Why do we sleep?

The question

We spend about a third of our lives asleep. For hours we don't eat, don't look after our young and barely notice danger. For an animal in the wild, that is an enormous risk, and yet evolution has not got rid of sleep. Fish, fruit flies, worms and even jellyfish, which have no brain at all, go through sleep-like states.

Anything that costly and that universal must do something essential. The question is what. We know very well what happens when we don't sleep; we know much less about what sleep itself does.

What we know

Sleep is regulated like hunger: the longer we stay awake, the stronger the pressure to sleep, and after deprivation we sleep longer and deeper. A separate internal clock sets when during the day this happens.
Total sleep deprivation is lethal. Rats kept awake in the classic experiments of the 1980s died within two to three weeks. In fruit flies, death after sleep loss has been linked to the build-up of harmful oxidising molecules in the gut, and was prevented by antioxidants — an unexpected sign that sleep matters for the whole body, not only the brain.
Sleep helps memory. After sleep, people remember learned material better than after the same time awake, and during deep sleep the brain seems to replay recently learned patterns.
Some animals have found ways around full sleep but not ways out of it: dolphins and some birds sleep with one half of the brain at a time.
Even jellyfish, with scattered nerve nets and no brain, have a sleep-like state with rebound after deprivation. Sleep appears to be as old as nervous systems.

Main hypotheses

Resetting the synapses

While awake, learning strengthens connections between neurons. That costs energy and space, and it can't go on forever. According to the synaptic homeostasis hypothesis, sleep scales connections back down, keeping the important ones and letting the noise fade.

Consolidating memory

In deep sleep the brain may transfer fresh memories from the hippocampus to the cortex for long-term storage. This fits the evidence for sleep's role in memory, but doesn't explain why animals with very simple nervous systems also sleep.

Cleaning the brain

In 2013 a study in mice reported that during sleep the space between brain cells widens and fluid washes out waste products, including proteins linked to Alzheimer's disease. The idea became very popular, but in 2024 another mouse study found the opposite: clearance slowed during sleep and anaesthesia. The dispute is not yet settled.

Adaptive inactivity

Perhaps sleep is not primarily a repair process but a way of being still when activity is useless or dangerous, and of saving energy. Repair and memory functions would then be added later to time that was already "free".

Where understanding ends

We have many functions of sleep but no core one. Each hypothesis explains part of the evidence, and none explains why sleep in almost all animals requires switching off awareness of the outside world. Many of these tasks — cleaning, saving energy, strengthening memory — could in principle be done while awake and alert, as some organs do all the time.

We also don't know what exactly accumulates during waking and is removed during sleep. Something builds up the pressure to sleep, and the molecule adenosine plays a part, but the underlying debt that sleep "pays off" hasn't been identified.

Finally, it isn't clear whether sleep is one thing. The deep sleep of mammals, their REM sleep with its dreams, and the rest of a jellyfish or a worm may share a name without sharing a purpose. It is possible that sleep began with one function and later collected others.

What would settle it

Identifying the physical "debt" that builds up while awake and is paid off during sleep, and showing that removing it without sleep removes the need to sleep.
Comparing sleep across many species, from jellyfish to humans, to find what is common to all of them.
A healthy animal that doesn't sleep, created by switching off sleep circuits — or proof that such an animal can't exist.