Why Do We Dream — and Why Do We Forget Our Dreams?

You wake up with a story in your head. You were flying, or late for an exam in a school you left twenty years ago, or talking to someone who died long ago as if nothing had happened. It felt completely real. Then you reach for your phone, or you get up to make tea, and by the time the kettle boils, the story is gone — not faded, but vanished, as if it had never been there at all.

This happens to almost everyone, almost every night. Two questions hide inside it. Why does the brain put on these elaborate shows in the first place? And why does it then throw the recording away? Neuroscience now has good answers to both, and the two answers turn out to be connected: we forget our dreams largely because of the same brain chemistry that makes dreaming possible.

The night shift: what sleep actually looks like

First, a correction to a common idea. Sleep is not one long shutdown. It is a repeating cycle of stages, and dreaming belongs mostly to one of them.

Through the night, your brain cycles between non-REM sleep (three stages, from light dozing to deep, slow-wave sleep) and REM sleep — rapid eye movement sleep, named for the darting of the eyes beneath closed lids. You enter your first REM period about ninety minutes after falling asleep, and the cycles repeat four to six times a night, with REM periods getting longer towards morning. That is why the dreams you remember are usually the ones from the last hours of sleep: they happen closest to waking.

REM sleep is strange territory. Your brain is almost as active as when you are awake — the visual cortex, the emotional centres, and the memory systems are all firing. At the same time, your body is effectively paralysed: the brainstem sends signals that relax the muscles that hold your posture, so you cannot act out whatever you are dreaming. Your breathing and heart rate become fast and irregular. This combination — a waking brain in a sleeping body — is the stage on which dreams are performed.

It was only in 1953 that researchers Eugene Aserinsky and Nathaniel Kleitman, watching sleeping volunteers, first described REM sleep and linked it to dreaming. Before that, dreams were largely the property of philosophers and psychoanalysts. After it, they became something you could measure.

Why do we dream? Four serious answers

Sigmund Freud famously called dreams the “royal road to the unconscious” and treated them as disguised wishes. Modern neuroscience is kinder to Freud than it used to be about one thing — dreams do reflect our emotional lives — but it has not kept his system of hidden symbols. The evidence points instead to several jobs that dreaming seems to do, and they are not mutually exclusive.

Dreams consolidate memory. During sleep, the brain replays and reorganises the day’s experiences, transferring important ones from short-term holding areas towards long-term storage. This work is shared between deep non-REM sleep and REM sleep. Dreaming may be the conscious side-effect of that replay: as the sleeping brain reactivates fragments of recent experience and links them to older memories, you experience the result as a story. The story is often scrambled because the replay is not a faithful recording — it is the brain extracting patterns, testing connections, and filing things away.

Dreams process emotion. This is the job most strongly tied to REM sleep. During REM, the brain reactivates emotional memories, but in an unusual chemical state: stress neurochemicals are dialled low. Psychologist Matthew Walker has described this as a form of overnight therapy — the brain replays a difficult experience without the full alarm response, so that by morning the memory remains but its sting has softened. That may be why a bad day often feels more manageable after a night’s sleep, and why people deprived of REM sleep tend to react more strongly to emotional events the next day.

Dreams rehearse threats. The Finnish neuroscientist Antti Revonsuo proposed that dreaming is an ancient simulation system: the sleeping brain stages threatening situations — being chased, falling, confronted — so that we can practise responses safely. Children, who dream more about threats and animals than adults do, fit this picture, as does the sheer frequency of chase and danger dreams across cultures. You do not need to believe every dream is a rehearsal to accept that a brain that practises danger in simulation is better prepared for danger awake.

Dreams explore loose connections. A more recent theory, called NEXTUP and proposed by researchers Antonio Zadra and Robert Stickgold in 2021, argues that dreaming lets the brain wander through weak, unusual associations that waking focus would suppress. Awake, your brain favours strong, useful links: keys go with doors, rain goes with umbrellas. Asleep, it wanders into the margins of its own knowledge, linking things that rarely belong together. That wandering may be where creative insight comes from — the strange dream combination that, the next morning, turns into a new idea. It also explains why dreams feel so odd: they are built from exactly the links your waking mind would filter out.

None of these theories claims to be the whole story, and researchers still argue about their relative importance. But together they paint a picture of dreaming as work, not entertainment: the sleeping brain is sorting, soothing, rehearsing, and exploring.

Why do we forget? The chemistry of a missing save button

If dreams are so important, why do they evaporate? The answer lies in how memory is made.

Making a lasting memory is a two-step process. First the experience must be encoded — tagged as worth keeping. Then it must be consolidated — transferred from fragile short-term holding into stable long-term storage. Both steps depend on brain chemistry, and during REM sleep that chemistry is wrong for saving.

The key player is norepinephrine, a chemical messenger closely related to adrenaline that helps the brain pay attention and lay down memories. During REM sleep, the brainstem neurons that release norepinephrine go almost silent. At the same time, another messenger, acetylcholine, surges — and acetylcholine drives the vivid imagery of dreams. So the dreaming brain is in a peculiar state: brilliant at generating experiences, poor at recording them. One researcher memorably compared it to filming a movie without pressing save.

A second mechanism actively erases. In 2019, a team funded by the US National Institute of Neurological Disorders and Stroke reported in the journal Science that a group of neurons in the hypothalamus — cells that produce a messenger called melanin-concentrating hormone, or MCH — fire during REM sleep and actively weaken recently formed memories. In experiments with mice, switching these neurons off during REM sleep improved the animals’ memory; switching them on made it worse. The effect was specific to REM sleep. Since dreams are generated during the same stage when these forgetting neurons are active, the researchers suggested that MCH activity may prevent dream content from being stored in the hippocampus, the brain’s memory hub — and that is why the dream slips away.

A third factor is simpler: waking up destroys the trace. Dreams are fragile, state-dependent memories — they belong to the sleeping brain’s chemistry and lose their footing when the waking brain takes over. The moment you wake, your senses flood in, your attention shifts, and new experiences compete for the same memory machinery. That is why dreams vanish within seconds or minutes unless you deliberately rehearse them immediately: lie still, keep your eyes closed, and run the dream through your mind before doing anything else. People who keep a notebook by the bed and write immediately on waking often report a dramatic increase in recall — not because they dream more, but because they catch the dream before the waking world overwrites it.

Wake timing matters too. If you wake directly from a REM period — which is common in the morning, when REM periods are longest — you are far more likely to remember a dream than if you wake from deep sleep. An alarm that drags you out of deep sleep can erase a night of dreaming from memory entirely, even though the dreaming certainly happened. Forgetting a dream, in other words, is no evidence that you did not dream. Everyone who reaches REM sleep dreams; not everyone wakes at the right moment to catch it.

Is forgetting a flaw or a feature?

It is tempting to see dream forgetting as a design flaw. But there are good reasons to think it is protective.

First, consider what would happen if every dream were stored as a real memory. Dreams mix real people with impossible events, familiar places with invented ones. If those combinations were filed alongside genuine memories, they could contaminate your picture of what actually happened. Forgetting the dream keeps your memory of reality clean. Stickgold’s version of this argument is blunt: the brain must forget its nightly fictions, or it would start believing them.

Second, the forgetting machinery does useful work beyond dreams. The same MCH system that erases dream traces also helps the brain prune unimportant daytime information, preventing overload. Sleep is not just for keeping memories; it is also for discarding them. A brain that remembered everything would be as dysfunctional as one that remembered nothing.

And third, the emotional work of dreaming does not require remembering the dream. If REM sleep softens the emotional charge of a difficult memory, you benefit the next morning whether or not you can describe the dream that did the softening. The therapy works without the transcript.

The edge cases: nightmares, lucid dreams, and remembering more

A few exceptions sharpen the picture. Nightmares are remembered more often than ordinary dreams, largely because they wake you directly from REM sleep — the fear jolts you awake at exactly the moment when recall is possible, and the strong emotion tags the experience as important. Recurring nightmares after trauma are a sign that the emotional-processing system is struggling, not that it is broken beyond repair; therapies that rewrite the nightmare’s ending while awake have good evidence behind them.

Lucid dreaming — knowing you are dreaming while still asleep — is real and has been verified in sleep laboratories, where trained dreamers signalled their awareness with pre-agreed eye movements while their brain waves showed they were still in REM sleep. Lucid dreamers remember more, probably because the awareness itself engages the attention and memory systems that are usually quiet during REM. It can be trained, though the evidence on techniques is mixed, and deliberately chasing it can disturb the very sleep it depends on.

And if you want to remember ordinary dreams, the method is unglamorous but effective: keep a notebook and pen by the bed, and when you wake, stay still and replay whatever fragments remain before you move or look at your phone. Most people find their recall improves within a week or two. What improves is not the dreaming — that was always happening — but the catching.

Why this matters

Dreaming sits at the intersection of two things we still do not fully understand: consciousness and memory. Every night, your brain builds an entire world, populates it with people, convinces you it is real, and then — in most cases — deletes it before breakfast. That nightly performance is not a glitch. It is the visible part of work your brain does to keep your memories organised, your emotions bearable, and your picture of reality intact.

There is something quietly reassuring in that. The dream you lost this morning was not wasted. Its job may already be done: a memory filed, an emotion softened, a connection tested. You do not need the transcript to keep the benefit. And the forgetting itself is part of the system working as it should — a brain that knows the difference between a story worth keeping and a story worth letting go, and that makes that judgement, night after night, without asking your permission.

What happens next

Dream science is moving from description to intervention. Researchers are testing whether sounds or smells delivered during sleep can strengthen specific memories — a technique called targeted memory reactivation — and whether treating nightmares can ease post-traumatic stress. Brain imaging is beginning to decode fragments of dream content from patterns of activity, though reading a dream like a film remains far out of reach. And the forgetting machinery itself, the MCH system and its neighbours, is a target of interest for diseases of memory: understanding how the healthy brain discards may eventually help when diseased brains discard too much, or too little.

The oldest questions about dreams — why we have them, why they vanish — are turning into practical ones: how to sleep better, remember what matters, and let go of what does not. That is a fitting next chapter for a mystery humans have argued about for thousands of years, from temple priests to psychoanalysts to neuroscientists. The tools have changed. The nightly show has not.

Sources

• Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD), “What happens during sleep?”: https://www.nichd.nih.gov/health/topics/sleep/conditioninfo/Pages/what-happens.aspx
• National Institutes of Health, “REM sleep may help the brain forget” (2019, on the Science study of MCH neurons): https://www.nih.gov/news-events/nih-research-matters/rem-sleep-may-help-brain-forget