We often imagine the night as a tunnel: we enter on one side, exit on the other, and nothing happens in between. This is far from what sleep recordings show. A night is a succession of waves, with ascents and descents, a recognizable order, and stages that have neither the same role nor the same effect on us.
Understanding this architecture is not a theoretical exercise: it explains why the first hours are the most restorative, why dreams mainly occur in the early morning, and why some awakenings are so difficult.
Four to six cycles of about ninety minutes
Sleep is organized into cycles of about 90 minutes, linked four to six times per night depending on its duration. Each cycle goes through the same stages, in the same order, but their proportions change throughout the night. This is the point that most simplified diagrams forget to mention, and it is precisely what makes the whole intelligible.
Also note that ninety minutes is an average. The actual cycle varies from person to person and from one moment of the night to another. Any method that promises to calculate your ideal wake-up time to the minute largely overestimates the regularity of living things.
N1, the threshold
This is the moment of falling asleep, the gateway. We detach from the world, the slightest noise brings us back to wakefulness, and if we are woken up at this stage, we will often swear we haven't slept at all. Jean-Paul Tassin, a neurobiologist at Inserm, notes that this stage is itself subdivided into several sub-stages.
It is also the realm of hypnagogic imagery: geometric shapes, faces, snippets of scenes, sometimes a voice or a sensation of falling. It's not yet a dream, it's the threshold. These images have a name, hypnagogia, and an experimental protocol has shown that one minute spent in this stage greatly increased the resolution of a problem with a hidden rule, whereas a deeper sleep caused this advantage to be lost. The window is narrow, and it has intrigued many artists.
This is also where the hypnic jerk occurs, that jolt accompanied by a sensation of falling that about two-thirds of people experience. Harmless, favored by stress, fatigue, and caffeine.
N2, the body of the night
Light slow-wave sleep occupies most of the early night and, in volume, the largest part of the total time. The arousal threshold rises, muscle activity decreases, and awareness of the outside world further diminishes.
It is a discreet yet central stage: this is where the night spends most of its time, and this stage is suspected of playing an important role in consolidating what has been learned during the day.
N3, deep slow-wave sleep
This is the most physically restorative stage. Brain waves become slow and ample, the arousal threshold is very high: this is the sleep from which one emerges groggy, disoriented, unable to know what day it is.
This deep sleep is concentrated in the early part of the night. This gives some truth to the idea that "sleep before midnight counts double." Nothing magical happens at midnight, but going to bed early mechanically provides more deep sleep, simply because one enters their own night earlier.
It is also from this stage that phenomena often misattributed arise: sleepwalking and night terrors occur during deep slow-wave sleep, thus in the early part of the night, and not during dreams. A night terror is not a nightmare: the child has open eyes, does not recognize loved ones, and retains no memory of it the next day.
REM sleep, the strangest stage
Its name says it all about the paradox: the brain is almost as active as when awake, while the body is completely paralyzed. It was the Lyon neurobiologist Michel Jouvet who discovered and characterized this phase in the late 1950s.
This paralysis, called muscle atonia, has a specific function: it prevents physically acting out dreams. Jouvet demonstrated this in cats by removing the small area that controls it: the animal, deeply asleep, then begins to mimic hunting scenes. Only three things escape atonia: eye muscles, breathing, and smooth muscles.
This ocular exception has a remarkable consequence. Since the eyes still move, a sleeper who knows they are dreaming can send a pre-arranged signal to the outside. This is how lucid dreaming was validated in the laboratory, starting in the 1970s and 1980s. Our REM sleep page details this phase.
Why early morning dreams
Here is the keystone of this entire architecture: deep slow-wave sleep dominates in the early night, REM sleep dominates in the late night. REM sleep phases, short in the first cycle, lengthen considerably over subsequent cycles.
Two very concrete consequences follow. The first hours are the most physically restorative, which gives meaning to a regular bedtime. And the dreams we remember are mostly those from the early morning, because they are the longest and closest to waking. Waking up a little earlier, paradoxically, increases the chances of emerging from a dream and remembering it. If you want to try, submit the first dream you remember upon waking.
Waking up at the right time
Hence the intuition, correct in principle, that it is better to wake up at the end of a cycle. Emerging from deep slow-wave sleep is unpleasant and leaves an inertia that can last half an hour; emerging from light or REM sleep is significantly gentler.
A caveat, however, on minute-by-minute calculations. As cycle duration varies, aiming for an exact multiple of ninety minutes is a comfortable approximation, not precision. What truly works is regularity: with a stable wake-up time, the body eventually organizes its night to finish a cycle at the right moment, all by itself.
An architecture that is not a long, calm river
Last stage, this one invisible: between cycles, the brain briefly returns to wakefulness. About ten times per night for good sleepers, more for others, and most often without any memory of it. These micro-awakenings are part of normal functioning.
They are even at the heart of a lively scientific debate. The dominant model places dreaming in REM sleep. Jean-Paul Tassin defends a minority and counter-intuitive hypothesis: since neuromodulators are inactive during this phase, there would be no consciousness, and the dream would actually be formed at the moment of micro-awakening, in reverse and in a few hundred milliseconds. A research hypothesis, not a consensus, and one of the most beautiful disagreements in the discipline. We present it in detail in our article on sleep phases.
A night, then, is not a tunnel but a landscape: deep valleys at the beginning of the journey, dreamy peaks in the early morning, and brief outcrops to the surface between each stage. Knowing it changes how we approach it.
Content for informational purposes, not a substitute for medical advice.
