What the brain does while you sleep

A sleeping brain is not an idle one. Measured by energy use, it is nearly as expensive as a waking brain, and in REM it can be more active by some measures than quiet wakefulness. Whatever sleep is, it is not the brain being switched off. Three lines of research describe what it appears to be doing instead — and they differ sharply in how solid the evidence behind them is, which is worth saying up front.

The best-evidenced is replay. In a series of rodent experiments, Matthew Wilson and Bruce McNaughton recorded from hippocampal place cells — neurons that fire when an animal is at a particular spot — while rats ran a track, then recorded again while the rats slept. The same sequences of cells fired again during sleep, in the same order, compressed in time. The brain was, in a literal and measurable sense, running the day's route back through itself. Replay has since been observed in other tasks and species, and it is the most concrete mechanism anyone has for how sleep might turn experience into durable memory.

The second is downscaling. Giulio Tononi and Chiara Cirelli's synaptic homeostasis hypothesis proposes that waking drives a net increase in synaptic strength across the brain, which is metabolically expensive and cannot continue indefinitely, and that slow-wave sleep scales connections back down proportionally — weakening everything so that the strongest, most meaningful traces stand out more clearly against a quieter background. It is an elegant account and has experimental support, but it remains a hypothesis with active critics, and it sits in an unresolved relationship with replay: one describes strengthening what matters, the other describes weakening everything else.

The third is cleaning, and here the hedging has to be explicit. Maiken Nedergaard's group described what they named the glymphatic system: a route by which cerebrospinal fluid moves along channels around blood vessels through brain tissue, carrying metabolic waste products out. Their work reported that this flow increases substantially during sleep, along with an increase in the space between cells that would make such flow easier. If that picture holds, it gives sleep a plainly physical job — the brain flushing out the by-products of a day's operation.

The caution is important. The strongest glymphatic evidence comes from rodents, where the necessary imaging and manipulation are possible. Extending it to humans is an active research area rather than a settled result, some findings have been contested, and the popular summary — that sleep washes the brain and skipping it lets waste accumulate toward disease — runs well ahead of what has actually been demonstrated in people. It is a promising line of work, not an established human mechanism, and this site would rather say so than repeat a confident version.

These processes are not competing explanations of one thing. They operate on different timescales and are tied to different stages: replay is prominent in slow-wave sleep, downscaling is attributed to the same slow oscillations, and REM has its own distinctive chemistry — noradrenaline at its lowest point of the day — that other researchers link to emotional processing. A single night plausibly does all of it, in the layered structure described in this section's article on sleep cycles.

Where does dreaming fit? Less neatly than popular accounts suggest. It is tempting to read dreams as the subjective side of replay — the felt experience of memories being processed — and there is something to that, since dream content does draw on recent waking material. But dreams rarely reproduce events accurately, which is exactly what replay does; they recombine, distort and invent. Whatever the relationship is, it is not a simple screening of the day's footage.

The honest summary is that the sleeping brain is running maintenance work whose outlines are visible and whose details are genuinely unfinished. That is not a disappointing answer. It is a fair description of where an active field currently stands.

Frequently asked

What this is based on

• Wilson and McNaughton's rodent research on hippocampal replay during sleep
• Tononi and Cirelli's synaptic homeostasis hypothesis
• Maiken Nedergaard's group's research on the glymphatic system