Sleep
The function we still don't fully understand
Description
Humans spend roughly a third of their lives asleep. The average adult needs between seven and nine hours per night to function normally. Sustained sleep deprivation — a few days without adequate sleep — produces cognitive impairment comparable to alcohol intoxication. Weeks of partial deprivation degrade memory, mood, immune function, and metabolic regulation. Years of chronic insufficient sleep are associated with elevated risk of cardiovascular disease, type 2 diabetes, dementia, and early death. Sleep is not optional, not negotiable, and not something the body can catch up on easily once the debt accumulates. It is one of the few genuinely non-compressible demands in human biology.
And yet, despite roughly a century of serious scientific investigation, we still do not fully understand what sleep actually does. The neuroscience has advanced enormously. We know a lot about the architecture of sleep (REM versus non-REM, the stages within non-REM, the specific brain circuits involved), the hormones that regulate it (melatonin, cortisol), the physiological changes that accompany it (blood pressure falls, body temperature drops, growth hormone releases). What we don't have is a single, clean answer to the question a seven-year-old would ask: why do we sleep at all? Multiple theories exist, each explains some of the evidence, none explains all of it.
The gap between what we know and what we don't produces a specific cultural situation. Sleep is simultaneously the subject of scientific caution (researchers are honest about what remains unknown) and popular over-confidence (sleep-hacking content fills bookstores and podcasts with specific claims the research does not fully support). Matthew Walker's 2017 Why We Sleep became a bestseller, generated some controversy among sleep researchers over specific claims, and helped make sleep a prestige topic for high-achievers. Understanding what the science actually shows, where the popular coverage overreaches, and what the practical implications are remains useful — particularly because most people are not getting the sleep they need, and the consequences are substantial.
● The question we're asking: what does sleep actually do, what do we still not know, and what do the practical implications look like?
● What we'll see: the architecture of sleep, the competing theories, the science of sleep loss, and the practical stakes.
Table of contents
01The architecture
Sleep is not a uniform state. It cycles through multiple stages roughly every 90 minutes, and each stage has distinct neural signatures, physiological patterns, and probable functions. The main division is between non-REM sleep (which makes up roughly 75% of the night) and REM sleep (the remaining 25%). Non-REM has three stages of its own — N1, N2, and N3 — ranging from light drowsy sleep to deep slow-wave sleep where the brain's electrical activity synchronizes into large, slow waves. REM sleep is the one with vivid dreams, rapid eye movements under closed lids, and a specific pattern of muscle paralysis that prevents the sleeper from acting out their dreams.
The architecture changes across the night. Early in the night, the cycles emphasize deep non-REM sleep — the slow-wave sleep thought to be important for memory consolidation and physical restoration. Later in the night, as the cycles progress, REM periods lengthen, which is why most vivid dreams occur in the hours before waking. The specific sequence matters: waking mid-cycle (particularly during deep sleep) produces the grogginess called sleep inertia. Waking at the natural end of a cycle, which most people do without noticing, feels comparatively easy. This is one reason fragmented sleep is worse than shorter continuous sleep at the same total duration.
02The competing theories
The oldest theory is that sleep restores the body — tissue repair, waste clearance, energy replenishment. Evidence is clear for specific restorative processes. Growth hormone is released in pulses during deep non-REM. The brain's glymphatic system, discovered in the 2010s, actively flushes metabolic waste (including beta-amyloid associated with Alzheimer's) more during sleep than wakefulness. Immune function is calibrated during sleep. These restorative processes are real but do not fully explain why we need so much of this specific state rather than reduced activity while remaining alert.
The memory consolidation theory, dominant since the 1990s, argues that sleep is essential for processing and storing memories. The evidence is substantial. Subjects who sleep after learning perform better on subsequent recall tests than subjects kept awake. Specific types of memory are consolidated during specific sleep stages — procedural memory during REM, declarative memory during slow-wave sleep. The hippocampus appears to replay recent experiences during sleep, transferring them to longer-term storage in the cortex. The theory explains a lot of specific findings but struggles to explain why species with vastly different memory demands (jellyfish, for instance, which has no brain but still sleeps) still need sleep.
03The science of sleep loss
Sleep deprivation produces a specific and reliable sequence of effects. After 24 hours without sleep, cognitive performance on most tasks degrades to roughly the level of someone legally drunk. Reaction times slow, attention lapses, error rates rise. After 48 hours, hallucinations and perceptual distortions become common. Longer sustained deprivation produces specific physical symptoms and eventually catastrophic failure — Randy Gardner's eleven-day 1964 record remains the best-documented extended deprivation experiment, and he experienced substantial cognitive and perceptual disruption well before that point. Rats kept awake experimentally die within a few weeks, though the specific cause of death is not fully understood.
Chronic partial sleep restriction — the pattern most adults actually experience — produces cumulative effects that people systematically underestimate. Subjects restricted to six hours per night for two weeks perform on cognitive tests at the level of subjects kept awake for 48 hours straight, but the six-hours-per-night subjects consistently rate themselves as feeling fine. The subjective sense of having adapted to short sleep is a specific illusion; the objective performance shows no adaptation, just a lower baseline. This is one of the most important findings in modern sleep research and has specific implications for how people evaluate their own sleep needs.
04The practical stakes
The single most reliable finding in sleep research is that most adults in developed countries are not getting enough sleep, and most of them don't realize it. Estimates suggest 30-40% of American adults sleep less than seven hours per night regularly, and the share rose over the past several decades as work hours increased and smartphones arrived at the bedside. The productivity lost to sleep deprivation is substantial — estimated at hundreds of billions of dollars per year in the US alone — and the health consequences compound over decades. Fixing sleep at the population level would likely produce benefits comparable to or exceeding those of any single public health intervention available.
Individual sleep hygiene has accumulated substantial evidence. Regular sleep and wake times help synchronize the circadian system. Reduced evening light exposure (particularly blue light from screens) supports melatonin release. Cool, dark, quiet bedrooms produce better sleep architecture. Avoiding caffeine after noon reduces its interference with evening sleep. Avoiding alcohol in the three hours before bed preserves REM sleep, which alcohol disrupts. These recommendations are well-supported but are also easy to ignore — most people know them and follow them inconsistently, which is part of why sleep remains widely impaired even as the knowledge has spread.
05Conclusion
Sleep is one of the most important features of human biology and one of the least fully understood. What the science shows clearly is that it is non-negotiable, that most people are getting less than they need, and that the consequences of chronic insufficient sleep compound across domains — cognition, mood, metabolism, cardiovascular health, mortality. What it shows less clearly is exactly why sleep is so essential; multiple theories explain parts of the evidence, and the complete picture is probably a combination of several functions running simultaneously during a state that evolved to serve all of them at once.