The Sleep Duration Number Everyone Misreads: What "Four Hours Is Enough" Actually Measured

There is a story that circulates, usually delivered with a note of envy, about people who need only four hours of sleep and function beautifully. Margaret Thatcher is the patron saint of it. Tech founders invoke it. The implication is always the same: sleep duration is negotiable, and the disciplined among us can simply need less. The story leans on a real piece of science — a 2009 genetics paper — and it gets that science almost entirely wrong. So let's start with the actual number, because the number is where the confusion lives.

The number was 6.25, and it belonged to two people

In 2009, the geneticist Ying-Hui Fu and colleagues published a paper in Science describing a mother and daughter who slept far less than average and seemed no worse for it. The pair carried a point mutation in a gene called DEC2 (also written BHLHE41). Fu's team then engineered mice and fruit flies with the same mutation and found they, too, slept less.

Here is the part that gets sanded off in retelling. The two human short sleepers did not sleep four hours. They slept about 6.25 hours on average, compared with 8.06 hours for their unaffected relatives — a difference of roughly an hour and forty-five minutes. That's the real finding. Not "some people thrive on four hours." A rare genetic variant, in a handful of documented families, shaves under two hours off a normal night.

Fu's lab has since found other mutations — in ADRB1 (Shi et al., 2019, Neuron) and NPSR1 — associated with natural short sleep. Across all of them, the pattern holds: these people sleep somewhat less, around six hours, and the trait is heritable and vanishingly rare. Estimates put true genetic short sleepers at well under 1 percent of the population, possibly a small fraction of that. The odds that you are one, and simply haven't noticed, are not good.

What actually happens when you cut the hours

To see why four hours is a different animal from six, it helps to follow the body through a shortened night in the order it happens.

While you're awake, a molecule called adenosine accumulates in the brain. It's a byproduct of your neurons spending energy all day, and it builds what researchers call sleep pressure — the growing physical urge to sleep. Adenosine binds to receptors that quiet arousal-promoting circuits. This is exactly the machinery caffeine hijacks: caffeine is an adenosine-receptor antagonist, so it doesn't remove the pressure, it just blindfolds you to it. The adenosine is still there.

When you finally sleep, the early hours are dominated by slow-wave sleep — the deepest, most restorative stage, when adenosine gets cleared and the pressure resets. This is why the first block of sleep feels so structurally important: it's front-loaded with the deep stuff. Cut the night to four hours and you'll capture much of that early slow-wave sleep, which is part of why short nights don't feel catastrophic at first. You got the good structural sleep.

But the back half of the night is where REM sleep clusters, in longer and longer stretches toward morning. REM is disproportionately involved in emotional processing and memory consolidation. Amputate the night at four hours and you preferentially lose REM. You don't lose sleep evenly — you lose a specific, hard-to-replace kind.

There's also the plumbing. The glymphatic system, described by Nedergaard's group beginning around 2013, clears metabolic waste — including beta-amyloid — from the brain, and it runs most actively during sleep. The evidence here is mostly from mice, and the human translation is still being worked out, so file this under plausible but not fully nailed down. But the direction of the finding is consistent: sleep is not idle downtime. It's a maintenance shift, and a shortened shift leaves work undone.

What the number doesn't measure

The DEC2 story fails as life advice for a reason that has nothing to do with genetics. It measures the wrong thing for the question people are actually asking.

Most people who claim to thrive on little sleep are reporting how they feel, and feeling is a broken instrument here. The definitive work is Van Dongen et al. (2003), Sleep. Researchers restricted subjects to 4, 6, or 8 hours in bed for two weeks and tested them daily on a reaction-time task. The 6-hour group's performance declined steadily across the fortnight until, by the end, it matched people who'd been kept awake for 24 hours straight. The 4-hour group did worse.

The devastating detail: subjective sleepiness plateaued while objective performance kept sliding. People adapted to feeling tired. They did not adapt to being impaired. They rated themselves as only slightly sleepy while performing like the legally-drunk-equivalent. This is why "I do fine on five hours" is not evidence of anything except that the part of you keeping score has gone quiet.

So the four-hours myth commits two errors. It inflates a genetic six-hour finding into a universal four-hour permission slip. And it trusts self-report, the one measure the research specifically shows you can't trust.

How much sleep do adults actually need?

For most healthy adults, the answer is seven to nine hours per night. That's the range from the National Sleep Foundation's 2015 consensus panel (Hirshkowitz et al.) and echoed by the American Academy of Sleep Medicine, which recommends seven or more hours for adults on a regular basis. "Seven or more" is doing quiet work in that sentence: seven is a floor for typical adults, not a target to shave down to. A smaller share of people sit comfortably at the high or low edge of the range, and a rare few genuinely need less. But needing less is not the same as tolerating less, and almost everyone who thinks they're in the first camp is in the second.

Individual need shifts with age, illness, physical exertion, and recovery from prior deprivation. The range is a range for a reason. What the science does not support is the idea that you can train yourself down to four hours through discipline. You can get used to it. That's different.

An honest rule of thumb for tonight

Give yourself a sleep opportunity — time actually in bed, lights off — that's about thirty to sixty minutes longer than the sleep you want, because nobody sleeps 100 percent of their time in bed. If you want seven and a half hours of sleep, be in bed closer to eight and a quarter.

Then run one cheap test. On a stretch of days with no alarm, note when you naturally wake. If you're routinely sleeping an hour or two longer than your workweek, you've been carrying a sleep debt, and your weekday duration is a compromise, not your true need.

One caution on that last row: consistently long sleep can signal an underlying problem — depression, fragmented sleep, a disorder degrading quality so you need more quantity. More is not automatically better, and the U-shaped associations between sleep duration and mortality in large observational studies are real, if messy and confounded.

Where the science is still open

Here's what genuinely isn't settled. The DEC2 and ADRB1 short sleepers don't just sleep less — some evidence hints they may be protected against the usual costs of short sleep, sleeping efficiently rather than merely sleeping little. If that holds up, the interesting question isn't how they sleep less. It's whether their sleep is somehow more concentrated, doing eight hours of maintenance in six.

Which leaves the question nobody can answer yet: when a natural short sleeper thrives on six hours, is their brain finishing the same overnight work faster — or have they simply, like the rest of us, stopped being able to feel what's left undone?