Sleep cycles explained, in plain language

A night of sleep is not one long state. It is a sequence of repeating sleep cycles, each one running through lighter sleep, deep sleep and REM before starting over, and each one shaped a little differently from the last. Almost everything that feels strange about sleep, why the same seven hours lands so differently on two mornings, why 3am awakenings feel so specific, why you can dream vividly right before an alarm, comes out of that structure.

What one sleep cycle actually contains

Sleep splits into two families. NREM sleep is everything that is not REM, and it is conventionally divided into three stages. N1 is the shallow drift right after sleep onset, the stage you can be woken from and swear you were never asleep. N2 is the workhorse of the night, the stage you spend more time in than any other. N3 is deep sleep, also called slow-wave sleep, the heavy, hard to rouse stage where the brain’s electrical activity slows into large rolling waves.

Then there is REM sleep, where most vivid dreaming happens, the eyes move under the lids, and the body is largely paralysed so you do not act out what you are dreaming.

A cycle runs roughly N1 into N2 into N3 and back up through N2 into REM, then repeats. During N2 the brain produces short bursts of rhythmic activity called sleep spindles, which are one of the signatures scorers look for and are associated with memory consolidation and with holding sleep steady against noise.

Why your sleep cycles are not identical

If you drew every cycle the same, you would be drawing the wrong night. The mix inside each cycle changes as the night goes on, and the pattern is consistent enough to be predictable.

Early cycles are heavy on deep sleep. Sleep pressure is at its peak when you fall asleep, and your body spends it on N3 first. By the third or fourth cycle there may be very little deep sleep left in the mix at all.

Late cycles are heavy on REM. REM periods start short and get longer as the night goes on, which is why the dream you remember is almost always the one that got interrupted by your alarm. It is also why cutting a night short does not remove sleep evenly. Lose the last 90 minutes and you lose disproportionately REM, not deep sleep.

This overall shape, the arrangement of stages across the whole night, is what people mean by sleep architecture. It changes with age, with alcohol, with a short night, with illness, and with what time you went to bed relative to your body clock.

The 90 minute number is an average, not a timetable

You have seen the advice: sleep in multiples of 90 minutes, set your alarm for 6 hours or 7.5, wake between cycles. The arithmetic is tidy and the biology is not.

Cycles are commonly quoted at about 90 minutes because that is roughly the average. Real ones run shorter and longer, often somewhere in the region of 70 to 120 minutes, they vary between people, and they vary within one night for the same person. Your first cycle is usually not the same length as your fourth. Add the time it takes you to fall asleep, which itself varies, and any countdown you do at 23:00 has accumulated enough error by 06:00 to land you anywhere.

So the underlying idea is right, waking from a lighter stage is genuinely easier, but the method is wrong. You cannot schedule your way to it in advance. What works is watching the last stretch of the night and acting on what is actually happening, which is the whole reason smart alarms exist.

What a hypnogram shows you

A hypnogram is the chart of stages over time, drawn as a stepped line falling into deep sleep and climbing back up into REM. Once you can read one, a night stops being a number and becomes a shape.

Three things are worth looking for. First, the depth and timing of the early descent, because deep sleep clustered in the first couple of hours is the normal, healthy pattern. Second, whether REM blocks lengthen towards morning. Third, the gaps. Brief awakenings show up as spikes to the top of the chart, and the total of the time spent awake after first falling asleep is WASO, wake after sleep onset.

Some of those spikes are entirely normal. Brief awakenings between cycles are a standard part of a healthy night, and most of them are never remembered. A nocturnal awakening only becomes a problem when you come fully around and then cannot get back to sleep, which is a different pattern with different causes. Why you wake up at 3am, and what it means goes through those.

What your phone can see, and what it cannot

This is the part most sleep apps are vague about, so here is the honest version.

The gold standard for staging is polysomnography: a lab study recording brain activity, eye movement and muscle tone, scored in 30 second windows. That is the only method that directly sees the difference between N2 and N3, because the difference is defined in terms of brain waves.

Everything else infers. Actigraphy uses movement to work out when you were asleep and when you were awake, which it does well, and it is the basis of most consumer tracking. Wearables improve on it by adding heart rate and heart rate variability from your wrist, which gives them a genuine, if imperfect, signal about stages.

A phone on the nightstand has movement and sound and nothing on your body. That is enough for good sleep and wake detection, and enough for a reasonable guess at stages, and it is not measurement. Tock is explicit about this: on a phone only night, the stages you see are estimates. Measured stages need Apple Health or a wearable, and Tock reads a Watch’s data through Apple Health rather than running on the Watch itself. What estimated sleep stages really mean is the longer explanation, and sleep tracking without a wearable covers what a phone genuinely does well.

What a phone is unusually good at is the audio side. Tock classifies snores, coughs and sleep talking on device with Apple SoundAnalysis, and saves the clips, so in the morning you can hear the thing rather than read a chart about it. No wearable does that.

What waking mid cycle actually does

The reason any of this matters practically is sleep inertia, the slowed, foggy state right after waking. The general finding is consistent: inertia is worst when you are woken out of slow-wave sleep, which for most people means the first half of the night, or any night short enough that deep sleep is still being made up when the alarm goes.

That is the whole mechanism behind the mornings where you cannot form a sentence for a quarter of an hour. Not weakness, not laziness, just an alarm that happened to land in the wrong stage. It usually clears within 15 to 30 minutes, and what sleep inertia is, and how to shorten it covers what actually shortens it.

The corollary is more encouraging. If the alarm catches you in N1 or N2, or in the brief awakening between cycles, you are functional almost immediately, on the same total amount of sleep. Same night, same hours, entirely different morning.

What this changes tonight

Not much, and that is the point. Sleep cycles are a thing to understand rather than a thing to manage. You cannot make yourself have more deep sleep by deciding to, and the fastest route to a healthy architecture is unglamorous: enough hours, at a consistent time, in a dark and quiet room.

Two things do follow from the structure, though. First, going to bed earlier protects deep sleep and waking later protects REM, so a short night is not just a smaller version of a full one. Second, since you cannot predict your cycles in advance, the sensible way to catch a lighter moment is to watch for one. Tock’s smart wake looks at the 30 minutes before your set alarm and rings a little early if it sees you stirring, with your set time as the fallback so nothing gets missed. How to wake up easier in the morning puts that in context with the things that matter more.