The Sleeping Brain

Sleep is one of the most common things people raise with me, and the request is nearly always the same: something to fix it. A tablet, ideally.

It is a reasonable thing to want. Poor sleep is miserable, and the misery is the part you notice.

What is easier to miss is that sleep is not simply rest. It is when the brain does a set of jobs it cannot do at any other time, and the consequences of not doing them run a lot further than feeling tired the next day.

Sleep is when memory is made permanent

The clearest link between sleep and cognition is not about waste clearance or dementia risk. It is about how memory works.

Experience is first encoded in the hippocampus, which acts as something like a temporary buffer. During deep sleep, those newly formed traces are replayed in bursts and gradually transferred into the neocortex for long-term storage. The process is coordinated by three rhythms locking together — the slow oscillations of deep sleep, thalamic sleep spindles, and hippocampal sharp-wave ripples. When that coordination is disrupted, less of the day’s learning is retained.

The effect runs in the other direction too. A night of poor sleep reduces the hippocampus’s capacity to encode new information the following day, so the deficit is both retrospective and prospective. Anyone who has tried to learn something after a bad night has felt this.

REM sleep appears to do different work again — consolidating procedural skills and processing emotional memory.

So the most immediate cognitive cost of poor sleep is not a theoretical risk decades away. It is that you learn less, retain less, and think less clearly now.

The overnight clean

The longer-term argument rests on clearance.

The brain is metabolically the most active organ in the body, generates waste accordingly, and has no conventional lymphatic drainage. The glymphatic hypothesis, proposed just over a decade ago, holds that channels running alongside the brain’s blood vessels carry cerebrospinal fluid through the tissue and flush waste out — including amyloid-beta and tau, the proteins that accumulate in Alzheimer’s disease. The system was reported to work far better during deep sleep than during waking.

It is an elegant idea and it has been widely repeated, including by me. The mechanism itself is still contested, with recent work disagreeing over whether clearance rises or falls during sleep, and almost all of that work is in mice.

What is better established is the human evidence sitting alongside it. A single night of sleep deprivation measurably increases amyloid-beta in the brain on PET imaging. Experimental disruption of deep sleep for one night raises amyloid in cerebrospinal fluid the next morning. Cognitively unimpaired older adults with less slow-wave activity accumulate amyloid faster over subsequent years. And loss of slow-wave sleep with age is associated with a higher risk of later dementia.

One encouraging finding: among people with similar amounts of amyloid already deposited, those getting more deep sleep perform better on memory testing. Deep sleep appears to offer some resilience against pathology that is already there.

One caveat. The relationship is bidirectional — Alzheimer’s pathology disrupts the brain structures that generate sleep, so disturbed sleep is also a feature of the long preclinical phase. When a study finds poor sleepers develop more dementia, some of that is the disease announcing itself.

Sleep apnoea

If there is one thing here to act on, it is obstructive sleep apnoea.

It is common, substantially underdiagnosed, and associated with both cognitive impairment and faster amyloid accumulation. It fragments sleep, strips out slow-wave sleep specifically, and produces repeated overnight falls in oxygen saturation. It also raises blood pressure, which brings the vascular mechanisms from an earlier article into play.

Unlike most of what this series covers, it has a defined test and a defined treatment. Snoring, witnessed pauses in breathing, waking unrefreshed and daytime sleepiness are the classic features, though significant apnoea can exist without dramatic symptoms. A collar size above about 43 centimetres in men or 40 in women, a raised BMI and hypertension all raise the probability.

If any of that is familiar, a sleep study is a reasonable request.

Treating insomnia

This brings me back to the tablet.

For chronic insomnia, the first-line treatment is not medication. It is cognitive behavioural therapy for insomnia, and the evidence for it is strong — better than hypnotics over any time frame beyond a few weeks, and without the dependence, tolerance and next-day impairment.

It is more demanding than a prescription. It involves restricting time in bed to consolidate sleep, using the bed only for sleep, getting up when not sleeping, and working on the anxious thinking that keeps people awake. The sleep restriction component feels counterintuitive and is unpleasant for the first fortnight. It also works.

It does not require a psychologist. Several digital programmes deliver it effectively, which matters in a country where access to psychology is uneven and expensive.

Benzodiazepines in particular produce sedation while suppressing slow-wave sleep, so someone taking them may be unconscious for eight hours while getting less of the stage that does the work. They also carry their own association with dementia risk in observational data — confounded, but not obviously wrong. I have written elsewhere on the site about sleep architecture in more detail.

How much

The relationship between sleep duration and dementia risk is U-shaped. Both short and long sleep are associated with higher risk, though the long-sleep association is probably largely reverse causation.

Seven to eight hours is where the evidence is most reassuring. What matters as much as the total is continuity, because deep sleep is concentrated in the first part of the night and is the first thing lost to fragmentation.

Practical steps for this week

1. Protect the first half of the night. Deep sleep is front-loaded, so a consistent bedtime does more work than a consistent wake time. Going to bed at a similar hour, seven nights a week, is the highest-yield change for most people.
2. Get assessed for sleep apnoea if there is any suggestion of it. Snoring, witnessed pauses, waking unrefreshed, or a partner who has moved to another room. It is treatable and the treatment works.
3. If you have chronic insomnia, start with CBT-I rather than a tablet. Ask your GP, or look into one of the digital programmes. It is harder than a prescription and it is the treatment that lasts.
4. Move alcohol earlier, or preferably reduce it. Alcohol reliably suppresses slow-wave sleep in the first half of the night, precisely when it is most needed. A drink at six does far less damage than one at ten.
5. If you are on long-term sleeping tablets, raise it at your next appointment. Not to stop abruptly, which is unwise, but to discuss whether what you are getting is genuinely sleep.

References

1. Girardeau G, Lopes-Dos-Santos V. Brain neural patterns and the memory function of sleep. Science. 2021.
2. Sharma RA et al. Obstructive sleep apnea, cognition and Alzheimer’s disease: a systematic review. Sleep Medicine Reviews. 2020.
3. Skoglund H et al. Digital cognitive behavioural therapy for insomnia: a large-scale randomised controlled trial. Sleep Medicine. 2024.