Late in 1951, in a laboratory at the University of Chicago, a graduate student named Eugene Aserinsky stuck wires to the head and the skin around the eyes of his eight-year-old son and recorded him through the whole night. The wires led to a machine that drew the brain’s electrical activity and the movements of the eyes as wavy lines on a moving strip of paper. Aserinsky expected the eyes to lie still for long stretches. He had watched sleeping babies and seen their eyes go quiet for about twenty minutes in every hour, and he wanted to know whether older sleepers did the same.
The pens told him otherwise. Several times in the night, for about twenty minutes at a time, the boy’s eyes jerked about under their lids, and the brain’s lines showed it busy, not resting. Aserinsky’s first name for these stretches was “Jerky Eye Movement periods”. His professor, Nathaniel Kleitman, played the skeptic and demanded proof beyond doubt, and Aserinsky later thought Kleitman’s feelings must have been mixed: Kleitman’s own theory was that sleep is simply the brain going quiet. In 1953 the two of them published what they had found, under the better name that stuck, rapid eye movement, or REM.
Most of what follows grew from that discovery. Sleep isn’t the brain switching off. It is a busy state of its own, with a structure you can draw, a clock that times it, and a purpose that scientists still argue about.
A night in stages
Sleep comes in two kinds. Most of the night is non-REM sleep, which has three stages, from light to deep. The rest is REM sleep, the kind Aserinsky found. Scientists tell them apart with the same three measures his machine used: brain waves (an EEG, electroencephalogram), eye movements (EOG) and the tension in the muscles (EMG).
Stage 1 is the doorway, a few minutes of drifting in which breathing and the heart slow, the muscles relax, and a person woken up will often insist they weren’t asleep at all. Stage 2 is light sleep: the eyes stop moving, the body cools a little, and the slow brain waves are broken now and then by short bursts of quicker activity. Most of a night is spent here. Stage 3 is deep sleep, also called slow-wave sleep, because the brain’s waves become big and slow. The heart and breathing reach their slowest, and someone in stage 3 is hard to wake.
Then, about an hour and a half after falling asleep, the brain climbs back up and something strange happens. The waves become fast and mixed, like those of a waking brain. Breathing turns quick and uneven, the heart speeds up, and the eyes dart about. Yet the muscles of the arms and legs go limp, held still by signals from the brainstem. This is REM sleep, sometimes called paradoxical sleep: the brain is at its busiest while the body is at its most relaxed. It is when the most vivid dreams happen, and the paralysis seems to be there so that we don’t act them out.
The night repeats this climb and fall, each round taking about ninety minutes, four or five times. But the rounds aren’t all alike. Deep sleep comes mostly in the first half of the night. REM periods start short and get longer toward morning. So a night cut short in the morning loses mostly REM sleep.
This pattern changes over a lifetime. Newborn babies sleep 16 to 18 hours a day, and a far larger share of it is REM sleep (in babies it’s called active sleep) than in adults, for whom REM is about a fifth to a quarter of the night. As people grow old they spend less of the night in REM. Whether all this REM matters for a growing brain is a good question; nobody has a firm answer.
Two clocks
In June 1938 Kleitman and a younger student, Bruce Richardson, climbed down into Mammoth Cave in Kentucky and stayed there for about a month. Deep in the cave there was no sunrise and no sunset, and the air stayed at about 12 °C. They wanted to know whether a person could live a 28-hour day, sleeping and working by the clock instead of the sun. Richardson, who was 25, settled into the 28-hour day. Kleitman, 43, never did: his body temperature, which rises and falls through a day, kept going up and down on a 24-hour beat. Two men prove little, but the lesson has held up in far bigger studies since. The body keeps time by itself.
That inner timekeeping is a circadian rhythm, from the Latin for “about a day”. It runs through body temperature, hormones, hunger and alertness as well as sleep. Measured carefully in people kept away from all clues to the time of day, the human clock runs a little long, about 24 hours and 11 minutes on average. So every day it needs nudging back into step with the real day.
The nudge comes mostly from light. The master clock is a small cluster of nerve cells deep in the brain called the suprachiasmatic nucleus, sitting just above where the nerves from the two eyes cross. Special cells in the eyes report light straight to it. When the light fades in the evening, the clock signals the pineal gland, which releases the hormone melatonin, and melatonin helps the body toward sleep. Light is the strongest of the cues that set the clock, but there are others: meals, work, school, the people around you.
When the clock and the world disagree, you feel it. Fly from Kathmandu to New York, where the clocks are 10 hours and 45 minutes behind, and your body still thinks it’s in Kathmandu: wide awake at 2 a.m. and drooping at lunch. That’s jet lag, and it fades as daylight resets the clock. Night-shift workers live with a version of it all the time, trying to sleep while their clock is telling them to be up.
There is a second process too, quite separate from the clock. Every hour you are awake, a pressure to sleep builds up, and sleep lets it drain away. Part of this pressure seems to come from a chemical called adenosine, which builds up in the brain while you are awake and makes you feel sleepy. Caffeine works by blocking adenosine’s effect, which is why tea or coffee keeps sleepiness away for a while. Bedtime, in this picture, is where the two processes meet: the pressure is high, and the clock has opened the gate.
There’s a Nobel Prize behind the clock’s machinery, given in 2017 for finding the genes that make it tick in flies.
The teenage clock
If you have ever been told you are lazy for wanting to sleep till nine, here is the science. Around puberty, the body clock shifts later. Teenagers start to feel sleepy later at night and to wake later in the morning, and it isn’t only habit or phones. In one early study of nearly 460 sixth-graders in the United States, the girls further into puberty leaned more toward being “evening people”, whatever their school or family, and the boys showed the same trend more weakly. Later work found the second process changing too: in older teenagers, the pressure to sleep builds more slowly through the day, so staying up late gets easier.
School doesn’t move with the clock. So on school days many teenagers sleep too little, and on free days they sleep long and late to make up. The jump between the two is called social jet lag: like flying across time zones and back every week, without leaving home. Laboratory studies of teenagers who lost sleep on school nights and caught up at weekends found that the catching up didn’t fully repair the loss.
How much do teenagers need? A panel of the American Academy of Sleep Medicine, after weighing the studies in 2016, put it at 8 to 10 hours in every 24 for ages 13 to 18. That is an expert judgment from a range of evidence, not a measured law, and people of the same age differ a good deal.
If you’d like small things to try, two tips on this site are about it: keeping the same bedtime and sleeping before an exam.
What short sleep does, and a book that scared the world
The clearest evidence on short sleep comes from experiments, where people are randomly given more or less of it. In one in Philadelphia, 48 healthy adults lived in a laboratory and were allowed 8, 6 or 4 hours in bed each night for two weeks, while a fourth group stayed awake for three nights running. Those held to 6 hours a night got steadily worse at tests of alertness and thinking, until after two weeks they did about as badly as people kept up for one or two whole nights. The odd part is that they hardly noticed. Their feeling of sleepiness rose in the first days and then leveled off, while their performance kept falling. Short sleep, it seems, is easy to get used to and hard to judge from the inside.
Outside the laboratory the picture is less clear. People who sleep little for years are more likely to have high blood pressure, heart disease, diabetes, depression and obesity. But these come from watching people, not from experiments, and they leave the old question open: does short sleep lead to illness, or does illness, worry or a hard life cut sleep short? Probably some of both; the scientists who study it list this as one of the field’s open questions.
In 2017 a neuroscientist at the University of California, Berkeley, Matthew Walker, published a book called Why We Sleep. It became a bestseller, and most reviewers liked it: it explained the science of sleep with lively comparisons. It also made sleep sound like an emergency. Routinely sleeping less than six or seven hours, it said, more than doubles your risk of cancer; the World Health Organization had declared a sleep loss epidemic; “the shorter your sleep, the shorter your life span”.
In 2019 Alexey Guzey, an independent researcher, posted a long essay online that went through the book’s first chapters claim by claim and listed what he called errors. Walker replied publicly later that year. He accepted several: short sleep is not linked to every form of cancer, and the book would say so; the “epidemic” statement came from the US Centers for Disease Control, not the WHO; a graph of sleep and sports injuries had left out its group with the least sleep, which he said was too small to give a firm estimate; some numbers and sample sizes were wrong, and a few percentages that were impossible would be fixed. He defended his main case: that six hours or less is too little for an average adult, and that short sleep goes with a shorter life. Guzey had pointed out that most studies find long sleepers dying sooner too, so that the link is U-shaped. Walker answered that a later chapter of his own book says so, and that long sleep may be a sign of poor health rather than a cause of it. Walker holds that the corrections leave his main case standing; Guzey holds that the errors run through the book. How much of the short-sleep link is cause hasn’t been settled.
The book also had a graph showing people’s sleep falling by more than two hours since the 1940s, numbers Guzey couldn’t find any data for. Behind it lies a common idea: that people slept far more before electric light, and that lamps, television and the internet have stolen hours from everyone. Some scientists tested it. In 2015 a team strapped movement recorders to the wrists of people in three societies with no electricity: the Hadza hunter-gatherers of Tanzania, the San of the Kalahari in Namibia, and the Tsimané, who hunt and farm in Bolivia. They slept between 5.7 and 7.1 hours a night on average, about the same as people in industrial countries, or a little less. They didn’t fall asleep at sunset, either, but about three hours after it, and they woke before sunrise. Neither the San nor the Tsimané had a word for insomnia. It’s only three small groups, but they don’t suggest a lost paradise of nine-hour nights.
Closer to home, in 2021 researchers asked 390 students in Grades 9 and 10 at two schools in Kirtipur about their internet use and their sleep. About three in ten slept poorly by a standard questionnaire, and those whose internet use looked like an addiction were much more likely to be among them. It’s tempting to read that as “the internet ruins sleep”, but a survey taken at one moment can’t say which came first. A student who can’t sleep may reach for the phone, and something else, such as worry, may drive both. The authors say so themselves. It’s a useful first look at teenagers in Nepal, who are rarely studied, and a good example of why a link isn’t a cause.
Why sleep, and why dream
For all this, nobody knows for sure what sleep is for. There are several good ideas, and they may all be partly right.
The oldest is restoration: sleep lets the body and brain recover from the day. It fits the way the pressure to sleep builds with waking and drains with sleep, and the way deep sleep deepens after a long day. Against it, much of the body’s repair, of muscles for instance, happens just as well during quiet rest while awake, so the idea has to explain why that isn’t enough.
The second is memory. For more than a century, experiments have found that sleep after learning helps people keep what they learned. In today’s leading account, memories made during the day are replayed by the brain in deep sleep and moved into longer-lasting storage, and REM sleep may then steady them. The basic finding, that sleeping on something helps you keep it, has been seen again and again; how it happens is still being worked out.
The third made headlines. In 2013 a team in Rochester, New York, found that in sleeping mice, the spaces between brain cells grew by about 60%, letting fluid wash through and carry away waste, including beta-amyloid, a protein that can harm nerve cells. Sleep, it seemed, was the brain’s night-time cleaning shift. Then in 2024 a team in London measured it differently, also in mice, and found the opposite: clearance was lower during sleep, not higher. The two can’t both be right as they stand, and the argument is still going on.
Dreams are harder still. Everyone dreams, perhaps two hours a night, though most dreams are forgotten. The most vivid come in REM sleep, but people woken from non-REM sleep sometimes report dreams too, so dreaming isn’t only a REM affair.
The Austrian neurologist Sigmund Freud, in The Interpretation of Dreams (1899), argued that dreams are wishes in disguise. The story you remember, the manifest content, hides a meaning underneath, the latent content, which an expert can uncover. It became one of the most famous ideas of the twentieth century, but it is very hard to test: two interpreters can read the same dream in quite different ways, and nothing decides which is right.
In 1977 two Harvard psychiatrists, Allan Hobson and Robert McCarley, offered a plainer account, the activation-synthesis theory. During REM sleep the brainstem sends out bursts of activity, while the senses are shut off from the world and the body from moving. The thinking parts of the brain then do what they always do with signals: they try to make sense of them, weaving them into a story from memories. The bizarre jumps of dreams fit this well. What it struggles with is that many dreams aren’t random at all: they are about the people and worries of the dreamer’s real life. That has led others to link dreaming with memory, and to the idea that dreams are partly the brain sorting through the day. None of these accounts has won.
Sleep can also go wrong in ways psychologists study. Insomnia is trouble falling or staying asleep. In sleepwalking, which comes out of deep sleep, a person gets up and does things with little awareness and little memory of it afterwards. Sleep paralysis is the REM paralysis arriving or lingering while the mind is awake: for up to a few minutes you can’t move, and you may see or feel things that aren’t there, which is frightening but harmless in itself. In REM sleep behavior disorder the paralysis fails and people act out their dreams. Narcolepsy brings overwhelming sleepiness by day, sometimes with sudden muscle weakness set off by a strong feeling, such as laughter or anger. In sleep apnea, breathing stops again and again in the night, so sleep is broken, often without the sleeper knowing. Doctors can tell these apart with an overnight sleep study, which records the same waves, eyes and muscles that Aserinsky’s machine did.
So sleep is not nothing. It is a busy, well-ordered state: light sleep, deep sleep and REM, repeating about every ninety minutes, with deep sleep early in the night and REM late. Two processes time it, a body clock set by light and a pressure that builds while you’re awake, and in teenagers the clock runs later, which school times don’t allow for. Too little sleep makes people worse at things in ways they hardly notice, though how much the long-term health links are caused by sleep is still open, and a famous book’s warnings about them were disputed, some corrected by its own author. Why we sleep at all, and why we dream, nobody can yet say for sure, and that may be the most surprising thing about something we spend a third of our lives doing.
Found a mistake? Please tell us.
