Enel Health®Atlas
Sign in
Enel Health®Atlas
Browse libraries

Not medical advice. The Atlas is not a substitute for a clinician who can examine you.

Sleep Architecture

Sleep Stages & The 90-Minute Cycle

Evidence grade 5 of 5, Settled. Read how the Atlas grades evidence.

0 human trials, about 2000 studies, Foundational. The written rationale for this grade is still being prepared.

FDA
Foundational sleep science
WADA
Not prohibited

Reference only. Not medical advice.

What can go wrong

Sleep architecture is descriptive physiology and carries no intrinsic risk, but several practical cautions apply. Many substances shift stage distribution: acute alcohol suppresses REM and fragments the second half of the night, and acute THC in naive users tends to increase slow-wave sleep and reduce REM, with these effects subject to tolerance and a REM rebound on withdrawal. Consumer trackers estimate stages imperfectly and a poor nightly score should not be treated as a diagnosis. Persistent unrefreshing sleep, loud snoring or witnessed breathing pauses, acting out dreams, or daytime sleepiness warrant evaluation by a qualified clinician or sleep-medicine provider rather than self-management from wearable data. Anyone considering changes to medication, alcohol, or cannabis use for sleep should consult a qualified healthcare provider.

Sleep is not a uniform state but cycles through distinct stages roughly every 90 minutes: lighter non-REM (N1, N2), deep slow-wave sleep (N3), and REM. Their proportions shift across the night, deep N3 dominates the first half and is associated with the night's largest growth-hormone pulse, while REM lengthens and concentrates in the second half. This front-loading means cutting sleep short with an early alarm disproportionately removes REM, the stage associated with emotional and memory processing, which is one reason modest sleep loss can affect mood and learning beyond simple tiredness. Sleep timing and continuity therefore matter alongside total hours. Consumer wearables estimate these stages only roughly, so a nightly stage score is best read as a trend signal rather than a clinical measurement.

Deep sleep in the first half, REM in the second. Cutting sleep short cuts REM most.

What it is invoked to explain

Claims this idea is used to support. Whether it supports them is the rest of the entry.

  • Understanding sleep quality and continuity, not just total quantity
  • Explaining why both sleep timing and duration matter
  • Providing context for how alcohol and some medications shift sleep stages
  • Interpreting consumer sleep-tracker stage data honestly as a trend rather than a measurement
  • Framing why an early alarm preferentially removes REM

What people get wrong

  • Often saidAs long as I get enough total hours, when I sleep doesn't matter

    ActuallyStage distribution shifts across the night, so timing changes what kind of sleep you get. Slow-wave deep sleep is front-loaded and the largest growth-hormone pulse occurs minutes after the first slow-wave period, while REM lengthens toward morning. Truncating the end of sleep with an early alarm removes REM disproportionately (AASM; Van Cauter, growth-hormone studies).

  • Often saidMy wearable accurately measures my sleep stages

    ActuallyIn a multicenter validation of 11 consumer trackers against polysomnography, agreement varied widely (Cohen's kappa roughly 0.21 to 0.53, fair to moderate), with REM typically estimated better than deep sleep and one device labeling 59% of epochs deep when only about 11% truly were. Useful for tracking trends over time, not for precise per-night staging (JMIR mHealth uHealth 2023).

  • Often saidDeep sleep and REM are basically the same restorative state

    ActuallyThey are distinct. Slow-wave N3 is associated with physical restoration and the nocturnal growth-hormone pulse and declarative-memory consolidation; REM is associated with emotional and procedural memory processing and is when most vivid dreaming occurs (AASM staging; sleep-and-memory reviews).

  • Often saidA normal night has a fixed, even split of each stage

    ActuallyProportions are not even and shift cycle by cycle: early cycles are rich in N3 with little REM, later cycles are rich in REM with little N3. This ordering, not a flat average, is what makes truncated sleep cost REM specifically (Rechtschaffen and Kales; AASM).

Who profits, and who pushes back

  • Dr. Matthew Walker

    Promotes itUC Berkeley

    Popularized sleep science widely; core architecture content is sound (some specific mortality claims in his book have been critiqued). Limited commercial ties.

  • Sleep-medicine community

    Promotes itClinical

    Sleep staging is foundational, well-established polysomnography science.

  • Consumer Sleep Tracker validation researchers (Chee, Lee et al., JMIR mHealth uHealth)

    Pushes backMulticenter device-validation study, 349,114 epochs vs polysomnography

    Across 11 consumer trackers, macro F1 ranged from 0.69 down to 0.26 and agreement with polysomnography was only fair to moderate, with deep sleep estimated less reliably than REM; the devices are suited to tracking trends, not to precise per-night staging.

  • Borbely et al. (two-process model reappraisal, 2016)

    Pushes backOriginators of the governing sleep-regulation framework

    Sleep is governed by the interaction of a homeostatic process S that builds with time awake and a circadian process C; the framework, now widely cited, frames why both how long you have been awake and what time it is shape sleep, which is exactly why timing and architecture cannot be reduced to total hours.

  • Cannabis-and-sleep meta-analysts (Systematic review, 2025)

    Pushes backSystematic review and meta-analysis of cannabis and sleep architecture

    In contrast to early high-dose studies, modern trials administering THC generally did not report REM suppression, plausibly reflecting pharmacological tolerance, so blanket claims that cannabis 'kills REM' overstate a dose- and tolerance-dependent effect.

Where the research comes from

Share of published research by country of origin.
PlaceShare of published research
USA40%
Europe34%
Other26%

Research published from: Global.

How we got here

Periodic eye movements in sleep were noted by Russian physiologists in the 1920s, but modern sleep architecture dates to 1953, when graduate student Eugene Aserinsky, working under Nathaniel Kleitman at the University of Chicago, recorded regularly recurring rapid-eye-movement periods tied to vivid dreaming and published the finding in Science. William Dement then helped link REM to dreaming and map its cyclic structure across the night. In 1968 Allan Rechtschaffen and Anthony Kales standardized stage scoring, the framework the American Academy of Sleep Medicine later revised into its current manual. In 1982 Alexander Borbely of Zurich proposed the two-process model, framing sleep as the interaction of a homeostatic sleep pressure (process S) and a circadian rhythm (process C). Together these advances overturned the old assumption that sleep was passive and uniform, and grounded sleep-disorder diagnosis and the modern understanding of why timing and continuity matter as much as duration.

  1. 1920sRussian physiologists observe periodic eye movements during sleep
  2. 1953Aserinsky and Kleitman record and report REM sleep in Science
  3. 1957-1960sDement and Kleitman map the REM/NREM cycle and its distribution across the night
  4. 1968Rechtschaffen and Kales publish the first standardized sleep-stage scoring manual
  5. 1982Borbely proposes the two-process model (homeostatic process S plus circadian process C)
  6. 2007AASM renames stages to N1-N3 and R, consolidating the old stages 3 and 4 into N3
  7. 2010s-2020sConsumer wearables popularize nightly stage estimates of varying accuracy
  8. 2023AASM Manual for the Scoring of Sleep and Associated Events Version 3 takes effect

At a glance

Cycle
~90 minutes
Deep sleep (N3)
First half of night
REM
Second half of night
Alcohol/THC
Suppress REM

What this entry does not yet cover

What follows is unfinished work on our side, not a finding about Sleep Architecture. A short entry is not a safe entry.

  • Cost and market. We have not documented what it sells for or who sells it.
  • Supply chain. We have not mapped where it is made.
  • Who uses it. We have not documented the user groups.
  • What could change. No forward-looking summary yet.
  • How it compares. We have not set it against the alternatives.
  • Children and teenagers. We have not checked what is known about younger people.

Sources