Sleep

Why would an animal spend a third of its life unable to notice a predator?

Consider what an animal gives up by sleeping. It stops foraging, stops watching for danger, stops defending territory, and lies immobile and largely insensible for hours — in the open, where things eat animals. Evolution is ruthless about waste; a trait that costs this much and delivers nothing should have been trimmed away in some lineage somewhere. Yet no such lineage has been found. Every animal examined with a nervous system, from mammals down to fruit flies and even jellyfish, shows some sleep-like state. What is being bought that is worth that price?

The logic of the question constrains the answer before we know it. If sleep were merely useful, some species under heavy predation would have abandoned it. That none has suggests the benefit is not an optional extra but something the nervous system cannot do while awake — a genuine incompatibility, not an inconvenience.

Look for evidence of that incompatibility and it turns up in the compromises. Dolphins and some birds sleep one cerebral hemisphere at a time, keeping one eye open and continuing to swim or fly. The pressure to stay alert is real enough to have produced an elaborate workaround — and yet the need itself was not abolished, only halved and taken in turns. That is what you would expect if sleep were doing something obligatory to brain tissue.

So what could that be? Two accounts lead the field, and I want to be careful to present them as leading candidates rather than as the answer.

The first is clearance. The brain has no lymphatic drainage of the ordinary kind, and work published in 2013 reported that in sleeping mice the interstitial space between cells expands substantially, letting cerebrospinal fluid flush through brain tissue and carry away metabolic waste, including amyloid-beta, far faster than during waking. That is the glymphatic account, and if it holds it gives sleep exactly the character the puzzle demands: a housekeeping process incompatible with the packed, active tissue of a working brain. The caveats are serious, though. The work is largely rodent, done under anaesthesia in places, the human evidence is indirect, and at least one later study reported the opposite direction of effect. It is an attractive hypothesis under active dispute, not an established mechanism.

The second is memory consolidation. During slow-wave sleep, hippocampal cell sequences that fired during the day's experience replay in compressed form, and experiments that cue particular memories during sleep — by replaying an associated sound or smell — can measurably improve later recall of those specific items. Something is being reprocessed. A related proposal is that sleep globally weakens synapses strengthened during waking, restoring the capacity to learn tomorrow. Here too, the caveats matter: much of the evidence is correlational, sleep deprivation experiments confound the loss of sleep with stress and impaired attention, and it is unclear how much of the consolidating work strictly requires sleep rather than merely quiet time free of new input.

Ask which is right and you may have the wrong question. Sleep is not one state — it cycles through stages with different electrical signatures and different physiology — and there is no reason a process this old and this universal should serve a single purpose. Energy conservation, immune function, and metabolic and hormonal regulation all have supporting evidence too. The honest position is that sleep is clearly necessary, that several plausible functions have real support, and that no single one has been established as the reason.

What is not in doubt is the necessity. Rats prevented from sleeping die within weeks. In humans, fatal familial insomnia — a rare prion disease that progressively destroys the ability to sleep — is uniformly fatal. Whatever the function turns out to be, it is not optional.

Two clarifications, both aimed at the same misconception.

The first is that sleep is not the absence of activity. Large parts of the brain are as active asleep as awake, and some circuits more so; what changes is the pattern — coordinated slow oscillations rather than the fragmented activity of waking. Sleep is a different mode of operation, not an off switch, which is also why it saves so little energy relative to quiet waking, and why the pure energy-conservation account has never been sufficient on its own.

The second is that "insensible" overstates it. Sleep is reversible by sufficient stimulus — the feature that distinguishes it from coma or anaesthesia — and the sleeping brain retains some sensitivity to salient signals, which is why a parent wakes at their own child's cry. Evolution did not accept blindness; it accepted a raised threshold, and hedged further with burrows, roosts, group watches and hemispheric sleep. That softens the tradeoff without dissolving it: the cost is still large enough that the benefit must be substantial.

• Why REM sleep and slow-wave sleep have such different physiology, and whether they serve different purposes.
• How circadian timing and sleep pressure interact to decide when sleep happens rather than why it exists.

Every term the collection defines is gathered in the glossary.