NREM and REM sleep

Jump to: navigation, search

This text is part of: "Science of sleep" by Piotr Wozniak (2017)

NREM and REM alterations

In the course of the night, we alternately enter two phases of sleep:

The interplay of those NREM and REM episodes is vital for understanding memory optimization in sleep, creativity and problem solving.

Using EEG measurements, scientists are able to distinguish 4 phases of NREM sleep which correspond to progressively deeper sleep. In newer literature you may read of three stages due to the fact that Stages 3 and 4 of NREM have been bundled together as a single stage of slow-wave sleep.

NREM-REM cycling through a typical night

NREM-REM cycling through a typical night.

As we close our eyes, it takes 3-15 minutes to enter Stage 1 NREM sleep (in a healthy and well-regulated individual). In this stage we will often experience little jerks associated with the impression of falling. Minor disturbances will wake us up and often we will even deny being asleep! Once State 1 NREM solidifies, we move towards Stage 2 NREM sleep which is still relatively light. After that we move to Stage 3 (and Stage 4) NREM, which is also called deep sleep or slow-wave sleep (SWS).

Historically, the importance of REM sleep for memory and learning was documented before we became truly aware of the role of slow-wave sleep. Consequently, articles and books on sleep are peppered with an overemphasis on the role of REM sleep in learning as compared with SWS. Over time, REM deprivation studies received lots of criticism. Today, we know that the natural harmonious interplay of uninterrupted NREM and REM sleep is essential for memory, learning and creativity (Salzarulo et al. 2000[1]).

Cruel sleep deprivation studies actually show that sleep deprived rats can live longer if REM deprived than if NREM deprived. Rats deprived of sleep survive for 2-3 weeks. Rats deprived of REM sleep only survive for some five months.

Napping human subjects reported that it is Stage 4 NREM that feels most restorative. The release of norepinephrine, serotonin and histamine is inhibited during REM. During dreaming, the primary visual cortex is not active, while its secondary areas are active. This is similar to the situation in which people are asked to imagine a visual scene as opposed to a situation in which they actually see the scene. Blind people have dreams that are more auditory and more tactile. This seems to confirm the role of REM sleep in the replay of experiences and in optimization of memories. They do not show the typical eye movement pattern in REM sleep either. Those observations led to an idea that REM sleep is vital for creativity (more than NREM sleep). During REM, cholinergic modulation suppresses the flow of information from the hippocampus to the neocortex. This is supposed to help build new associations within the neocortical areas.

Evolution of NREM and REM

REM sleep is phylogenetically younger than NREM sleep. Fish, amphibians or reptiles do not show typical REM sleep. Yet, interestingly, REM sleep is present in both mammals and birds. This made some scientists hypothesize that REM sleep has been invented twice by evolution! Clearly, REM sleep plays a role critical for survival of creatures with bird-mammal IQ levels (see: How much do animals sleep?). However, Dr Siegel who studied REM sleep in echidna concluded that this animal's sleep combines aspects of both REM and NREM sleep. As a result, he suggested that REM and NREM might have evolved from a phylogenetically older unified form of sleep (Siegel et al. 1996[2]). If REM sleep is as disparate from NREM in its function and as complex as implied by the theories on the neural optimization in sleep, the re-invention factor might be used by evolutionists as an argument against neural optimization. However, like aerial flight, re-invention combined with complexity could equally well add weight to emphasizing the vital neural function of sleep.

NREM and REM deficits

Cruel sleep deprivation studies actually show that sleep deprived rats can live longer if REM deprived than if NREM deprived. Rats deprived of sleep survive for 2-3 weeks. Rats deprived of only REM sleep survive for some five months. After sleep deprivation, it is the SWS deficit that is repaid first. SWS deficit is a result of NREM sleep deprivation. REM deficits are paid off later. A frequent scenario is that the SWS deficit is paid fully on the first night of recovery, while REM sleep deficit may persist through to the second recovery night.

Forced desynchrony protocols are sleep protocols in which subject sleep is dissociated from its natural circadian cycle. In condition of forced synchrony, we can observe that slow-wave activity that characterizes NREM sleep is associated with homeostatic sleep propensity, while the proportion of REM sleep in sleep episodes depends on both homeostatic pressure and the circadian cycle.

NREM control

NREM sleep is primarily controlled via a homeostatic mechanism. During the waking day we build a pressure to initiate sleep and its deeper NREM stages. If sleep is initiated without a contribution of the circadian component, it is likely to be short and NREM-only. One of the signals correlating with homeostatic sleepiness is the buildup of adenosine. It is the adenosine receptors that are affected by caffeine resulting in its short-lived impact on reducing the homeostatic sleep pressure. One of the consequences of the buildup of adenosine is the inhibition of the aminergic wake centers, inhibition of the basal forebrain, and the disinhibition of the VLPO: the chief brain center responsible for the initiation of sleep. REM sleep also has a homeostatic component, however, in times of deficit, it is NREM sleep deficit that is compensated for first. There is also more evidence indicating that REM sleep increases NREM sleep pressure (Beersma et al. 1990[3]). In addition to adenosine, other signals such as interleukin-1, tumor necrosis factor, interferon, prostaglandin D2, NO, GHRH, and others have also been associated with the increase in homeostatic sleep propensity (Krueger et al. 2008[4]).

Sleep deprivation increases both NREM and REM sleep propensity. Short sleepers have less NREM 2, but there is little data on the actual quality and effectiveness of their sleep. Thomas Edison or Nicola Tesla, on one hand, are well-known for sleeping relatively little, while Einstein is a well-known long sleeper, who, supposedly, slept over nine hours per night. Interestingly, all these geniuses also belonged to notable nappers. It is true that by getting less sleep you compress the less critical NREM 2 sleep, but there is no evidence that this can become your regular habit without hurting the quality of your NREM-REM interplay. With the currently available sleep data the conclusion is: do not try to compress NREM 2 by sleeping less. You are likely to hurt the memory optimization process occurring in sleep.

Some scientists believe that during sleep, an ultradian oscillator in the mesopontine junction controls the regular alternation of NREM and REM sleep. However, the term oscillator is rather misleading as the mechanism of NREM-REM mutual interaction is more of a flip-flop nature, and the timing of alternation is pretty irregular indicating significant internal and external homeostatic influences that ultimately culminate in the extinction of the sleep cycle.

Neuromodulation in sleep

The AIM model of sleep ((Hobson et al. 2000[5]) emphasizes the importance of cholinergic modulation that dominates REM in the absence of aminergic activity.

In NREM sleep, cholinergic systems in the brainstem and the forebrain are less active than in waking. Serotonergic raphe and noradrenergic LC are also less active. On the other hand, in REM sleep, these aminergic structures are strongly suppressed, while the cholinergic systems flare up. Release of histamine is also down in REM. It has been hypothesizes that cholinergic modulation suppresses the flow of information from the hippocampus to the neocortex. This is supposed to play an important role in the dual network model of learning in which the hippocampus plays a role in building up new associations on the basis of old information (see: Neural optimization in sleep).

Dr Siegel, who does not believe in the role for sleep in memory and learning, believes that REM sleep serves recovery as serotonergic, noradrenergic, and histaminergic neurons stop firing. It is as if they were overused and attempted to replenish their resources. This interpretation might pass the shutdown test (see: Sleep theories) as many of these neurons are vital for maintaining arousal. However, it is hard to imagine that evolution would not find a way to re-design the brain in which neurotransmitter replenishment would be possible without the shutdown. Some areas of the brain keep firing in waking and as well as in all stages of sleep. The neurons involved are able to replenish their resources without going offline.

Growth hormone and cortisol

In addition to changes in firing patterns of neurons releasing different types of neurotransmitters, circulation of systemic hormones also changes during sleep. Of these, growth hormone and cortisol are of particular importance as they impact glucose metabolism. Growth hormone increases at sleep onset and peaks in deeper stages of NREM. On the other hand, cortisol levels increase in the later stages of the night dominated by REM sleep. Unlike the release of cortisol, which is largely circadian, the increase in GH is associated with a sleep onset (Van Cauter et al. 1997[6]). Sleep deprivation or sleeping in a wrong phase are both involved in major disruption of glucose metabolism for different reasons. This is why healthy sleep is vital for preventing obesity.

REM Homeostasis

In the same way as sleep in general, REM is controlled via homeostatic and circadian processes. Acrophase of the circadian REM cycle comes late in the subjective night. Benington and Heller proposed that it is the presence of NREM sleep rather than the absence of REM sleep that leads to an increase of REM sleep propensity. Slow-wave sleep builds homeostatic REM propensity, and the best REM comes from the combination of slow-wave "exhaustion" and the circadian REM peak which comes in the last hours of sleep. There is also a strong homeostatic link between learning and the demand for REM sleep. The more you learn, the stronger the drive towards REM. There is an increase in both the number of minutes of REM sleep and the density of REM sleep following intensive learning (De Koninck et al. 1989[7]). It is not clear if learning affects REM demand directly or via NREM demand. However, it is more than clear that heavy learners should be heavy sleepers!

Stimulating the basal forebrain causes a release of acetylcholine, which induces wakefulness and is also conducive to REM sleep. This means that the basal forebrain that takes part in the initiation of sleep is also involved in NREM/REM transitions. Similarly, a subset of VLPO cells contribute to generating REM sleep.

The impact of adenosine antagonists, such as caffeine, is also important. Adenosine agonists infused into the basal forebrain increase c-Fos in the VLPO as well as increase the release of acetylcholine by the basal forebrain. Acetylcholine is known to induce the states of wakefulness and REM sleep. As a result of the agonist infusion, both the total amount of NREM and the total amount of REM sleep increase (Satoh et al. 1999[8], Scammell et al. 2001[9]).

Of other homeostatic hormonal influences, increased levels of VIP and prolactin in sleep promote REM. It is possible that substance abuse, delaying sleep, as well as the use of alarm clocks can all read to REM sleep deficits (see: REM rebound hypothesis).

One of the sleep theories says that REM sleep helps the brain recover from NREM sleep to speed up the responses in waking. This theory fails the shutdown test as the same recovering might simply be taking place in a waking state unless the hard work of the networks in dreaming is a faster recovery method for some unknown reason. However, why would a brain experience a REM rebound in conditions of full "recovery" to waking? The claim that histamine, serotonin and noradrenaline neurons need recovery time sound more plausible, however, it does not explain why we would need different populations of neurons with different neurotransmitters with different restoration and recovery strategies.

Transition to REM

After an hour or so of healthy NREM sleep during the subjective night sleep, there is a gradual increase in the activity of cells in the pontine tegmentum which is responsible for triggering REM sleep. Structures responsible for triggering REM sleep might include pedunculopontine tegmental nucleus (PPN) and sublaterodorsal tegmental nucleus (SLD). GABAergic SLD and cholinergic PPN send their signals in multiple directions. One of the outcomes is muscle atonia. Another is the activation of the thalamus, hippocampus, and the cortex with an appearance of the typical REM EEG. As a result the brain behaves as if it woke up internally! Injections of acetylcholine into the pons during an ongoing NREM episode may trigger REM sleep, which illustrates the importance of this neurotransmitter in sleep cycle regulation. During REM sleep, the cortex behaves as in the state of wakefulness. Dreams experienced at that stage seem to be generated by random impulsation sent from the brainstem to the cortex. The cortex produces best possible and most coherent imagery of that chaotic input. During dreams we experience connected events, real people, realistic scenery, etc. However, all these are put together in most improbable configurations as if the brain was testing "what if" scenarios. Yet we cannot act upon our dreams (except for people who suffer from violent sleeping). Pontine structures responsible for REM control make sure that the cerebral output gets cut off from motor nuclei that move the muscles. It happens often that we want to act in sleep (e.g. to escape a ferocious dog), and yet we remain motionless as if mired in molasses. At that time, only the eyes move rapidly, while the muscles in the middle ear also twitch.

The movements of eyeballs that gave REM its name is controlled by impulsation generated in the pontine nucleus that projects to the superior colliculi. That impulsation is associated with generating of the ponto-geniculo-occipital waves (PGO) that are also used to detect REM sleep.

During dreaming, the primary visual cortex is not active, while its secondary areas are active. This is similar to the situation in which people are asked to imagine a visual scene as opposed to a situation in which they actually see the scene. Blind people have dreams that are more auditory and more tactile. This seems to confirm the role of REM sleep in the replay of experiences and in optimization of memories. They do not show the typical eye movement pattern in REM sleep either. Those observations lead to an idea that REM sleep is vital for creativity (more than NREM sleep).

REM flip-flop

The interplay between NREM and REM sleep is most likely controlled by a REM flip-flop. Some scientists believe that during sleep, an ultradian oscillator in the mesopontine junction controls the regular alternation of NREM and REM sleep. However, a flip-flop model is a better analogy considering the timing of the alternations.

As it is the case with the sleep-wake flip-flop, the REM flip-flop causes a continuous switches between a relatively stable NREM and REM states, however, the flip-flop is under far greater influence of various homeostatic inputs resulting in a somewhat chaotic succession of NREM/REM states that gradually become dominated by REM circadian peak towards the end of the subjective night.

For many years, an oscillation between cholinergic and monoaminergic states seemed like a final answer to the control of REM sleep. However, some inconsistencies and new research lead to a newer similar model involving GABAergic structures (see the next section).

Models of REM flip-flop

In 1962, Jouvet showed that stimulation of the caudal mesencephalic region or pontine tegmentum in cats produced a state similar to REM sleep. This led to a hypothesis that mesopontine cholinergic structures are responsible for the activation of the thalamus and the cortex in REM sleep. The hypothesis would also be supported by the fact that injections of cholinergic agonists into the pontine reticular formation would enhance REM sleep.

The original reciprocal interaction model in which pontine aminergic and cholinergic neurons have formed a classical REM-on/REM-off flip-flop has been accepted as a fact for a quarter of a century. New research has identified GABAergic populations that might be part of the REM flip-flop on both on and off sides of the switch.

The old REM flip-flop included cholinergic PPT and LDT, which are particularly active in REM (and wakefulness), as well as the BRF (brainstem reticular formation). The REM-off component was composed of DR (serotonin) as well as the LC (NA) (Saper et al. 2001[10]).

Newer research questioned some inconsistencies in the model. For example, selective lesions to cholinergic or monoaminergic nuclei of the brainstem have only limited effect on REM sleep. Low c-Fos expression in REM-on structures during REM was also troubling. Instead, it was suggested that the key REM-on area is the GABA-ergic SLD (sublaterodorsal tegmental nucleus)(Lu et al. 2007[11]). As the SLD does not directly inhibit DR-LC, their direct participation in the flip-flop was questioned as well. GABA-ergic neurons of SLD project to the vlPAG (ventrolateral periaqueductal grey matter) and LPT (lateral pontine tegmentum), which thus became REM-off suspects. Lesions of the SLD cause a loss of REM sleep.

The precise nature of the REM flip-flop must yet be determined. Components of the old and new models show some interaction as well. For example, PPT/LDT do excite SLD neurons, while DR/LC may inhibit the SLD or activate the REM-off components. However, that interaction is not directly mutual. Hence the exclusion of the old components from the core of the new model.

Termination of sleep

The sleep control system would act as an infinite seesaw were it not for the circadian component of the sleep drive. Towards the end of sleep, the circadian sleepiness determined by the suprachiasmatic nucleus (SCN) will produce decline in sleep propensity, and the sleep will be terminated after one of the REM sleep episodes. It is the SCN which provides the link between the strongest zeitgeber, the light, and the circadian cycle. SCN generates the rhythm endogenously, but is able to get reset by light. Light impulses from the retina travel to the hypothalamus and the SCN to produce a stop&reset signal. End of sleep will see the end of melatonin release. Instead, another neurohormone starts building up: serotonin. A hypothesis says that it is the high level of serotonin that we feel as the morning sunshine happiness. High serotonin combines with the alertness hormone cortisol to give us a good alert start into a new day. Unless you suffer from sleep phase advancement, always make sure the sunshine streams into your sleeping room in the morning to wake you up.


  1. Ficca G., Lombardo P., Rossi L., and Salzarulo P., "Morning recall of verbal material depends on prior sleep organization," Behavioural Brain Research / Volume 112 / Issues 1-2 (July 2000): 159-163, doi: 10.1016/S0166-4328(00)00177-7
  2. Siegel J.M., Manger P.R., Nienhuis R., Fahringer H.M., and Pettigrew J.D., "The Echidna Tachyglossus aculeatus Combines REM and Non-REM Aspects in a Single Sleep State: Implications for the Evolution of Sleep," The Journal of Neuroscience / Volume 16 / Issue 10 (May 15, 1996): 3500-3506
  3. Beersma D.G.M., Dijka D.J., Bloka C.G.H., and Everhardusa I., "REM sleep deprivation during 5 hours leads to an immediate REM sleep rebound and to suppression of non-REM sleep intensity," Electroencephalography and Clinical Neurophysiology / Volume 76 / Issue 2 (August 1990): 114-122, doi: 10.1016/0013-4694(90)90209-3
  4. Krueger J.M., Rector D.M., Roy S., Van Dongen H.P.A., Belenky G., and Panksepp J., "Sleep as a fundamental property of neuronal assemblies," Nature Reviews Neuroscience / Volume 9 / Issue 12 (December 2008): 910-919, doi: 10.1038/nrn2521
  5. Hobson J.A., Pace-Schott E.F., and Stickgold R., "Dreaming and the brain: Toward a cognitive neuroscience of conscious states," Behavioral and Brain Sciences / Volume 23 / Issue 6 (2000): 793-842
  6. Van Cauter E., Polonsky K.S., and Scheen A.J., "Roles of Circadian Rhythmicity and Sleep in Human Glucose Regulation," Endocrine Reviews / Volume 18 / Issue 5 (October 1, 1997): 716-738, doi: 10.1210/er.18.5.716
  7. De Koninck J., Lorrain D., Christ G., Proulx G., and Coulombe D., "Intensive language learning and increases in rapid eye movement sleep: evidence of a performance factor," International Journal of Psychophysiology / Volume 8 / Issue 1 (1989): 43-47
  8. Satoh S., Matsumura H., Koike N., Tokunaga Y., Maeda T., and Hayaishi O., "Region-dependent difference in the sleep-promoting potency of an adenosine A2A receptor agonist," European Journal of Neuroscience / Volume 11 / Issue 5 (May 1999): 1587–1597, doi: 10.1046/j.1460-9568.1999.00569.x
  9. Scammell T.E., Gerashchenko D.Y., Mochizuki T., McCarthy M.T., Estabrooke I.V., Sears C.A., Saper C.B., Urade Y., and Hayaishi O., "An adenosine A2a agonist increases sleep and induces Fos in ventrolateral preoptic neurons," Neuroscience / Volume 107 / Issue 4 (November 28, 2001): 653–663, doi: 10.1016/S0306-4522(01)00383-9
  10. Saper C.B., Chou T.C., and Scammell T.E., "The sleep switch: hypothalamic control of sleep and wakefulness," Trends in Neurosciences / Volume 24 / Issue 12 (December 1, 2001): 726-731, doi: 10.1016/S0166-2236(00)02002-6
  11. Fuller P.M., Saper C.B., and Lu J., "The pontine REM switch: past and present," The Journal of Physiology / Volume 584 / Issue 3 (November 1, 2007): 735-741, doi: 10.1113/jphysiol.2007.140160