Appearance
2.8 — Sleep, Dreams and the Mind
Keep a rat permanently awake and it dies in about two to three weeks — faster than it would die of starvation. Nobody has fully explained the cause of death.
You will not be doing that. But the milder version is running in millions of people, and its effects on the mind are larger and arrive faster than almost anyone assumes. After roughly seventeen to nineteen hours awake, performance on attention and reaction-time tasks falls to a level comparable to being legally drunk in many countries. After two weeks of six hours a night, people perform about as badly as those who have gone a full night without sleep — and, critically, they rate their own alertness as only slightly reduced.
That last part is the dangerous one. The instrument that would tell you that you are impaired is itself impaired.
Volume V covers sleep physiology. This page is what sleep does to the mind, and why almost every technique in the rest of this volume degrades without it.
The architecture of a night
Sleep is not one state. You cycle through stages roughly every ninety minutes, four to six times a night, and the composition of the cycles changes as the night goes on.
Light non-REM (stages 1 and 2) — the majority of the night. Stage 2 contains brief bursts of activity called sleep spindles, which are associated with consolidating what you learned during the day.
Deep non-REM (slow-wave sleep) — large slow brain waves, very hard to wake from. Concentrated in the first half of the night. This is when growth hormone is released, when clearance of metabolic waste from around brain cells appears to increase substantially, and when facts and events are transferred from the hippocampus towards longer-term cortical storage.
REM sleep — rapid eye movements, brain activity resembling waking, and near-total paralysis of the voluntary muscles. Concentrated in the second half of the night, with REM periods lengthening towards morning. Most vivid dreaming happens here, along with emotional processing and consolidation of skills and patterns.
The practical consequence of that asymmetry is important and rarely stated. Sleeping from 1 am to 7 am instead of 11 pm to 7 am does not simply remove a quarter of your sleep — it removes a disproportionate share of your deep sleep, and with it the memory consolidation and physical restoration that happen there. Waking at 4 am for an early flight removes REM instead, which is why that particular deprivation leaves you emotionally raw rather than merely tired.
What one bad night does to your mind
Not "makes you tired". Specific, measurable, and directly relevant to everything in Parts 4 and 5.
Prefrontal function degrades first. From 2.1, this is impulse control, planning, holding a goal, and stopping yourself from saying the thing. It is the most metabolically demanding region and the most sleep-sensitive.
Emotional reactivity goes up sharply. In one much-cited study, sleep-deprived participants showed markedly stronger amygdala responses to unpleasant images, together with weaker connectivity between the amygdala and the prefrontal regions that normally regulate it. Bigger alarm, weaker brake. That is a mechanical description of a bad day, and it means that after poor sleep you are not imagining that everything is worse — your threat detector is running hot with less supervision.
The negativity bias increases. Tired people recall negative material better than positive, read neutral faces as more hostile, and interpret ambiguous remarks less charitably. This alone is worth knowing before any difficult conversation.
Attention becomes intermittent. Rather than uniform slowing, sleep-deprived attention shows brief lapses — hundreds of milliseconds where nothing is processed. You cannot feel them. They are what kills people on motorways.
Learning is impaired at both ends. A sleep-deprived brain encodes new material substantially worse, and then the following night's sleep is what would have consolidated it.
And judgement of your own state is impaired. Self-rated sleepiness plateaus while objective performance keeps falling, which is the mechanism that lets chronic short sleep become normal. You will not notice. You will simply become a slightly worse-tempered, slightly less capable version of yourself and conclude that is who you are.
How much, honestly
Adults need roughly seven to nine hours, and the range is real — some people genuinely do well on seven, some need nine.
The exception people claim is much rarer than the number of people claiming it. A small number of individuals carry gene variants associated with naturally short sleep and appear to function normally on around six hours. The prevalence is very low. Almost everyone who believes they are one of them is instead someone whose self-assessment has adapted while their performance has not, which is exactly what the studies above predict.
On catching up. A weekend lie-in recovers some deficit, particularly the deep-sleep component, and does not fully restore attention and metabolic measures after a week of restriction. Sleep is not a bank account with a flexible overdraft; some of the debt is not repayable.
On naps. Genuinely effective and underused. Ten to twenty minutes gives alertness without grogginess. Sixty to ninety minutes gives a full cycle including REM and helps consolidation, but a nap that ends in deep sleep produces twenty minutes of feeling terrible. Avoid after about 3 pm, since the pressure to sleep at night is built up by hours awake and a late nap spends it.
What dreams are, as far as anyone can tell
Everybody dreams, several times a night, in every sleep stage though most vividly in REM. Most is forgotten within minutes of waking because the consolidation machinery is largely offline.
Freud's version — dreams as disguised wish fulfilment requiring interpretation — has no supporting evidence and is not taken seriously in research. Symbol dictionaries are worthless.
Three accounts with some support, not mutually exclusive.
Consolidation and integration. Sleeping brains replay the day's activity, and people frequently dream about tasks they have been practising. Learning a spatial task and then dreaming about it predicts better performance the next day.
Emotional processing. REM is the only state in which the brain is highly active while noradrenaline — the alerting, stress-associated chemical — is essentially absent. One influential proposal is that this lets the brain revisit emotional material without the accompanying physiological alarm, gradually stripping the charge from the memory while keeping its content. It is not proven, and it fits several observations: emotional events dominate dream content, REM deprivation worsens emotional regulation, and in PTSD this process appears to fail, with the same nightmare recurring without ever losing its force (14.4).
Simulation and rehearsal. Dream content is heavily weighted towards social scenarios and towards threats — being chased, losing something, being unprepared, conflict with people you know. One proposal is that this is safe rehearsal of situations that mattered.
What to make of your own dreams. They are not messages, and no fixed symbol means anything in particular. What is genuinely informative is the recurring theme, especially if it changes, because that tracks what you are currently processing. Someone who dreams repeatedly about examinations during a difficult period at work is not receiving a coded instruction; they are watching their own emotional processing use an old template for the same feeling.
Where sleep and mental health meet
The relationship runs both ways, and this is one of the most useful things in this Part.
Insomnia is a strong risk factor for later depression, not only a symptom of it. Treating the sleep improves the depression, and in several trials treating insomnia reduced depressive symptoms substantially even when depression was not the target.
Sleep loss triggers episodes in bipolar disorder, reliably enough that protecting sleep is a central part of managing it (14.5).
Anxiety and sleep form a loop: worry delays sleep, poor sleep raises next-day threat sensitivity, which produces more worry.
And the treatment that works is not a pill. Cognitive behavioural therapy for insomnia — CBT-I — outperforms sleeping medication over the medium and long term in trials and is recommended as first-line treatment by major clinical guidelines. Sleeping tablets work for a few weeks and then stop working, while tolerance and dependence build. This is one of the clearest cases in this volume where the psychological treatment beats the drug, and most people have never heard of it.
The core of CBT-I is counter-intuitive enough to be worth stating, because people do the opposite by instinct: spend less time in bed, not more. Lying awake for hours teaches your brain that the bed is a place for lying awake. Restricting time in bed to roughly your actual sleep time concentrates the sleep, rebuilds the association, and the window is then widened gradually. It is uncomfortable for about a week and it works.
What to do with this
Fix the timing before anything else. A consistent wake time, seven days a week, does more than any other single change, because your body clock is set mainly by when light hits your eyes after waking. Going to bed at a consistent time is much less powerful than getting up at one.
Get outdoor light in the first hour awake, ideally ten to twenty minutes. Outdoor light is many times brighter than indoor lighting even on a cloudy day, and it is the main signal that sets the clock that will make you sleepy roughly sixteen hours later.
Protect the last hour. Dim light, no work, no arguments, nothing that raises alertness. Screens matter less for their blue light than for what is on them — an argument in a message thread at 11 pm will cost you an hour whatever the colour temperature.
Caffeine has a half-life of around five to six hours, which means an afternoon coffee still has a quarter of its dose active at midnight. It reduces deep sleep even when you fall asleep fine. A cut-off eight to ten hours before bed is a large intervention that costs nothing.
Alcohol is the great deceiver here. It shortens the time to fall asleep, and then suppresses REM, fragments the second half of the night, and produces exactly the emotional rawness described above. A nightcap is a trade of tonight's onset for tomorrow's temper.
If you cannot sleep, get up. Twenty minutes awake in bed and you leave the room, do something dull in dim light, and return when sleepy. This feels like giving up and it is the single most effective behavioural instruction in CBT-I.
And build the diagnostic habit. Before you conclude that you are angry at your partner, unsuited to your job, or a fundamentally impatient person, check the last three nights. The number of psychological problems that turn out to be sleep problems wearing a costume is genuinely large, and this check costs you nothing and takes five seconds.
Next: 2.9 closes Part 2 with the thing that runs almost continuously while you are awake — the voice in your head, where it came from, and how to change what it says.