Sleep is not a single state. It is a cycle through several distinct stages, each with different physiology, and the structure explains a great deal about why sleep advice takes the form it does.
The stages
Light sleep occupies the transition from waking and a substantial portion of the night.
Deep sleep, characterised by slow brain waves, is where the body is most difficult to rouse. It is concentrated in the first half of the night.
Rapid eye movement sleep, where most vivid dreaming occurs and where the body is temporarily paralysed, is concentrated in the second half and lengthens through the night.
A full cycle takes roughly ninety minutes, and the proportion of each stage shifts across the night rather than repeating identically.
Why the timing distribution matters
Because it means cutting sleep short removes disproportionately from one stage.
Going to bed late but waking at the usual time removes deep sleep. Waking early after a normal bedtime removes rapid eye movement sleep.
Which means two people both losing two hours can lose quite different things, and the effects differ.
What each stage appears to do
Stated carefully, because this is an area where confident claims outrun the evidence.
Deep sleep is associated with physical recovery, growth hormone release and, in more recent research, with clearance of metabolic waste from the brain.
Rapid eye movement sleep is associated with emotional processing and with certain kinds of memory consolidation.
Both appear involved in memory, in different ways, and the research on consolidation is active and not settled.
The two-process model
The framework that explains most sleep timing effects.
Sleep pressure builds during waking, driven by accumulating adenosine, and dissipates during sleep. The longer you are awake, the stronger the drive.
The circadian process is an internal clock producing alertness signals on a roughly twenty-four hour cycle, entrained primarily by light.
Sleep occurs readily when high pressure coincides with the circadian low, and poorly when they are misaligned.
Which explains jet lag, shift work difficulty, and why a nap at the wrong time makes evening sleep harder.
Caffeine
Works by blocking adenosine receptors, which means it masks sleep pressure rather than removing it.
Its half-life is around five hours in most adults, with wide individual variation driven partly by genetics.
Which means an afternoon coffee leaves a meaningful amount in the system at bedtime, and it can affect sleep structure even in people who fall asleep normally.
Alcohol
Reduces the time to fall asleep and degrades sleep quality substantially.
It suppresses rapid eye movement sleep early in the night, followed by a rebound later, which produces the fragmented second half that people recognise.
Which is why it feels like it helps and measurably does not.
Light
The primary signal setting the circadian clock, and the effect is dominated by morning light more than by evening light.
Which means getting outside early has a larger effect on sleep timing than avoiding screens at night, though both operate.
The blue light question specifically has weaker evidence than the marketing around filters suggests, and total light exposure and what you are doing with the device both matter.
Insomnia
Persistent difficulty is a condition with an effective evidence-based treatment, and it is not sleep hygiene advice.
Cognitive behavioural therapy for insomnia has substantial evidence, is recommended as first-line treatment in clinical guidelines ahead of medication, and is available in various delivered forms.
Which is worth knowing because most people with chronic insomnia are never told it exists.
Anyone with persistent sleep problems should raise it with a doctor, since several treatable conditions present as poor sleep, and sleep apnoea in particular is commonly missed.
Sleep trackers
Worth assessing since so many people rely on them.
Consumer devices infer sleep stages from movement and heart rate rather than from brain activity, which is what clinical sleep staging measures.
Validation studies find reasonable accuracy for total sleep time and considerably weaker accuracy for stage classification.
Which means the stage breakdown displayed each morning should be treated as an approximation rather than a measurement.
A recognised phenomenon has emerged where anxiety about tracker readings itself worsens sleep, and clinicians have begun advising some patients to stop tracking.
Naps
The evidence is more specific than general advice suggests.
Short naps of around twenty minutes improve alertness without producing the grogginess that follows waking from deep sleep.
Longer naps risk entering deep sleep, and waking from it produces sleep inertia lasting up to half an hour.
Naps also reduce sleep pressure, which makes falling asleep at night harder, so late-afternoon naps are counterproductive for anyone with sleep difficulty.
The variation between people
Sleep need varies genuinely, and the population distribution is narrower than people claim about themselves.
A small number of people carry genetic variants associated with short sleep without apparent cost, and they are rare enough that most who believe they qualify do not.
Chronotype — whether someone is naturally early or late — is also partly genetic and shifts across the lifespan, peaking latest in adolescence.
Which is the basis for arguments about school start times, and the evidence there is reasonably strong.