For most of history, sleep was treated as the brain switching off — a passive state where not much happened. We now know the opposite is true. Some of the brain's most important maintenance work happens precisely when you're unconscious. And one of the most intriguing discoveries of the last 15 years is a cleaning system that may run primarily while you sleep.
The discovery of the brain's cleaning system
In 2012, a team led by neuroscientist Maiken Nedergaard at the University of Rochester described a previously unknown network in the brain. They called it the glymphatic system — a play on the lymphatic system (the body's waste-clearance network) and glial cells (the brain's support cells that help run it).
The basic idea: throughout the day, your brain's intense metabolic activity generates waste — leftover proteins and byproducts of cellular work. The body has the lymphatic system to clear waste elsewhere, but the brain, sealed behind the blood-brain barrier, needed its own solution. The glymphatic system is that solution. It uses cerebrospinal fluid (CSF) to flush through brain tissue, picking up waste and carrying it away to be cleared from the body.
The waste it clears includes two proteins of particular interest: beta-amyloid and tau. These are the same proteins that, when they accumulate and clump together, form the plaques and tangles that are hallmarks of Alzheimer's disease.
Why sleep specifically?
The most striking part of the discovery was the timing. In the original animal studies, glymphatic clearance was dramatically more active during sleep than during wakefulness. The proposed explanation: during deep sleep, brain cells appear to shrink slightly, widening the spaces between them and allowing cerebrospinal fluid to flow through more freely — like opening channels for the cleaning fluid to reach more of the tissue.
This clearance seems to be concentrated in slow-wave sleep, the deepest stage of non-REM sleep. It's the stage associated with the large, slow brain waves that dominate the first few hours of the night. Slow-wave sleep also declines naturally with age — which may be part of why dementia risk rises as we get older and our deep sleep diminishes.
There's an important wrinkle here: the relationship runs in both directions. Poor sleep may contribute to amyloid buildup, but amyloid buildup also disrupts sleep. This bidirectional loop — where each problem worsens the other — is one reason sleep is taken seriously in dementia research. Breaking the cycle at the sleep end is something you can potentially influence.
What the human evidence shows
Most of the early glymphatic research was done in mice. The natural question was whether the same thing happens in people. Studying it in humans is hard — you can't easily watch fluid move through a living brain — but the evidence has been building.
A 2026 study published in Nature Communications offered some of the most direct human evidence yet. In a randomised crossover trial, 39 cognitively healthy adults were monitored over nights of normal sleep versus nights of sleep deprivation. After normal sleep, participants had higher morning blood levels of beta-amyloid and tau than after sleep deprivation — consistent with the idea that during normal sleep, the brain was clearing these proteins out into the bloodstream.
An honest note: the science is still evolving
It would be easy to present all of this as settled fact. It isn't, and you should be wary of any source that does. The glymphatic system is a genuinely active area of scientific debate.
In particular, a study published around 2024 challenged the core claim, suggesting that brain clearance might actually be more efficient during wakefulness than during sleep — the reverse of the original finding. This sparked an ongoing debate among sleep and neurology researchers, including a featured debate at the 2025 SLEEP annual meeting. The mechanisms, the timing, and even the basic direction of the effect are still being worked out.
What good sleep actually looks like
If sleep matters for your brain — and the weight of evidence says it does — the practical question is what to actually do. The research points less to a magic number of hours and more to consistency and quality. Most adults need 7–9 hours, but how you sleep matters as much as how long.
Sleep in the bigger picture
A note on how sleep fits the broader framework: sleep was not included among the 14 modifiable risk factors in the 2024 Lancet Commission report. That doesn't mean sleep is unimportant — it reflects that the Commission's list focuses on factors with the strongest causal evidence at population scale, and sleep's role is still being established. The association between poor sleep and dementia is well documented; the precise causal contribution is what's still being quantified.
Sleep also interacts with nearly every other brain health domain. Poor sleep worsens mood (linked to depression, a Lancet factor), reduces next-day physical activity, impairs food choices, and raises blood pressure. Improving sleep often improves several domains at once. The FINGER trial's multi-domain approach reflects this interconnection — the domains reinforce each other.