Your activity: 238 p.v.
your limit has been reached. plz Donate us to allow your ip full access, Email:

Stages and architecture of normal sleep

Stages and architecture of normal sleep
Douglas Kirsch, MD, FAASM
Section Editor:
Susan M Harding, MD, FCCP, AGAF
Deputy Editor:
April F Eichler, MD, MPH
Literature review current through: Dec 2022. | This topic last updated: Sep 12, 2022.

INTRODUCTION — Sleep is a rapidly reversible state of reduced responsiveness, motor activity, and metabolism [1]. It is a phenomenon observed in all animals in some form; this universality suggests that the act of sleeping likely has some evolutionary relevance. Humans spend approximately one-third of their life, or about eight hours per night, sleeping. The purpose of sleeping is poorly understood, however, and multiple theories exist. These theories include restoration, energy conservation, and memory consolidation.

The polysomnogram is the primary tool for assessing sleep in the laboratory for both clinical and research purposes. During a polysomnogram, electroencephalography (EEG) and other sensors are used to categorize sleep in discrete stages. Initial sleep staging was described in the 1930s, and formal rules for staging sleep were first propagated in 1968 [2]. Since 2007, most sleep laboratories use terminology and scoring rules from the American Academy of Sleep Medicine (AASM) manual for the Scoring of Sleep and Associated Events, which is routinely updated [3]. Sleep laboratories accredited by the AASM are required to use the AASM scoring manual, and these guidelines are being increasingly adopted worldwide [4].

This topic will review the current guidelines for adult sleep staging, the architecture of sleep, common causes of sleep stage abnormalities, and theories around the purposes for sleep. The classification of sleep disorders is reviewed separately. (See "Classification of sleep disorders".)

SLEEP STAGING — Sleep can be broadly segmented into rapid eye movement (REM) sleep and non-REM (NREM) sleep. Scoring of sleep stages occurs in 30-second epochs based on current American Academy of Sleep Medicine (AASM) scoring rules [3]. The current rules mandate the use of electroencephalography (EEG), electromyography (EMG) for muscle tone, and electro-oculography (EOG) for eye movements, to determine the stage of sleep. This may change in the future to better align with sleep as a continuous process, by focusing attention on shorter epochs of sleep or through spectral analysis of sleep EEG.

A typical 30-second sleep study data epoch is provided in the figure (figure 1). The EEG data is derived from electrodes placed on the head in frontal, central, and occipital locations and referenced to bone according to the international 10-20 system (figure 2). As part of this system, odd numbers refer to the left side of the head and even numbers are right sided; typically, EEGs require bilateral monitoring, as left and right cerebral hemispheres may not provide identical data. While full EEG monitoring requires placement of all 10-20 electrodes, sleep staging requires only the partial grouping mentioned above (F3 and F4, C3 and C4, O1 and O2). Brainwaves are assessed by amplitude and frequency; different frequencies are associated with different stages of sleep (table 1).

Wake — Adults are typically awake for at least two-thirds of the 24-hour day. Behavioral cues, including open eyes, movement, and conversation, demonstrate alertness. However, as activities wind down, people recline and close their eyes. It is at this time that the brainwaves slow to a stable posterior alpha rhythm (figure 1). This rhythm is the bridge between wake and sleep. When this rhythm slows further, sleep has arrived. Some adults do not generate the alpha posterior rhythm, making staging more difficult.

NREM sleep — Most adults will enter sleep from the drowsy state via NREM sleep. NREM sleep is divided into three sub-stages: stage N1, stage N2, and stage N3. Of note, older rules had four stages of NREM sleep [2]; in the current rules, NREM stage 3 and NREM stage 4 are combined as stage N3 [3].

Stage N1 — Stage N1 sleep is the typical transition from wakefulness to sleep. It is characterized by low amplitude mixed EEG frequencies in the theta range (4 to 7 Hz) for at least 50 percent of the epoch (figure 3). Eye movements are typically slow and rolling. Stage N1 is the lightest stage of sleep; patients awakened from it usually do not perceive that they were actually asleep. Stage N1 sleep typically accounts for 5 to 10 percent or less of the total sleep time in young adults [5].

An increased proportion of stage N1 sleep during a polysomnogram may suggest a sleep fragmenting disorder, such as obstructive sleep apnea. However, it may also represent a "first night effect" of a patient psychologically adjusting to being monitored in a sleep laboratory [6].

Stage N2 — Stage N2 generally comprises the largest percentage of total sleep time in a normal middle aged adult, typically 45 to 55 percent of the night [5]. It is characterized by theta EEG frequency.

There are two distinct features of NREM sleep that appear for the first time on the EEG during stage N2: sleep spindles and K-complexes.

Sleep spindles are generally short (though at least 0.5 seconds) and have an EEG frequency of 11 to 16 Hz (most commonly 12 to 14 Hz) (figure 4). They occur most prominently in the central (vertex) EEG leads.

K-complexes are well-delineated, negative, sharp waves immediately followed by a positive component that stand out from the background EEG and have a total duration of ≥0.5 seconds (figure 5). A K-complex usually has maximal amplitude in the frontal regions of the EEG.

Benzodiazepines increase stage N2 sleep. Typically, benzodiazepines are associated with increased spindle activity [7]. (See "The effects of medications on sleep quality and sleep architecture", section on 'Benzodiazepines and nonbenzodiazepine receptor agonists'.)

Stage N3 — Stage N3 sleep is frequently referred to as "deep sleep" or "slow wave sleep." It is characterized by low frequency (0.5 to 2 Hz), high amplitude delta EEG waves with an amplitude >75 microvolts, comprising at least 20 percent of a given sleep epoch (figure 6). Stage N3 sleep typically accounts for 10 to 20 percent of the total sleep time in young to middle age adults and decreases with age [5].

Stage N3 tends to occur more in the first half of the night and particularly at the beginning of the night, since slow wave activity during sleep represents the homeostatic drive to sleep, which is maximal after the waking period. It is often more difficult to arouse sleepers during stage N3 sleep compared with stages N1 and N2, and stage N3 is a typical time for NREM-parasomnias to occur. (See "Parasomnias of childhood, including sleepwalking".)

REM sleep (stage R) — REM sleep, also called stage R, is characterized by three primary features that require EEG, electro-oculography (EOG) and electromyography (EMG) to capture [3].

EEG demonstrates a low voltage, mixed EEG pattern. Sawtooth waves are a common finding during REM sleep; these 2-6 Hz wave patterns are sharply contoured and occur in brief bursts.

Rapid eye movements are the defining feature of the stage. These are defined on EOG by conjugate, irregular, sharply-peaked eye movements with an initial phase less than 500 milliseconds.

EMG demonstrates atonia, indicating inactivity of all voluntary muscles (except the extraocular muscles and the diaphragm). Atonia is the result of direct inhibition of alpha motor neurons.

There are two phases of REM sleep, phasic and tonic. Phasic REM sleep contains bursts of rapid eye movements, respiratory variability, and brief EMG activity (occasionally seen as muscle twitches) (figure 7). More limited motor activity occurs during tonic REM sleep, with few eye movements (figure 8).

REM sleep has typically been associated with vivid dreaming, based on early studies in which patients were awoken out of REM sleep [8]. Although REM sleep accounts for less than a quarter of total sleep time (ranging from 18 to 23 percent), the function of this stage of sleep is still an area of debate [5]. One hypothesis suggests that REM sleep is a time of memory consolidation during which important memories are retained and less important neural connections are pruned [9].

REM sleep is sub-segmented into phasic REM sleep and tonic REM sleep. Phasic REM sleep is the portion of REM sleep during which there are bursts of rapid eye movements, which may be associated with brief bursts of EMG activity (sometimes called twitching) and/or sudden increases in sympathetic activity. Tonic REM sleep is the portion of REM sleep that exists between the phasic bursts, in which low muscle tone is consistent.

Several sleep disorders may be related to abnormalities in REM sleep or the physiological events associated with REM sleep:

Narcolepsy – Patients with narcolepsy often have REM sleep very early in their sleep architecture and even during brief daytime naps [10]. (See "Clinical features and diagnosis of narcolepsy in adults".)

Obstructive sleep apnea – REM-related muscle atonia may impair upper airway patency, significantly exacerbating the frequency of obstructive respiratory events. Poor response of the brain to oxygen signals during REM sleep may also cause prolongation in respiratory events.

Pulmonary disease – REM-related atonia extends to the intercostal and accessory muscles of breathing. This has potentially important implications for patients whose abnormal respiratory mechanics render them dependent upon their intercostal and accessory muscles to maintain an acceptable level of ventilation (eg, patients with chronic obstructive pulmonary disease, a chest wall disorder, or a neuromuscular disorder). During REM sleep, such patients may develop hypoventilation and oxyhemoglobin desaturation [11]. (See "Sleep-related breathing disorders in COPD" and "Evaluation of sleep-disordered breathing in patients with neuromuscular and chest wall disease".)

REM sleep behavior disorder – Individuals may act out their dreams if the muscle atonia of REM sleep is incomplete or absent. This can result in injuries to the patient or bed partner [12]. (See "Rapid eye movement sleep behavior disorder".)

REM sleep may be delayed or suppressed by alcohol, sedative-hypnotic drugs, barbiturates and other antiepileptic drugs, beta antagonists, monoamine oxidase inhibitors, selective serotonin reuptake inhibitors, and stimulants. Medications with prominent anticholinergic effects (eg, tricyclic antidepressants) may also delay or suppress REM sleep. The clinical significance of these changes in REM sleep is unclear, however, as patients do not demonstrate typical findings of REM sleep deprivation [13]. Conversely, REM sleep can be increased by withdrawal from alcohol, benzodiazepines, tricyclic antidepressants, or monoamine oxidase inhibitors. (See "The effects of medications on sleep quality and sleep architecture".)

Arousals — Sleep stages shift in one of two ways. As sleep progressively deepens, specific EEG, EOG, and EMG findings will become visible, as described in the preceding sections. Alternatively, arousals will occur, bringing the individual from deeper to lighter sleep or to wakefulness (figure 9).

Arousals are defined by specific criteria. During sleep stages N1, N2, N3, or R, an arousal is scored if there is an abrupt shift of EEG frequency including alpha, theta and/or frequencies greater than 16 Hz (but not spindles) that lasts at least 3 seconds, with at least 10 seconds of stable sleep preceding the change [3]. Scoring of arousal during REM requires a concurrent increase in submental EMG lasting at least 1 second.

Consumer wearable devices — An increasing number of consumer wearable devices for fitness and other purposes purport to measure sleep, including sleep stages, based on proprietary algorithms making use of accelerometry and, in some cases, heart rate. Validation and interpretation of these devices for clinical purposes is discussed separately. (See "Actigraphy in the evaluation of sleep disorders", section on 'Consumer wearable devices'.)

SLEEP ARCHITECTURE — Sleep is not a homogenous process and appears to go through multiple discrete cycles during any given night. These cycles occur in fairly typical patterns of non-rapid eye movement (NREM) and rapid eye movement (REM) sleep, with a single cycle lasting approximately 90 to 120 minutes. As an example, the first sleep cycle will typically encompass the time from initial sleep onset until the patient exits from the first REM period. Four to five cycles occur during a typical eight-hour night of sleep (figure 10).

Sleep cycles trend through a standard night in the following ways:

The first cycle of the night starts with transition from wake to stage N1, then into stage N2, stage N3, and then REM.

As the cycles continue during the night, the percentage of REM sleep in each cycle generally increases.

The percentage of stage N3 tends to decrease over the course of the night, with the largest amount of N3 in the first half of the night.

Sleep architecture also varies across the lifespan (figure 11). Newborn infants sleep 16 to 18 hours per day in short blocks of time, without a clear circadian phase. They tend to enter sleep through REM as opposed to NREM sleep. Around three months of age, they begin to develop a day/night cycle and enter sleep through NREM sleep. Total sleep time slowly decreases, eventually reaching adult norms post-adolescence. (See "Sleep physiology in children", section on 'Maturation of sleep architecture'.)

Young adults typically sleep approximately eight hours per night with an elevated percentage of stage N3 sleep; as humans transition to middle age and beyond, the percentage of N3 decreases and the percentage of wake and N1 increases. It is notable, however, that the percentage of REM sleep is fairly stable throughout adulthood. Though often assumed to be reduced in older adults, the total amount of sleep time required by young adults and older adults does not appear to be dramatically different.

Normal sleep architecture is relevant to certain sleep disorders. As an example, NREM-related parasomnias are more likely to occur in the first half of the night, often within the first hour or two, when stage N3 sleep is most common. In contrast, REM-related parasomnias tend to occur more commonly in the second half of the night, when REM sleep percentage is higher. Obstructive sleep apnea may also be more prominent in the second half of the night, when REM-related physiological changes worsen sleep-disordered breathing.

It is important to recognize that sleep architecture on a polysomnogram may not reflect the patient's normal sleep at home; thus, reduced REM sleep on a study night may not necessarily be clinically relevant. In other cases, however, alterations in sleep architecture reflect an underlying sleep disorder, medical disorder, or the effect of a substance. Examples including the following:

REM sleep occurring soon after initial sleep onset suggests narcolepsy, depression, REM-rebound following medication withdrawal, or a circadian rhythm disorder.

Sleep fragmenting disorders, such as obstructive sleep apnea or periodic limb movements, often increase the number of sleep stage changes and may completely disrupt the normal cycling of sleep.

Prior acute or chronic sleep deprivation may cause increases in stage N3 sleep and REM sleep.

Ingestion or withdrawal from certain medications (eg, tricyclic antidepressants, monoamine oxidase inhibitors) or substances (eg, caffeine, alcohol) can have dramatic effects on sleep architecture.

Mood can affect sleep, such that patients with major depression may have shortened latency to REM sleep [14].

SLEEP AND PHYSIOLOGY — The stage designations discussed above are not solely EEG-based; there appear to be physiological differences between non-rapid eye movement (NREM) and rapid eye movement (REM) sleep. In general, deep NREM sleep (N3) tends to be a time of respiratory and cardiovascular stability. REM sleep is associated with more irregularities in heart rate, respiratory rate, blood pressure, and ventilation (table 2).

POTENTIAL FUNCTIONS OF SLEEP — Sleep is essential for brain function and physiologic health in a variety of ways.

Restorative function – A restorative theory proposes that the body repairs and revitalizes itself during the sleep state. Upon awakening from a full night of sleep, individuals typically feel restored. Conversely, insufficient sleep results in poorer daytime performance, a sensation of tiredness or sleepiness, and measurable effects on immune system function. (See "Insufficient sleep: Definition, epidemiology, and adverse outcomes".)

Sleep may have an important role in mitigating adverse consequences of stress. In a mouse model, neurons in the ventral tegmental area of the midbrain are activated by stress and drive the hypothalamus to induce sleep and inhibit corticotropin-releasing factor [15]. Growth hormone secretion also peaks during sleep; this may contribute to muscle growth and cell regeneration during the night.

Clearance function – Brain metabolism during sleep likely plays a restorative function via clearance of substances such as adenosine, which builds up over the course of the day and likely helps to induce deep NREM sleep [16]. The brain waste removal pathway has been termed the glymphatic system due to its dependence on glial cells [17,18]. In particular, sleep appears to be associated with an increase in interstitial space, leading to improved clearance of neurotoxic waste [19,20]. Experimentally, acute sleep deprivation in human volunteers impairs molecular clearance of an intrathecally-administered magnetic resonance imaging (MRI) contrast agent from the brain parenchyma, suggesting that transport of water-soluble metabolites excreted into the brain extravascular space is facilitated during sleep [21].

Brain plasticity and learning – Empirical and experimental data also suggest that sleep plays a role in brain plasticity by promoting learning-dependent synapse formation and maintenance [22-27]. Humans clearly learn less well when not sleeping enough; thus sleep must have an impact on cognitive function and memory.

Newborns spend more time asleep, and in particular, have a higher proportion of REM sleep. Some believe that the sensory input that occurs during dreaming, as well as motor cortex activity (which is blocked from expressing itself peripherally in normal REM sleep) plays an important role in brain development [28]. NREM sleep, meanwhile, may impact learning by returning saturated learning circuits back to baseline levels [29].


Sleep stages – Sleep is analyzed in 30-second epochs, each of which is categorized as rapid eye movement (REM) sleep or non-rapid eye movement (NREM) sleep. (See 'Sleep staging' above.)

NREM sleep – NREM sleep is divided into three stages: N1, N2, and N3. A fourth stage, N4, was previously recognized only under older scoring standards, but newer standards have incorporated stage N4 into stage N3. NREM sleep typically comprises the majority of the total sleep time in adults. (See 'NREM sleep' above.)

REM sleep – REM sleep occurs every 90 to 120 minutes. It is characterized by three main features: a low voltage mixed frequency EEG pattern, rapid eye movements, and voluntary muscle atonia, except the extraocular muscles and diaphragm. Though a minority of the sleep time, it has important roles in physiological homeostasis and cognition. (See 'REM sleep (stage R)' above.)

Sleep architecture – Sleep stages occur in cycles lasting 90 to 120 minutes each. Four to five cycles occur during a typical night of sleep. Shifting of stages occurs over the course of the night, typically with increased percentage of NREM sleep in the first half of the night (in particular, stage N3) and increased percentage of REM sleep in the second half of the night. Changes in typical sleep architecture may be representative of sleep disorders, though many other causes exist. (See 'Sleep architecture' above.)

Functions of sleep – The true purpose of sleeping is poorly understood. Empirical and experimental data support a variety of potential functions, including energy conservation, restoration and clearance of metabolites, and promotion of brain plasticity. (See 'Potential functions of sleep' above.)

  1. Siegel JM. Sleep viewed as a state of adaptive inactivity. Nat Rev Neurosci 2009; 10:747.
  2. Rechtshaffen A, Kales A (Eds). A manual of standardized terminology and scoring system for sleep stages of human subjects. 204, United States Government Printing Office; National Institutes of Health, Washington, DC 1968.
  3. Berry RB, Quan SF, Abreu AR, et al for the American Academy of Sleep Medicine. The AASM Manual for the Scoring of Sleep and Associated Events: Rules, Terminology and Technical Specifications, Version 2.6,, American Academy of Sleep Medicine, Darien, IL 2020.
  4. Magalang UJ, Chen NH, Cistulli PA, et al. Agreement in the scoring of respiratory events and sleep among international sleep centers. Sleep 2013; 36:591.
  5. Ohayon MM, Carskadon MA, Guilleminault C, Vitiello MV. Meta-analysis of quantitative sleep parameters from childhood to old age in healthy individuals: developing normative sleep values across the human lifespan. Sleep 2004; 27:1255.
  6. Agnew HW Jr, Webb WB, Williams RL. The first night effect: an EEG study of sleep. Psychophysiology 1966; 2:263.
  7. Proctor A, Bianchi MT. Clinical pharmacology in sleep medicine. ISRN Pharmacol 2012; 2012:914168.
  8. ASERINSKY E, KLEITMAN N. Regularly occurring periods of eye motility, and concomitant phenomena, during sleep. Science 1953; 118:273.
  9. Tononi G, Cirelli C. Perchance to prune. During sleep, the brain weakens the connections among nerve cells, apparently conserving energy and, paradoxically, aiding memory. Sci Am 2013; 309:34.
  10. American Academy of Sleep Medicine. International classification of sleep disorders: Diagnostic and coding manual, 2nd ed, Westchester, IL 2005.
  11. Weitzenblum E, Chaouat A. Sleep and chronic obstructive pulmonary disease. Sleep Med Rev 2004; 8:281.
  12. Howell MJ. Parasomnias: an updated review. Neurotherapeutics 2012; 9:753.
  13. España RA, Scammell TE. Sleep neurobiology from a clinical perspective. Sleep 2011; 34:845.
  14. Anderson KN, Bradley AJ. Sleep disturbance in mental health problems and neurodegenerative disease. Nat Sci Sleep 2013; 5:61.
  15. Yu X, Zhao G, Wang D, et al. A specific circuit in the midbrain detects stress and induces restorative sleep. Science 2022; 377:63.
  16. Porkka-Heiskanen T, Zitting KM, Wigren HK. Sleep, its regulation and possible mechanisms of sleep disturbances. Acta Physiol (Oxf) 2013; 208:311.
  17. Hauglund NL, Pavan C, Nedergaard M. Cleaning the sleeping brain - the potential restorative function of the glymphatic system. Curr Opin Physiol 2020; 15:1.
  18. Komaroff AL. Does Sleep Flush Wastes From the Brain? JAMA 2021; 325:2153.
  19. Xie L, Kang H, Xu Q, et al. Sleep drives metabolite clearance from the adult brain. Science 2013; 342:373.
  20. Lewis LD. The interconnected causes and consequences of sleep in the brain. Science 2021; 374:564.
  21. Eide PK, Vinje V, Pripp AH, et al. Sleep deprivation impairs molecular clearance from the human brain. Brain 2021; 144:863.
  22. Yang G, Lai CS, Cichon J, et al. Sleep promotes branch-specific formation of dendritic spines after learning. Science 2014; 344:1173.
  23. Diering GH, Nirujogi RS, Roth RH, et al. Homer1a drives homeostatic scaling-down of excitatory synapses during sleep. Science 2017; 355:511.
  24. de Vivo L, Bellesi M, Marshall W, et al. Ultrastructural evidence for synaptic scaling across the wake/sleep cycle. Science 2017; 355:507.
  25. Brüning F, Noya SB, Bange T, et al. Sleep-wake cycles drive daily dynamics of synaptic phosphorylation. Science 2019; 366.
  26. Noya SB, Colameo D, Brüning F, et al. The forebrain synaptic transcriptome is organized by clocks but its proteome is driven by sleep. Science 2019; 366.
  27. Girardeau G, Lopes-Dos-Santos V. Brain neural patterns and the memory function of sleep. Science 2021; 374:560.
  28. Siegel JM. Clues to the functions of mammalian sleep. Nature 2005; 437:1264.
  29. Tononi G, Cirelli C. Sleep and the price of plasticity: from synaptic and cellular homeostasis to memory consolidation and integration. Neuron 2014; 81:12.
Topic 7710 Version 32.0