What is rem sleep?

REM stands for rapid eye movement, and the name describes the most visible sign of it: the eyes move quickly beneath closed lids. Brain activity during REM looks strikingly similar to being awake, which is why it is sometimes called paradoxical sleep.

The other defining feature is muscle atonia. During REM the body is effectively paralysed from the neck down, with the muscles controlling breathing and the eyes exempted. This is generally understood as a mechanism that stops you acting out what you are dreaming, and its failure is the basis of REM sleep behaviour disorder.

Most vivid, narrative dreaming happens in REM, though dreaming is not exclusive to it. Breathing and heart rate become more irregular than in deep sleep, and body temperature regulation is reduced.

REM is not spread evenly through the night. The first REM period typically arrives around an hour to ninety minutes after falling asleep and is short, often only a few minutes. Later periods get progressively longer.

That distribution matters more than the total. Deep, slow-wave sleep is concentrated in the first half of the night; REM is concentrated in the second. Cutting a night short by two hours therefore does not remove a proportional slice of everything. It removes disproportionately more REM than deep sleep.

Across a full night, REM accounts for roughly a fifth to a quarter of total sleep in a healthy adult, and the proportion is much higher in infants.

Waking from REM generally produces less sleep inertia than waking from deep sleep. Because brain activity during REM already resembles wakefulness, the transition is shorter, and people woken from REM tend to feel more immediately coherent.

That single fact is the seed of a large amount of overstated advice, so it is worth being careful with. It is true that waking from light sleep or REM feels better than waking from slow-wave sleep. It does not follow that you can reliably arrange to be woken from it.

Since REM clusters in the second half of the night, a normal morning alarm is already reasonably likely to land near light sleep or REM rather than in deep sleep. That is part of why waking at 7am after a full night feels different from being woken at 2am, and it needs no app to arrange.

Two claims circulate more confidently than the evidence supports, and both come up constantly in the context of alarms.

The first is that you can time an alarm to a REM or light-sleep window using a phone on the mattress. Consumer sleep tracking infers stages from movement and sometimes heart rate, and while that can approximate broad patterns, it does not reliably identify sleep stage in the way polysomnography does. Advice built on the assumption that it does is building on sand.

The second is that REM is the important sleep and deep sleep is not, or the reverse. Both stages appear to serve distinct and necessary functions, and being deprived of either has consequences. Ranking them is a simplification that mostly exists to make an app feature sound meaningful.

The genuinely useful implication is duller and better supported: get enough total sleep, and keep the timing consistent, and the stage distribution largely takes care of itself.

Alcohol is the clearest example, and the effect is more specific than general sedation. Drinking in the evening tends to shorten the time it takes to fall asleep while suppressing REM in the first part of the night, then producing a rebound later with lighter, more fragmented sleep and more waking. This is why a night after drinking can be long in duration and poor in quality.

Sleep restriction disrupts it structurally rather than by suppression. Because REM is concentrated in the second half of the night, cutting the night short removes disproportionately more of it than of deep sleep. A person who consistently sleeps five hours is not getting five-sevenths of a normal night; they are getting most of their deep sleep and a fraction of their REM.

A number of medications affect REM, including some antidepressants, and stopping them can produce a rebound. This is worth knowing about rather than acting on alone, since it is a question for whoever prescribed them.

Sleep apnoea disrupts it in a particularly costly way. Because muscle tone is lowest during REM, apnoea events are often worst then, which means the stage that is hardest to accumulate is also the one most likely to be interrupted.

Most vivid, narrative dreaming happens during REM, though it is not exclusive to it and dreaming of a vaguer kind occurs in other stages. Dreams recalled on waking are disproportionately REM dreams for a simple reason: you are more likely to wake out of REM in the second half of the night, and recall is best immediately after the dream.

Nightmares are REM events, which is why they cluster towards morning rather than shortly after falling asleep. That timing is the most reliable way to distinguish a nightmare from a sleep terror, which is a disorder of arousal out of deep sleep, occurs early in the night, and is usually not remembered at all.

The muscle atonia of REM occasionally fails, and there are two familiar consequences. Sleep paralysis is atonia persisting briefly into waking, so the person is aware but unable to move; it is common, harmless in itself, and often frightening. REM sleep behaviour disorder is the opposite failure, where atonia is absent during REM and the person physically acts out dreams. Unlike sleep paralysis it is worth medical attention, both for injury risk and because it can be associated with other conditions.

None of these is caused by an alarm, but the timing is worth knowing, because waking during the second half of the night is more likely to interrupt REM and to produce a remembered dream than waking in the first.