, and more awakeningsduring a single sleep period.45 Chil-dren and adults with attention deficithyperactivity disorder (about 3%-5%of a population) may also have alower threshold and greater fre-quency of awakenings.46 Variabilityin chronotype, resulting in sleep sen-tinels, could have led to increasednight-time vigilance and survivability,possibly via group selection, as a wayof countering predation or defendingagainst night raids by rival groups.

The question of “normal” humansleep phasing has received attentionfrom both empirical and historicalresearchers. Ekirch47 used historicalrecords as the basis for his proposalthat biphasic sleep was common inEuropean civilizations before theIndustrial Revolution. In addition,experiments suggested that humansfrom Western populations can revertto a biphasic sleep pattern underconditions of long nights.48 FromWehr’s48 research emerges the“human sleep flexibility hypothesis,”according to which humans showcircadian rhythms and behavioralstrategies that facilitate flexible, pol-yphasic sleep patterns. The alterna-tive sleep consolidation hypothesisforwards the idea that a single, inte-grated sleep period, with a low fre-quency of daytime napping, bestcharacterizes traditional populations(Fig. 1) and therefore represents“normal” ancestral sleep.

In support of the sleep consolida-tion hypothesis, Yetish and col-leagues13 recently used actigraphy toquantify natural sleep patterns in

three traditional populations in Tan-zania, Namibia, and Bolivia. Strik-ingly, these populations exhibitedsimilar sleep parameters, with anaverage duration of 5.7-7.1 hours.Typically, sleep onset would occurseveral hours after sunset, with after-hours activity facilitated by firelight.However, wake onset, which wasassociated with sunrise, was lessvariable. The sleep efficiency in thesethree traditional populations wasbetween 81% and 86%, and thuscomparable to that in industrial pop-ulations. Sleep consistently occurredduring the period of night with thelowest temperature and individualsslept longest during the times of yearwith the coolest temperatures. Thesefindings highlight the role of envi-ronmental factors in sleep patterns.In summary, with temperature pro-posed as a major regulator of sleepduration and timing, data generatedin these equatorial populations sup-ports the hypothesis that ancienthumans’ sleep was consolidated intoone major sleep bout per 24 hours.

Multiple approaches exist to inves-tigate the phasing of human sleepmore quantitatively. First and fore-most, more data are needed on sleeppatterns in humans that lack accessto artificial light. This includes notonly hunter-gatherers and small-scale agriculturalists who live inequatorial habitats, but also popula-tions in high-latitude arctic areascharacterized by long summer day-light hours with moderate tempera-tures and long, cold winters withextremely low light exposure or no

light exposure. In addition, popula-tions in developing countries, whichexperience emerging “evolutionarymismatch” situations, in which theirrapidly changing environments differradically from ancestral environ-ments, may exhibit sleep phasingthat differs from that in traditionaland/or postindustrial populations.

Similarly, more research is neededon natural patterns of variation inape sleep, especially activity at night.Sleep research on apes has recentlygained traction because data-loggingequipment and infrared cameras canrecord vocalizations and activityaround great ape sleep sites in bothwild49,50 and captive sites.51 Compar-ative approaches are also needed torigorously measure sleep intensity ina broad set of primates and to placehumans in that comparative context.We next turn to an example of use ofthis comparative approach to deter-mine different aspects of humansleep architecture.

Sleep Duration, Depth, andPercentage of REM inComparative Perspective

Human sleep, when compared on agross level to that of other primates,appears to be characterized by severalunique traits. Based on a 7.0-hoursleep duration estimated in multiplepopulations over 66 studies, withthese populations cross-sectioned atfive-year intervals from the ages of 15through 45 years,32,45 human sleepduration is the shortest recordedamong primates (Fig. 2A). This is instark contrast to the primate“marathon sleepers” (for example, owlmonkeys, cotton-top tamarins, andmouse lemurs), which have total sleeptimes ranging from 13-17 hours.11 Thehuman REM to NREM ratio (22:78) isthe highest proportion of REM toNREM of all primates (Fig. 2B).

While these patterns are intrigu-ing, a more quantitative phylogeneticapproach is needed to assess thedegree to which human sleep differsfrom that of other primates.52 Weused such an approach – an evolu-tionary outlier analysis – to assesswhether human sleep is extraordi-nary as compared to variation

Figure 1. Is human sleep flexible? (A, B) Polyphasic sleep phasing, compared to (C), the rel-atively consolidated monophasic sleep characterized by populations living in postindus-trial developed economies. Compared to traditional populations, recent work points to asurprising convergence in the consolidation of “modern” sleep phasing with inferredancient patterns.

228 Samson and Nunn ARTICLE

among primates in general.53,54 Thisanalysis combines phylogeny withdata on predictor variables to makea phylogenetically informed predic-tion of a response variable for a par-ticular species on the phylogeny(Box 1). As shown in Box 1, ouranalyses demonstrated that humanshave exceptionally short sleep (Fig.3) with a greater proportion of REM(Fig. 4) than would be predicted fora primate of similar phenotypiccharacteristics.

Because of the challenges of usinginvasive EEG with endangered spe-cies, only 7% of primate species havehad their sleep architecture quanti-

fied.55 Even fewer have had detaileddescriptions of sleep environmentand species-specific sleeping pos-tures documented. Specifically, thereis a dearth of data recording the fullrange of critical sleep measures –light N2 sleep, deep N3 sleep, andREM – and behavioral variables suchas sleep efficiency, sleep motor activ-ity, and sleep fragmentation (seeGlossary). Therefore, variables thatare critically important for under-standing sleep intensity are currentlylacking for primates. As we will dis-cuss, given the link between sleephomeostasis and motor activity,56

several variables drawn from actigra-

phy may serve as behavioral proxiesfor deep sleep, thus opening the doorfor future studies of these variablesand their correlates.

ALLOMETRIC SCALING ANDECOLOGY OF SLEEP SITES

Humans are also characterized bysleeping terrestrially, which is unusualamong primates. Other primates willoccasionally sleep terrestrially, or someindividuals in some populations willshow more regular ground sleeping.For example, where predation isregionally low for chimpanzees, asmall proportion of males has beenobserved to sleep on the ground.57 Sim-ilarly, physically massive male gorillasthat live in predator-poor environ-ments often sleep on the ground.58 Butit is only in humans that all age and sexclasses habitually sleep terrestrially.

As suggested by these examples ofgorilla and chimpanzee male sleep pat-terns, physical characteristics and ecol-ogy are important in predicting primatesleep sites. Indeed, a consideration ofallometry helps shed light on someaspects of primate sleep site patterns.In a simple geometric model, the dou-bling of the length of an animal corre-sponds to disproportionate increases inarea and volume.59 The exponentialincrease in volume relative to staturecreates physical limits on where a pri-mate can successfully obtain the bene-fits of sleep.60

To investigate allometric scaling ofsleep-site use, we have identifiedfour major sleep-site categories thatprimates use to overcome the chal-lenges of finding security and com-fort (Box 2): fixed-point nests, treebranches, arboreal sleeping plat-forms, and terrestrial beds (Fig. 5).Each of these is related to what wecall a “sleep-scaling threshold” as away to overcome the exponentialincrease in body volume.

The ancestral primate was likelyarboreal,61 a way of life that presentsmajor challenges to most primates inlocating comfortable and secure sleepsites. Small-bodied primates oftenuse tree-holes or circular leaves thatmesh together to avoid predators andmosquito vectors that spread para-sites, as well as to thermoregulate,provision, and care for young.62–64

Figure 2. Human sleep relative to that of nonhuman primates. Humans have (A) theshortest total sleep times and (B) the greatest proportion of REM relative to total sleeptime. We investigate these patterns phylogenetically and quantitatively in Box (see Figs.(3 and 4).

ARTICLE Sleep Intensity and the Evolution of Human Cognition 229

Evolutionary reconstruction of pri-mate life-history traits points to anancestral sleep state that most likelyresembled that of extant galagos, asolitary, nocturnal animal that pro-duces a single offspring provisionedin a fixed-point nest.63 Early nest-likesleeping shelters may primarily haveprotected against predation and bit-ing insects,64 conferred a thermoreg-ulatory buffer, and increased overallsafety by reducing the risk of fallingfrom arboreal sleep sites.65 These val-

uable sites, especially tree-holes,would have required time to locateand potentially secure from other ani-mals or, in the case of leaves andbranches, to construct or parasitize.Open nests may provide some ofthese benefits but, when closed sitesare available, they considerablyenhance these benefits.

Paleocene and Eocene primates’body size, like that of many othermammals, steadily increased throughtime.66,67 As primate body mass

expanded beyond the capacity ofmost fixed-point nests, a major tran-sition occurred from fixed-point nestsleep to tree-branch sleep. Abandon-ment of the fixed-point nest sleep-sitestrategy in divergent primate lineagesmay also have been a result of thechange from nocturnal to diurnalactivity patterns, which resulted inlarger social groups as a defenseagainst diurnal predators.68 From themeasurements of postural behaviorof the few primates that have been

Box 1. Phylogenetic Outlier Analysis

A major challenge for compara-tive methods is to investigatechange in a quantitative trait alonga single branch of a phylogeny, asone might wish to do for humans,while also controlling for variationin factors that explain variation inthe trait of interest.52 A standardway to achieve this is by regressinga response variable on other traitsand testing whether humans are“outliers,” as has been done instudies of brain size.105 It is impor-tant in this context also to controlfor phylogeny, in terms of both theunderlying regression model andassessing whether humans are out-liers.106 For example, if apes lieabove the regression line, we mightexpect that humans do, too, sug-gesting that it is something aboutapes (not just humans) that differsfrom other primates.

We used recent Bayesian phyloge-netic methods53 for statistical model-ing of sleep durations in primates,then used the statistical model topredict sleep duration in humans.Sleep durations were used only ifthey passed the minimum require-ments for data quality as detailed inprevious work that compiled sleepquotas for phylogentic analy-sis.107,108 By comparing actual sleepduration in humans to the predictedoutcomes from the model, we caninvestigate whether humans are atypical primate (our observed sleepduration45 falls within the predictedinterval) or a “phylogenetic outlier”(our sleep duration falls outside the

predicted interval). More specifically,we estimated regression coefficientsand the degree of phylogenetic signalin the data using Markov chainMonte Carlo (MCMC) approachesand implemented Bayesian modelselection, which includes predictorvariables in the statistical model inproportion to their posterior proba-bility (body mass was forced into themodel at all iterations in the MCMCchain).53 When phylogenetic signalin the residuals from the statisticalmodel is high (estimated with thescaling parameters k and j),52 thehuman value is shifted based onphylogeny to reflect values of sleepduration in our close phylogeneticrelatives. We used a Bayesian poste-rior probability distribution of mod-els and phylogenies to generate aBayesian posterior prediction distri-bution, which is what would be pre-dicted in a primate with phenotypiccharacteristics similar to those ofhumans. We determined thathumans are exceptional if theobserved human value falls outsidethe 95% credible interval.

Analyses of sleep durationshowed that humans are exception-ally short sleepers, with the humanvalue (shown as a line in Figure 3)substantially below the 95% credi-ble interval of predicted sleep dura-tions (only 1 of 500 samples fromthe posterior prediction was moreextreme than observed). In fact,however, most of the coefficientsrelating these traits to total sleeptime in primates were not different

from zero, and often not includedin the model. Our predictions takeinto account that uncertainty aboutwhich variables to include in themodel, along with uncertaintyabout primate phylogeny. We alsofound good evidence of phyloge-netic signal; the branch length scal-ing parameter k for total sleep timehad a mean of 0.69 (SD 5 0.28).

We next examined the proportionof REM sleep across primates toassess whether humans are a phy-logenetic outlier in this regard aswell. Analyses of the proportion ofREM sleep also showed thathumans are a striking outlier (Fig.4), with the human REM propor-tion substantially above the 95%credible interval of predicted REMproportion (only 2 of 500 samplesin the posterior probability distri-bution were greater than theobserved value for humans). Again,we found good evidence of phylo-genetic signal based on the findingthat the branch-length scalingparameter k for REM proportionhad a mean of 0.59 (SD 5 0.27).

In summary, taking phylogenyand primate ecology into account,human sleep differs substantiallyfrom that of other primates: Weare exceptionally short sleepers andwe pack a higher proportion ofREM sleep into our short sleepdurations. It appears that evolutionhas whittled away sleep durationsalong our lineage, just as access toelectricity and lighting continues todo in the present day.

230 Samson and Nunn ARTICLE

recorded during sleep bouts, it isobserved that they typically huddletogether in a guarded position. Thisdynamic and precarious sleep envi-ronment may explain why, as com-pared to apes, monkeys have lessefficient, lighter sleep.69

Great apes are characterized by auniversal behavior that has helpedthem solve the challenge of sleepingsecurely and comfortably in the treesin spite of their massive bodies. Theyconstruct a “nest” or sleeping plat-form.70,71 Great apes build a new

sleeping platform each night, specifi-cally selecting trees for their firm,stable, and resilient biomechanicalproperties.72–74 In contrast, the lesserapes, the gibbons, do not build sleep-ing platforms; instead, they followthe ancestral primate pattern ofsleeping on tree branches, typicallylying or sitting on what is availablewithout altering their local environ-ment.75 Arboreal platforms provideseveral benefits to sleepers, includingkeeping individuals out of range ofterrestrial predators,76–78 repellingblood-sucking arthropods and/or ma-sking individual insect-attractingodors,64,79–81 providing added insula-tion to keep warm,65,81 and providing amore stable and secure environment.72

Phylogenetic reconstruction placesthe emergence of nest building some-time between 18 -14 mya.82 The inno-vation of ape nest construction,coinciding with the evolution of bodymass over the 30 kg threshold (Fig.6), suggests that larger body massmade sleeping on branches lessadvantageous for apes. At this size,apes would have benefited from moreresilient and stable sleeping sub-strates to reduce both physical stresson the body and the probability oflethal falls. This evolutionary eventcould have then established the pre-requisite adaptations to alter sleeparchitecture within the Hominidae.

In addition to the challenges of alarger-bodied animal sleeping onbranches, cognitive demands in greatapes may have favored nest-building.In particular, more stable sleepingsites provide physical support forlarge-bodied hominoids to maintaindeep and sustained sleep,60,83 whichmay be linked to enhanced cognitivefunction in the great apes.71 Thisidea has become known as the sleep-quality hypothesis. The alternative“engineering hypothesis” switchescausality to suggest that the greatercognitive performance of great apesenables them to build nests.83 Ratherthan viewing these two hypothesesas mutually exclusive, it could bethat increased complexity in sleepingplatform construction could havepositively affected cognition, whichthen enhanced nest building poten-tial, resulting in a positive feedbackloop.

Figure 3. We conducted a Bayesian analysis to predict total sleep duration based onphylogeny, body mass, activity period, endocranial volume, percentage of leaves in thediet, interbirth interval, and foraging group size. The distribution (top) reflects the posteriorprobability for predicted sleep duration relative to the observed variation in sleep dura-tions (bottom); the vertical line represents the observed human value. Because theobserved value falls outside the 95% credible interval, we conclude that human sleeplength is different from what one would predict for a primate with our traits and phyloge-netic position. In other words, humans, relative to other primates, have extremely shorttotal sleep durations. [Color figure can be viewed in the online issue, which is availableat wileyonlinelibrary.com.]

Figure 4. We conducted a Bayesian analysis to predict REM proportion based on phylog-eny, body mass, activity period, endocranial volume, percentage of leaves in the diet,interbirth interval, and foraging group size. The distribution (top) reflects the posteriorprobability for predicted REM proportion relative to the observed variation in REM pro-portion (bottom); the vertical line represents the observed human value. Because theobserved value falls outside the 95% credible interval, we conclude that human REM isdifferent from what one would predict for a primate with our traits and phylogeneticposition. In other words, humans, relative to other primates, have an extremely long pro-portion of REM to total sleep time. [Color figure can be viewed in the online issue, whichis available at wileyonlinelibrary.com.]

ARTICLE Sleep Intensity and the Evolution of Human Cognition 231

A few recent studies have begun toinvestigate how nests and sleepenhance cognitive performance innonhuman great apes. First, Samsonand Shumaker84 provided captiveorangutans with different materialsto construct a night nest. They quan-

tified the sleeping platform complex-ity each night, measuring it as anindex of the number of material itemsavailable to construct a bed, andfound that complexity covaried posi-tively with reduced night-time motoractivity, less fragmentation, and

greater sleep efficiency. Second, incaptive apes undergoing experimen-tal cognitive testing, sleep has beenshown to stabilize and protect memo-ries from interference.85 Futureresearch could investigate these ques-tions with similar approaches,

Box 2. Sleep Sites and Body Mass

A primary driver of primate sleep-site selection is circumscribed notonly by mass but also volume (as anallometric scaling function of bodylength), relative to local ecologicalpressure and functional morphology.Large-bodied primates must positionthemselves on relatively smaller sup-

ports. Even if larger primates posi-tioned themselves on geometricallyproportional supports, stress on thebody increases disproportionately.As body length increases, weightincreases at a cube length, whereasthe surface of the body supportingweight increases at only a square.

Thus, large primates disproportion-ately stress the skin, connective tis-sue, and skeleton when lyinghorizontally. More importantly, therelatively constant size of supportingbranches would no longer be appro-priate for the task of securing a sleepsite (Figs. 5 and 6).

Figure 5. (A) Fixed-point nests are used to avoid predators, stay within optimal temperatures, and store resources and offspring (photoaccredited to Manfred Eberle). (B) Tree branches are used to avoid predators, but are less stable than other sleep sites. (C) Sleepingplatforms are universally used by large-bodied great apes as stable and secure sleep sites that are warm and confer resistance to bit-ing insects and arboreal predators (photo accredited to Kathelijne Koops). Terrestrial beds are used by massive apes (male chimpan-zees and gorillas) and humans (photo of Hadza hunter accredited to Mathiew Paley/paleyphoto.com). [Color figure can be viewed inthe online issue, which is available at wileyonlinelibrary.com.]

232 Samson and Nunn ARTICLE

combined with quantitative measuresof sleep depth and quality.

EARLY HUMAN SLEEP ECOLOGY

Given the dramatic morphologicalchanges and exponentially increasingbody volume that took place duringthe Australopithecus-Homo transi-tion, inferential evidence supports H.erectus as the first fully terrestrialground-sleeping hominin. Homo’spredecessor, Australopithecus, sharesseveral anatomical features (such asa narrow scapula and long, curvedphalanges) with the great apes; thesefeatures reduced structural fatigueand are clear indicators of arborealadaptation.86 In contrast, it is gener-ally accepted that H. erectus is thefirst obligate biped,87 with resultingmass and stature estimates (H. erec-tus males 5 1.8 m, 66 kg; females 51.6 m, 56 kg) that are comparable tothose of modern humans.88–90 There-fore, with limb proportions thatwould make it difficult to facilitatearboreal sleep, Homo was likely thefirst full-time ground sleeper.

Regardless of the specific timingof the transition to the ground, acritical question remains: Once onthe ground, how did our ancestorscope without arboreal sleeping sitesand the benefits they confer? Asnoted, arboreal platforms provide

many benefits that would have beenlost when sleeping on the ground.Thus, one might predict that move-ment from tree to ground sleepwould have increased risks from pre-dation by large-bodied animals suchas leopards, hyenas and saber-toothed cats;91 disease transmissionby blood-sucking arthropods such asterrestrially hunting mosquitoes;92

and reduced themoregulatory home-ostasis when in contact with the con-ductive properties of the earth’srelatively cooler temperature.

Thus, to make the terrestrial sleeptransition, early hominins would likelyhave evolved numerous behavioraladaptations to counteract these risksand the loss of benefits of arborealplatforms. With respect to early homi-nin sleep ecology, possession of firehas been proposed as essential forenabling sleep in terrestrial environ-ments.10 It has been proposed thatthe first of our hominin ancestors touse fire was H. erectus,93 and al-though the claim remains uncertainbecause of fragmentary archeologicalevidence, it has been argued that firewas necessary for the transition toobligate terrestriality.94 Fire at nightwould have helped deter predators,94

kept individuals warm during coldnights, and fumigated sleep sites withsmoke to deter biting insects3,4 (Table1). Moreover, fire may have increased

group cohesion and promoted medita-tion.95 Possible costs in the habitualuse of fire may have includedincreased rates of respiratory dis-ease96 and attraction of intraspecificcompetitors.

The details of this transition aredifficult to pin down in the archeo-logical record. There is evidence ofthe use of smoke to repel malariavectors in Amazonian societies.97

Tests aimed at investigating the useof fire in relation to controlling bit-ing insects could be one way for-ward. Additional data obtained byexperimentally manipulating accessto fire in an outdoor environment,alongside mobile polysomnographyand insect-trap measures, coulddirectly test fire’s role in sleep safetyand quality. Finally, similar datagenerated with hunter-gathererscould clarify the frequency withwhich fires are used and the benefitsobtained for thermoregulation andpredator avoidance.

SLEEP INTENSITY HYPOTHESIS

Building on these and other emerg-ing findings in evolutionary anthro-pology and sleep biology, we proposethe “sleep intensity hypothesis”,which postulates that early humanssleeping in novel terrestrial environ-ments had characteristic sleep archi-tecture that fulfilled homeostaticneed in the shortest time possible.Such a shift in sleep ecology couldexplain the unique characteristics ofhuman sleep, resulting in an overallpattern of efficient sleep expressionwith subsequent cognitive and behav-ioral advantages emerging from high-quality sleep and increased net hoursof activity along the human lineage.We present this idea by first providingsome additional background on sleepbiology, then consider selective pres-sures on human sleep.

Vyazovskiy and Delogu14 advancethe theory that NREM and REMwork in a complementary system,with NREM enabling informationprocessing, synaptic plasticity, andcellular maintenance during a gen-eral “recovery phase” and REM ena-bling periodic excursions into anactivated brain, identifying networksthat have undergone recovery from

Figure 6. The effect of allometric scaling illustrated by body mass (y-axis) and sleep-sitecategory (x-axis), and accounting for sex (white 5 female, grey 5 male), shows how pri-mates have adapted their sleep environments relative to evolutionary increases in bodymass. Data from species’ averages in Smith and Jungers.110

ARTICLE Sleep Intensity and the Evolution of Human Cognition 233

the previous NREM period, thus per-forming a “selection phase.” In otherwords, REM “tags” which parts ofthe brain are sufficiently stabilized,differentiating them from those thatneed more SWA processing.

The net result of NREM-REMcycling is an efficient process. Itensures that functional requirementsfor current cognitive demands are metin the shortest possible time.22 Indeed,Hal!asz and colleagues98 note, “It maybe that the full-blown development ofthese syngergic regulations is ahuman-specific trait which was neces-sary due to the vulnerability of thecognitive functions of the frontal neo-cortex.”98 In full-time terrestrial envi-ronments, hominins, sleeping in large,sentineled groups on stable groundbeds protected by fire, would havebeen uniquely positioned to capitalizeon the adaptive advantage of deeper,more intense, REM-dominated sleep.

The terrestrial sleep transitionwould have eliminated the signifi-cant dangers of arboreal activity fora diurnal animal at night, one of themost important being the danger oflethal falls by large-bodied individu-als. In this situation, early humanscould have dedicated a greater pro-portion of time to behaviorally vul-nerable, yet highly consolidated,sleep states with high arousal thresh-olds, such as N3 (SWS) and REMsleep, with less time in light N2(NREM stages 1-2). This would haveincreased the relative proportion ofdeep sleep versus light sleep, whileshortening the total time individualsneeded to be inactive. Moreover,early humans may have been thefirst primates to exhibit a single inte-grated sleep period with a greaterproportion of deep sleep character-ized by a major initial period of N3(SWS) and coupled bouts of propor-tionately longer REM sleep.

In addition, with the use of fire,cooking would have reduced the timeinvested in chewing from 4-7 hours aday to about 1 hour.94 Indeed, onestudy has found that, relative to otherprimates, humans are evolutionaryoutliers in their chewing time,54

spending only 4.7% of the active dayfeeding, whereas Pan spends 37.5% ofthe day doing so. Including the reduc-tion in time spent chewing from the

Pan average (270 minutes) to ahuman average (35 minutes50) and arelease from the obligate inactivity ofarboreal sleep sites (720 minutes) to ahuman average (420 minutes), the netgain of activity in 24 hours could haveamounted to approximately 8.9hours. Moreover, fire may haveincreased the “artificial day,” sinceearly human activity may have beenextended past dusk by access to con-trollable light. In sum, the innova-tions of fire and high-quality sleepsites on the ground could haveincreased early Homo waking activityby 37% in a 24-hour period.

Increased sleep intensity conferredat least three cognitive benefits on early

humans. The first of these involvesthreat priming. By way of the phenom-enological content of sleep (dreaming),REM primes sleepers by rehearsinglikely threatening events or social sce-narios that may occur in their wakingenvironments.99 Increased innovationis a second benefit of increased sleepintensity. In particular, REM sleep andits contents may have allowed for awider networking of ideation, resultingin greater frequency of creativity,insight, and innovation.100 Moreover,increased sleep intensity likely en-hanced memory consolidation. Clearevidence exists regarding the role ofSWS and REM sleep in processingdaily information into long-term mem-ory stores. For example, SWS has been

associated with the consolidation ofprocedural memories (for example,visuospatial locations and stone-knapping skills)101 and the processingof emotionally valent declarative andepisodic memories.17 It is worth not-ing that Walker and Stickgold102 pro-posed a homeostatically drivendemand on sleep-dependent memoryconsolidation that reciprocally enhan-ces sleep depth; in other words, sleepenhances learning and, in turn, learn-ing enhances sleep.

With the increase in potentialactivity budget, significant group-level social activity could have beenexpanded to the night time. Thiscould have had important conse-quences for hominin socioecology byincreasing the total time available tobond communities,103 transmit cul-tural information, and augment wak-ing cognitive abilities. In turn, with areduction in total sleep time andincrease in sleep intensity, there mayhave been selective pressure for thecognitive and behavioral benefits ofimproved memory consolidation,increased creativity, and social intel-ligence, all of which would plausiblyimprove survival in challenging novelterrestrial environments.

KEY PREDICTIONS AND FUTUREDIRECTIONS

In exploring primate sleep ecologyand physiology, we presented a novelhypothesis that attempts to explainhow modern human sleep architec-ture evolved. While some existingdata led us to this hypothesis, futureresearch should test specific predic-tions arising from it. First, generat-ing descriptive statistics on sleeparchitecture in previously unre-corded captive and wild primates isessential. Our ability to test thesehypotheses is limited because of thedearth of nonhuman primate sleepstudies. As noted earlier, such stud-ies are difficult because of the inva-sive nature of EEGs and EMGs.However, cost-effective technology totest such hypotheses is now becom-ing available (Table 2). Key speciesto generate further data to test thesleep intensity hypothesis are lemurs(for example, Lemur catta, Propithe-cus spp., and Eulemur spp.), the

In full-time terrestrialenvironments, hominins,sleeping in large, senti-neled groups on stableground beds protectedby fire, would havebeen uniquely posi-tioned to capitalize onthe adaptive advantageof deeper, more intense,REM-dominated sleep.

234 Samson and Nunn ARTICLE

New World Cebidae, the Old Worldcolobines, and the apes.11 Moreover,sleep expression encompasses morethan just NREM and REM distribu-tion; to effectively move forward,research should also quantify behav-ioral measures such as sleep effi-ciency, motor activity, and sleepfragmentation, which can be meas-ured in primates via actigraphy andinfrared videography.

As data accumulate for more spe-cies, phylogenetic comparative meth-ods can be applied to reanalyze keyhypotheses and infer ancestral statesto assess characteristics of sleep inextant taxa, including humans. Sev-eral questions could be targetedusing comparative data on NREM,REM, sleep depth, and their corre-lates: Across mammals, how do evo-lutionary transitions to various kindsof sociality influence sleep? Do eco-logical variables associated withsleep ecology correlate with sleeparchitecture? Specifically, do animalswith fixed-point nests or arborealsleeping platforms have differentsleep characteristics than do thosethat sleep on branches? Do rates ofinnovation and social learningcovary with specific types of sleeparchitecture (more REM coupledwith less total sleep)? We predictthat the mechanisms that facilitatedifferences in sleep intensity anddepth will be functionally related tosleep environments, particularly dis-

crete parameters that augment secu-rity and comfort in sleep sites.

A critical test of the sleep intensityhypothesis will come from sleep datagenerated in traditional human pop-ulations characterizing normal sleeppatterns in varying climates and lati-tudes. The hypothesis would be dis-proved if traditional populationswere characterized by longer thanaverage sleep durations. A criticaltest of the sleep intensity hypothesiswas recently reported: Research thatgenerated sleep durations of 5.7-7.1hours for traditional populations13

discovered even shorter durationsthan the standard human averageused in our phylogenetic analysis.This finding supports the sleep inten-sity hypothesis. Furthermore, thesleep intensity hypothesis would bedisproved by comparative data show-ing that human sleep is less efficientthan that of our closest phylogeneticrelatives when sleeping is done insimilar conditions. However, even ifwe found that sleep across apes islargely similar, modern human reli-ance on high quality sleep is critical.Increased attention to sleep alongthe ape lineage may give a clearerunderstanding of why we sleep.

CONCLUSIONS

As Rechtschaffen104 noted, “Ifsleep does not serve an absolute vitalfunction, then it is the biggest mis-

take the evolutionary process evermade.” We advanced the hypothesisthat human sleep has unique charac-teristics and that those characteris-tics are intimately intertwined withother features that have madehumans so successful. We also pro-posed that the four known primatesleep environments are linked to theallometric sleep-scaling threshold,and that these discrete categorieslend support to H. erectus havingbeen the first full-time terrestrialsleeper among the primates.

These considerations led us toarticulate the sleep intensity hypothe-sis, which states that evolutionaryshifts in human sleep environmentshave allowed sleep quality to improveby permitting deeper sleep. This evo-lutionary model of early homininsleep builds on previous contribu-tions5 and dovetails with currenthypotheses related to the importanceof fire94 and group cohesion104 inhuman evolution. For early humans,a unique sleep architecture promot-ing information consolidation by wayof highly efficient, secure, and com-fortable sleep environments couldhave resulted in high degrees of cog-nitive and behavioral plasticity.

Empirical sleep research inanthropology is in its infancy. Thatbeing said, the window of possibilityto test many of the relevant hypothe-ses that could answer the greaterquestions of sleep’s role in humanevolution may be coming to a close.Wild populations of great apes areamong the most endangered in theworld, and hunter-gatherer culturesremain extremely vulnerable to mas-sive shifts in subsistence strategyand cultural norms, given increased(often hostile) exposure to largernation states. Therefore, generatingsleep data in an ecologically relevantcontext is acutely urgent.

ACKNOWLEDGMENTS

We thank Peter Gray, Kevin Hunt,Randi Griffin, Daniel Samson, andJames McKenna for constructive sug-gestions and provocative discussion.We thank Alexander Vining and Isa-bella Capellini for help in preparingthe manuscript; and Manfred Eberle,Katheligne Koops, Matthieu Paley for

TABLE 2. Recording Primate Sleep: Current Cost-Effective Technology That CanExpand Primate Sleep Datasets

Technology Gross Measure Specifics Invasive

Actigraphy Active/nonactivestates

Can measure activity at 1-secinterval resolution; in humans,algorithms may generateNRME/REM states at 90%efficacy

No

Videography/Eularianmagnification

Behavioralsignatures

Can measure respiration andgross body motor movement;in apes, shown to differentiatebetween NREM/REM statesat 80% efficacy

No

Piezoelectric Active/nonactivestates

Automatically records data;can measure breathingrhythms

No

EEG Gold standard“brain waveactivity”

Automatically records data; candifferentiate between NREM/REMstates and measure delta andtheta wave states

Yes

ARTICLE Sleep Intensity and the Evolution of Human Cognition 235

providing images of sleeping prima-tes; and Sarah Van Tassel for editingthe images. We are very grateful toAlyssa Crittenden for several hours ofdiscussion on Hadza sleep practice.Finally, we extend our gratitude forthe insightful, critical comments pro-vided by two anonymous reviewers, aswell as Frederick Coolidge, KristenHawkes, and John Fleagle. Thisresearch was supported by the NSF(BCS-1355902) and Duke University.

REFERENCES

1 Zepelin H, Siegel J, Tobler I. 2005. Mammaliansleep. In: Kryger MH, Roth R, Dement WC, edi-tors. Principles and practice of sleep medicine.Philadelphia, PA: Elsevier Suanders. p 91–100.

2 Lesku JA, Roth TC, Lima SL, et al. 2006. Sleeparchitecture in birds: an avian perspective on Zepe-lin and Rechtschaffen (1974). Sleep 29:A35–A35.

3 Worthman CM, Melby MK. 2002. Toward acomparative developmental ecology of humansleep. In: Carskadon MA, editor. Adolescentsleep patterns: biological, social, and psycho-logical influences. Cambridge, MA: CambridgeUniversity Press. p 69–117.

4 Worthman CM. 2008. After dark: the evolu-tionary ecology of human sleep. In: TrevathanWR, Smith EO, McKenna JJ, editors. Evolu-tionary medicine and health. Oxford: OxfordUniversity Press. p 291–313.

5 Coolidge FL, Wynn T. 2006. The effects ofthe tree-to-ground sleep transition in the evolu-tion of cognition in early Homo. Before Farm-ing 4:1–18.

6 Siegel JM. 2001. The REM sleep: memory con-solidation hypothesis. Science 294:1058–1063.

7 Durmer JS, Dinges DF. 2005. Neurocognitiveconsequences of sleep deprivation. Semin Neu-rol 25:117–129.

8 Drummond SPA, Gillin JC, Brown GG. 2001.Increased cerebral response during a dividedattention task following sleep deprivation.J Sleep Res 10:85–92.

9 McNamara P, Auerbach S. 2010. Evolutionarymedicine of sleep disorders: toward a science ofsleep duration. In: McNamara P, Barton RA,Nunn CL, editors. Evolution of sleep: phyloge-netic and functional perspectives. Cambridge:Cambridge University Press. p 107–122.

10 Coolidge FL, Wynn T. 2009. The rise ofHomo sapiens: the evolution of modern think-ing. West Sussex, UK: Wiley-Blackwell.

11 Nunn CL, McNamara P, Capellini I, et al.2010. Primate sleep in phylogenetic perspective.In: McNamara P, Barton RA, Nunn CL, editors.Evolution and sleep: phylogenetic and func-tional perspectives. New York: Cambridge Uni-versity Press. p 123–145.

12 Anderson JR. 1998. Sleep, sleeping sites,and sleep-related activities: awakening to theirsignificance. Am J Primatol 46:63–75.

13 Yetish G, Kaplan H, Gurven M, et al. 2015.Natural sleep and its seasonal variations inthree pre-industrial societies. Curr Biol 25:1–7.

14 Vyazovskiy VV, Delogu A. 2014. NREM andREM sleep: complementary roles in recoveryafter wakefulness. Neuroscientist 20:203–219.

15 Webb WB. 1988. Theoretical presentation:an objective behavioral model of sleep. Sleep11:488–496.

16 Kavanau JL. 1997. Memory, sleep and theevolution of mechanisms of synaptic efficacymaintenance. Neuroscience. 79:7–44.

17 Walker MP. 2009. The role of sleep in cognitionand emotion. Ann NY Acad Sci 1156:168–197.

18 Xie L, Kang H, Xu Q, et al. 2013. Sleepdrives metabolite clearance from the adultbrain. Science 342:373–377.

19 Achermann P, Dijk DJ, Brunner DP, et al.1993. A model of human sleep homeostasisbased on EEG slow-wave activity: quantitativecomparison of data and simulations. Brain ResBull 31:97–113.

20 Fisher SP, Foster RG, Peirson SN. 2013. Thecircadian control of sleep. Exp Pharmocol 217:157–83.

21 Saper CB, Fuller PM, Pedersen NP, et al.2010. Sleep state switching. Neuron 68:1023–1042.

22 Ackermann S, Rasch B. 2014. Differentialeffects of non-REM and REM sleep on memoryconsolidation? Curr Neurol Neurosci 14:1–10.

23 Vyazovskiy VV, Harris KD. 2013. Sleep andthe single neuron: the role of global slow oscil-lations in individual cell rest. Nat Rev Neurosci14:445–453.

24 Huber R, Deboer T, Tobler I. 2000. Effectsof sleep deprivation on sleep and sleep EEG inthree mouse strains: empirical data and simula-tions. Behav Brain Res 857:8–19.

25 Nishida M, Pearsall J, Buckner RL, et al.2009. REM sleep, prefrontal theta, and the con-solidation of human emotional memory. CerebCortex 19:1158–1166.

26 Sallinen M, Kaartinen J, Lyytinen H. 1996.Processing of auditory stimuli during tonic andphasic periods of REM sleep as revealed byevent-related brain potentials. J Sleep Res 5:220–228.

27 Ermis U, Krakow K, Voss U. 2010. Arousalthresholds during human tonic and phasicREM sleep. J Sleep Res 19:400–406.

28 Morin CM. 2011. Psychological and behav-ioral treatments for insomnia I: approachesand efficacy. In: Kryger MH, Roth T, DementWC, editors. Principles and practice of sleepmedicine, 5th ed. St. Louis, MO: Elsevier Saun-ders. p 866–883.

29 Thorpy MJ. 2011. Classification of sleep dis-orders. In: Kryger MH, Roth T, Dement WC,editors. Principles and practice of sleep medi-cine. 5th ed. St. Louis, MO: Elsevier Saunders.p 680–693.

30 Ekirch AR. 2006. At day’s close: night intimes past. New York, NY: WW Norton.

31 Everett DL. 2009. Don’t sleep, there aresnakes: life and language in the Amazonianjungle. New York: Vintage.

32 Knutson KL. 2014. Sleep duration, quality,and timing and their associations with age in acommunity without electricity in Haiti. Am JHum Biol 26:80–86.

33 McKenna JJ, McDade T. 2005. Why babiesshould never sleep alone: a review of the co-sleeping controversy in relation to SIDS, bed-sharing and breast feeding. Paediatr Respir Rev6:134–52.

34 Epstein R, Herer P, Tzischinsky O, et al.1997. Changing from communal to familialsleep arrangement in the Kibbutz: effects onsleep quality. Sleep 20:334–339.

35 Libert JP, Bach V, Johnson LC, et al. 1991.Relative and combined effects of heat and noiseexposure on sleep in humans. Sleep 14:24–31.

36 Reynolds B. 1968. The material cultures ofthe peoples of the Gwembie valley. New York:Praeger.

37 Turnbull CM. 1962. The forest people. NewYork: Simon and Schuster.

38 Wobber V, Hare B, Wrangham R. 2008.Great apes prefer cooked food. J Hum Evol 55:340–348.

39 Wrangham RW, Jones JH, Laden G, et al.1999. The raw and the stolen: cooking and theecology of human origins. Curr Anthropol 40:567–594.

40 Hammel HT, Elsner RW, Lemessurier DH,et al. 1959. Thermal and metabolic responsesof the Australian aborigine exposed to moder-ate cold in summer. J Appl Phys 14:605–615.

41 Gray PB. 2013. Evolution and human sex-uality. Am J Phys Anthropol 152:94–118.

42 Hewlett BS. 2014. Hunter-gatherers of theCongo Basin: cultures, histories, and biology ofAfrican pygmies. New Brunswick, NJ: Transac-tion Publishers.

43 Randler C. 2014. Sleep, sleep timing andchronotype in animal behaviour. Anim Behav94:161–166.

44 Viola AU, Archer SN, James LM, et al. 2007.PER3 polymorphism predicts sleep structureand waking performance. Curr Biol 17:613–618.

45 Ohayon MM, Carskadon MA, GuilleminaultC, et al. 2004. Meta-analysis of quantitativesleep parameters from childhood to old age inhealthy individuals: developing normative sleepvalues across the human lifespan. Sleep 27:1255–1273.

46 Owens JA. 2004. Sleep in children: cross-cultural perspectives. Sleep Biol Rhythms 2:165–173.

47 Ekirch AR. 2001. Sleep we have lost: pre-industrial slumber in the British isles. Am HistRev 106:343–386.

48 Wehr TA. 1992. In short photoperiods,human sleep is biphasic. J Sleep Res 1:103–107.

49 Zamma K. 2014. What makes wild chimpan-zees wake up at night? Primates 55:51–57.

50 Krief S, Cibot M, Bortolamiol S, et al. 2014.Wild chimpanzees on the edge: nocturnal activ-ities in croplands. Plos One 9:1–11.

51 Videan EN. 2006. Sleep in captive (Pan trog-lodytes): the effects of individual and environ-mental factors on sleep duration and quality.Behav Brain Res. 169:187–192.

52 Nunn CL. 2011. The comparative approachin evolutionary anthropology and biology. Chi-cago: University of Chicago Press.

53 Nunn C, Zhu L. 2014. Phylogenetic predictionto identify “evolutionary singularities.” In:Garamszegi LZ, editor. Modern phylogeneticcomparative methods and their application inevolutionary biology. Berlin: Springer. p 481–514.

54 Organ CL, Nunn CL, Machanda Z, et al.2011. Phylogenetic rate shifts in feeding timeduring the evolution of Homo. Proc Natl AcadSci USA 108:14555–14559.

55 Sri Kantha S, Suzuki J. 2006. Sleep quanti-tation in common marmoset, cotton top tam-arin and squirrel monkey by non-invasiveactigraphy. Comp Biochem Phys 144:203–210.

56 Tobler I. 2005. Phylogeny of sleep regula-tion. In: Kryger MH, Roth R, Dement WC, edi-tors. Principles and practice of sleep medicine.Elsevier Saunders. Philadelphia, PA. p 77–90.

57 Koops K, Humle T, Sterck EHM, et al.2007. Ground-nesting by the chimpanzees ofthe Nimba Mountains, Guinea: environmentallyor socially determined? Am J Primatol 69:407–419.

58 Yamagiwa J. 2001. Factors influencing theformation of ground nests by eastern lowlandgorillas in Kahuzi-Biega National Park: some

236 Samson and Nunn ARTICLE

evolutionary implications of nesting behavior.J Hum Evol 40:99–109.

59 Fleagle JG. 1999. Primate adaptation andevolution. 2nd ed. San Diego: Academic Press.

60 Samson DR. 2012. The chimpanzee nestquantified: morphology and ecology of arborealsleeping platforms within the dry habitat site ofToro-Semliki Wildlife Reserve, Uganda. Prima-tes 53:357–364.

61 Covert HH. 1995. Locomotor adaptations ofEocene primates: adaptive diversity among theearliest prosimians. In: Alterman L, Doyle GA,Izard MK, editors. Creatures of the dark; thenocturnal prosimians. New York: PlenumPress. p 495–509.

62 Mooring MC, Hart BL. 1992. Animal groupingfor protection from parasites: selfish herd andencounter-dilution effects. Behaviour 123:173–193.

63 Kappeler PM. 1998. Nests, tree holes, andthe evolution of primate life histories. Am J Pri-matol 46:7–33.

64 Nunn CL, Heymann EW. 2005. Malariainfection and host behaviour: a comparativestudy of Neotropical primates. Behav EcolSociobiol 59:30–37.

65 McGrew WC. 2004. The cultured chimpan-zee: reflections on cultural primatology. NewYork: Cambridge University Press.

66 Fleagle JG, Kay R. 1985. The paleobiology ofcatarrhines. In: Delson E, editor. Ancestors: thehard evidence. New York: Alan R. Liss. p 23–36.

67 Baker J, Meade A, Pagel M. 2015. Adaptiveevolution towards larger size in mammals. ProcNatl Acad Sci USA 112:5093–5098.

68 vanSchaik CP, Kappeler PM. 1996. Thesocial systems of gregarious lemurs: lack ofconvergence with anthropoids due to evolution-ary disequilibrium? Ethology 102:915–941.

69 Samson DR, Shumaker WR. 2015. Orangu-tans (Pongo spp.) have deeper, more efficientsleep than baboons (Papio papio) in captivity.Am J Phys Anthropol 157:421–427.

70 Goodall JM. 1962. Nest building behavior inthe free ranging chimpanzee. Ann NY Acad Sci102:455–467.

71 Fruth B, Hohmann G. 1996. Nest buildingbehavior in the great apes: the great leap for-ward? In: Marchant LF, Nishida T, editors.Great ape societies. Cambridge: CambridgeUniversity Press. p 225–240.

72 Samson DR, Hunt KD. 2014. Chimpanzeespreferentially select sleeping platform construc-tion tree species with biomechanical propertiesthat yield stable, firm, but compliant nests.PloS One 9:1–8.

73 van Casteren A, Sellers WI, Thorpe SKS,et al. 2012. Nest-building orangutans demon-strate engineering know-how to produce safe,comfortable beds. Proc Natl Acad Sci USA 109:6873–6877.

74 van Casteren A, Sellers WI, Thorpe SKS,et al. 2013. Factors affecting the complianceand sway properties of tree branches used bythe sumatran orangutan (Pongo abelii). PlosOne 8:1–9.

75 Reichard U. 1998. Sleeping sites, sleepingplaces, and presleep behavior of gibbons (Hylo-bates lar). Am J Primatol 46:35–62.

76 Anderson JR. 1984. Ethology and ecology ofsleep in monkeys and apes. Adv Stud Behav 14:165–229.

77 Pruetz JD, Fulton SJ, Marchant LF, et al.2008. Arboreal nesting as anti-predator adapta-tion by savanna chimpanzees (Pan troglodytesverus) in southeastern Senegal. Am J Primatol70:393–401.

78 Stewart FA, Pruetz JD. 2013. Do chimpan-zee nests serve an anti-predatory function? AmJ Primatol 75:593–604.

79 Herman N. 2010. Nest site selection in Bor-nean orangutans (Pongo pygmaeus wurmbii):the role of microclamatic factors and moquitoavoidance strategies. PhD dissertation, Univer-sity of Zurich.

80 Samson DR, Muehlenbein MP, Hunt KD.2013. Do chimpanzees (Pan troglodytes schwein-furthii) exhibit sleep related behaviors that min-imize exposure to parasitic arthropods? Apreliminary report on the possible anti-vectorfunction of chimpanzee sleeping platforms. Pri-mates 54:73–80.

81 Stewart FA. 2011. Brief communication:why sleep in a nest? Empirical testing of thefunction of simple shelters made by wild chim-panzees. Am J Phys Anthropol 146:313–318.

82 Duda P, Zrzavy J. 2013. Evolution of lifehistory and behavior in Hominidae: towardsphylogenetic reconstruction of the chimpanzee-human last common ancestor. J Hum Evol 64:424–446.

83 Baldwin PJ, Sabater Pi J, McGrew WC,et al. 1981. Comparisons of nests made by dif-ferent populations of chimpanzees (Pan troglo-dytes). Primates 22:474–486.

84 Samson DR, Shumaker RS. 2013. Docu-menting orang-utan sleep architecture: sleepingplatform complexity increases sleep quality incaptive Pongo. Behaviour 150:845–861.

85 Martin-Ordas G, Call J. 2011. Memory proc-essing in great apes: the effect of time andsleep. Biol Lett 7:829–832.

86 Hunt KD. 1996. The postural feedinghypothesis: an ecological model for the evolu-tion of bipedalism. S Afr J Sci 92:77–90.

87 Wood B, Collard M. 1999. Anthropology: thehuman genus. Science 284:65–71.

88 McHenry HM, Coffing K. 2000. Australopi-thecus to Homo: transformations in body andmind. Ann Rev Anthropol 29:125–146.

89 McHenry HM. 1991. Body size and propor-tions in early hominids. Am J Phys Anthropol84:407–431.

90 Anton SC, Snodgrass JJ. 2012. Origins andevolution of genus Homo: new perspectives.Curr Anthropol 53:S479–S496.

91 Werdelin L, Lewis ME. 2005. Plio-Pleisto-cene carnivora of eastern Africa: species rich-ness and turnover patterns. Zool J Linn SocLond 144:121–144.

92 Bodker R. 2003. Relationship between alti-tude and intensity of malaria transmission inthe Usambara Mountains, Tanzania. J MedEntomol 40:706–717.

93 Gowlett JAJ, Wrangham RW. 2013. Earliestfire in Africa: towards the convergence of

archaeological evidence and the cookinghypothesis. Arch Res Afr 48:5–30.

94 Wrangham R, Carmody R. 2010. Humanadaptation to the control of fire. Evol Anthro-pol 19:187–199.

95 Rossano M. 2006. The religious mind andthe evolution of religion. Rev Gen Psychol 10:346–364.

96 Kurmi OP, Lam KBH, Ayres JG. 2012.Indoor air pollution and the lung in low- andmedium-income countries. Eur Respir J 40:239–254.

97 Moore SJ, Hill N, Ruiz C, et al. 2007. Fieldevaluation of traditionally used plant-basedinsect repellents and fumigants against themalaria vector Anopheles darlingi in Riber-alta, Bolivian Amazon. J Med Entymol 44:624–630.

98 Hal!asz PB, B!odizs R, Parrino L, et al. 2014.Two features of sleep slow waves: homeostaticand reactive aspects from long term to instantsleep homeostasis. Sleep Med 15:1184–1195.

99 Revonsuo A. 2000. The reinterpretation ofdreams: an evolutionary hypothesis of the func-tion of dreaming. Beh Bra Sci 23:793–1121.

100 Wagner U, Gais S, Haider H, et al. 2004.Sleep inspires insight. Nature 427:352–355.

101 Peigneux P, Laureys S, Fuchs S, et al.2004. Are spatial memories strengthened in thehuman hippocampus during slow wave sleep?Neuron 44:535–545.

102 Walker MP, Stickgold R. 2004. Sleep-dependent learning and memory consolidation.Neuron 44:121–133.

103 Dunbar RIM, Gowlett JAJ. 2014. Firesidechat: the impact of fire on hominin socioecol-ogy. In: Dunbar RIM, Gamble C, Gowlett JAJ,editors. Lucy to language: the benchmarkpapers. Oxford: Oxford University Press. p 277–293.

104 Rechtschaffen A. 1971. The control of sleep.In: Hunt WA, editor. Human behaviour and itscontrol. Cambridge, MA: Schenkman. p 75–92.

105 Sherwood CC, Bauernfeind AL, Bianchi S,et al. 2012. Human brain evolution writ largeand small. In: Hofman M, Falk D, editors. Evo-lution of the primate brain: from neuron tobehavior. New York: Elsevier. p 237–254.

106 Gardland T, Ives AR. 2000. Using the pastto predict the present: confidence intervals forregression equations in phylogenetic compara-tive methods. Am Nat 155:346–364.

107 Capellini I, Barton RA, McNamara P, et al2008. Phylogenetic analysis of the ecology andevolution of mammalian sleep. Evol 6:1764–1776.

108 McNamara P, Capellini I, Harris E, et al.2008. The phylogeny of sleep database: a resourcefor sleep scientists. Open Sleep J 1:11–14.

109 Marlowe F. 2010. The Hadza: hunter-gatherers of Tanzania. Berkeley: University ofCalifornia Press.

110 Smith RJ, Jungers WL. 1997. Body mass incomparative primatology. J Hum Evol 32:523–559.

VC 2015 Wiley Periodicals, Inc.

ARTICLE Sleep Intensity and the Evolution of Human Cognition 237