Why we forget that we were asleep: the strange discontinuity of waking

A person falls asleep at eleven and wakes at seven. Subjectively, almost nothing happened. Not eight hours of darkness — no experience of duration at all. The two moments abut.

This is so ordinary that it is rarely remarked on, and it is genuinely strange. Substantial neural activity occurred throughout. Dreams happened, four to six times. None of it left a felt gap.

Two different absences

Worth separating, because they have different explanations.

The absence of dreams from memory. Dreams occurred and were forgotten — the standard account being that noradrenaline is at its nadir during REM, and encoding into long-term memory requires it. Content existed in working memory and never consolidated.

The absence of felt duration. Different, and less often discussed. Even where dreams are recalled, the intervals between them are not experienced as waiting. There is no sense of time having passed and no sense of having been unconscious.

The second is closer to the anaesthesia case, and the comparison is instructive.

What anaesthesia demonstrates

People emerging from general anaesthesia commonly report that no time passed — a four-hour operation feels instantaneous, and patients frequently ask when the surgery will start.

This differs from ordinary sleep, where people usually have some sense that a night has occurred, largely from context: it is light, the alarm is going, the body feels different.

The anaesthesia case suggests something important: the sense of elapsed time depends on there having been continuous experience to fill it, or at least on memory of experience. Remove both, and duration does not register — the two edges of consciousness are simply stitched together.

Sleep-onset amnesia

The specific mechanism at the near edge is documented. Retrograde amnesia for the period immediately before sleep onset means people typically cannot recall the last few minutes of wakefulness.

This is why the moment of falling asleep is almost impossible to observe — the memory of the approach is erased. People report drifting thoughts, then nothing, then morning. The transition itself is systematically unavailable.

It also explains the common experience of not remembering whether you finished a page, and of being unable to establish when you fell asleep, which makes self-reported sleep latency notoriously inaccurate.

Time inside dreams

A related question with an experimental answer.

The folk belief is that dream time runs faster than real time — that a long dream occupies moments. Stephen LaBerge tested this using lucid dreamers signalling with eye movements: instructed to signal, count ten seconds inside the dream, then signal again, the intervals recorded on the polygraph corresponded closely to ten seconds of real time.

Dream time, at least by this measure, runs at roughly real time.

Which means the impression of extended dream narratives compressed into moments is a reconstruction problem rather than a temporal one — the remembered dream is assembled from fragments, and assembly does not preserve duration.

Philosophical company

The nightly discontinuity has been used in arguments about personal identity for a long time.

If continuity of consciousness is what makes you the same person tomorrow, then that continuity is broken nightly, and what persists must be something else — memory, physical continuity, psychological connectedness. Derek Parfit's work on personal identity uses cases of interrupted consciousness in exactly this way.

Most people are entirely untroubled by this in practice, which is itself the interesting datum. We accept nightly discontinuity without concern while regarding much smaller interruptions in other contexts as significant.

The practical residue

Two things follow that are not merely philosophical.

Self-reported sleep is unreliable. Because the transitions are not encoded and duration is not felt, people are poor at estimating how long they took to fall asleep, how often they woke, and how long they slept. Insomniacs typically underestimate sleep; people with apnoea are unaware of hundreds of arousals. Anyone reasoning about their own sleep from impression alone is reasoning from a bad instrument.

The morning is not a continuation. Sleep inertia — the impairment on waking, particularly from slow-wave sleep — means the first period after waking is not a resumption of yesterday's mental state. Decisions made in the first twenty minutes are made by a system that is not fully online, which is a reasonable argument for not checking messages before getting up.