The Nightly Erasing

Sometime in the small hours, you lived a life that was not yours. There was a house you had never seen but somehow knew. A conversation with a person who was both a stranger and, impossibly, your mother. A staircase that led somewhere urgent. The narrative held together with the internal logic of fiction, and while it lasted, it was as real as anything. Then you opened your eyes, and within the time it took to reach for a glass of water, it was gone. By the time coffee was poured, you could not have sworn it happened at all.

This is the ordinary miracle of forgetting. On any given night, a healthy adult spends close to two hours dreaming, spread across four or five separate episodes. The brain composes elaborate scenes, casts characters, sustains suspense, and then, on waking, discards nearly all of it. Studies of dream recall suggest that the overwhelming majority of dream content evaporates within minutes of returning to consciousness. Some estimates put the loss as high as ninety-five percent.¹

For most of the twentieth century, this looked like a defect. If the sleeping brain went to the trouble of producing something so intricate, why would it fail so completely to keep a copy? The intuitive answer, that dreaming is a byproduct of a mind idling in the dark, has slowly given way to something stranger and more precise. The forgetting is not a failure of memory. It appears to be one of its functions. And the mechanism responsible turns on the presence, or rather the absence, of a single chemical.

A student watching sleeping eyes

The modern science of dreaming begins not with a theory but with an observation, made by a graduate student who was, by most accounts, not having a good year. In 1953, at the University of Chicago, Eugene Aserinsky was working under the physiologist Nathaniel Kleitman, a man who had already spent decades establishing that sleep was worth studying at all. Aserinsky rigged up equipment to record the eye movements of sleepers through the night, partly on his mentor’s suggestion and partly to justify his continued presence in the lab.²

What he saw did not fit the picture of sleep as a uniform descent into stillness. At intervals through the night, beneath closed lids, the sleepers’ eyes began to dart back and forth in rapid, coordinated bursts, as though they were watching something unfold. When Aserinsky and Kleitman woke sleepers during these episodes, the reports were remarkably consistent: vivid, story-like dreams, freshly interrupted. Wake them during the quieter stretches, and the dreams were vague or absent.

They had discovered what would come to be called REM sleep, for rapid eye movement, the phase in which most vivid dreaming occurs.³ The finding, published in the journal Science in 1953, reorganized the field. Sleep was not a single flat state but a cycle, and dreaming had a physiological address. Kleitman, often described as the founder of sleep research, had helped establish that the sleeping brain was doing structured, patterned work.

But identifying when we dream did not begin to explain why we forget. If anything, it deepened the puzzle. The dreams reported at the moment of waking were often detailed and emotionally charged, the sort of experience that, had it happened while awake, would have lodged itself firmly in memory. The problem was not that the dreams were faint. It was that something about the state in which they occurred prevented them from being written down.

The chemistry of keeping

To understand why a dream vanishes, it helps to understand what it takes for any experience to stay. Memory is not a passive recording. It is an active, chemically dependent process, and among the molecules that make it work is norepinephrine, a neurotransmitter that also functions as a stress hormone. Norepinephrine does several things at once: it sharpens attention, heightens arousal, and, crucially, flags an experience as significant enough to be preserved. When something startling or important happens while you are awake, a surge of norepinephrine helps stamp the moment into long-term storage. It is, in a sense, the chemical that decides an event is worth keeping.⁴

Here the picture turns. During REM sleep, the phase of the richest dreaming, norepinephrine levels fall to nearly zero. The locus coeruleus, the small cluster of neurons in the brainstem that is the brain’s primary source of the chemical, goes almost entirely quiet. The very molecule responsible for cementing experiences into memory switches off precisely when the mind is producing its most elaborate content.⁵

The metaphor writes itself. The brain generates the dream, projects it in full, and then finds it has no ink in the pen. The experience plays out across neural circuits with all the vividness of waking life, but the machinery that would normally save the file has been powered down. This is why the dream can feel so real in the instant of waking and dissolve so completely a moment later. It was never encoded in a durable form to begin with. What we experience on waking is less a memory than the fading echo of an event that was never properly recorded.

This chemical account also explains one of the most reliable features of dream recall: its dependence on timing. Norepinephrine does not return to full waking levels instantly. If you wake abruptly in the middle of a REM episode, before the chemical has flooded back, you catch the dream mid-scene, and for a brief window it can be transferred into ordinary memory. Drift back to sleep instead, and the dream is gone for good. This is why a morning alarm sometimes seems to interrupt a dream in progress, and why people who wake frequently through the night report far more dreams than those who sleep straight through. Recall, on this view, is largely a coincidence of when consciousness happens to return.⁶

The idea that we dream in order to forget

In 1983, the puzzle attracted an unlikely investigator. Francis Crick, who three decades earlier had co-discovered the structure of DNA, had turned his attention to the brain. Working with the mathematician Graeme Mitchison, he proposed a theory of dreaming that inverted the usual assumption. In a paper published in Nature, they argued that the function of REM sleep was not to consolidate memories but to remove them. We dream, they suggested, in order to forget.⁷

The reasoning drew on the emerging understanding of neural networks. A brain that learns by strengthening connections will, over time, accumulate spurious associations, parasitic patterns, and useless overlaps, the neural equivalent of clutter. Crick and Mitchison proposed that during REM sleep the brain runs a kind of reverse-learning process, generating the strange, disjointed imagery of dreams as a side effect of pruning away these unwanted connections. The bizarre quality of dreams, on this account, was not incidental. It was a symptom of the deletion in progress. The dream was the sound of the brain taking out the trash.

The theory was speculative and remains contested. But it reframed forgetting as something the brain might actively want to do, rather than a limitation it suffers. And it converged, from a completely different direction, with the chemical story. Whether you emphasize the absence of norepinephrine or the pruning of connections, the conclusion is similar: the sleeping brain is not merely failing to keep dreams. It is engaged in a process for which not keeping them is the point.

The recorder is unplugged

There is a third strand to the explanation, and it concerns the architecture of memory itself. New experiences are not filed directly into permanent storage. They pass first through the hippocampus, a seahorse-shaped structure that acts as a temporary buffer, capturing the events of the day before gradually transferring them to the cortex for long-term keeping. This handoff, the slow migration of memory from hippocampus to cortex, is thought to be one of the central tasks of sleep.⁸

During REM sleep, however, the normal traffic between these regions changes. Research on the neurochemistry of REM suggests that the flow of information runs largely outward, from the hippocampus toward the cortex, while the return pathway that would allow new dream content to be recorded is suppressed. Acetylcholine, another neurotransmitter, floods the system during REM and appears to block the hippocampus from taking in fresh input.⁹ The result is a curious asymmetry. The circuits that generate the dream are active, but the circuits that would capture it for later are not receiving. The film is playing, and the recorder is unplugged.

Into this territory came the work of Robert Stickgold, a cognitive neuroscientist at Harvard who has spent much of his career studying how sleep sorts memory from noise. Stickgold’s research supports a picture in which sleep is not a passive off-switch but an active curator, deciding what to keep and what to let go. Memories are tested, recombined, and either strengthened or allowed to fade.¹⁰ Dreaming, in this framing, may be part of how the brain reorganizes the day, extracting patterns and discarding detail. That we do not remember the process is consistent with its purpose. The sorting is not meant to be reviewed.

Stickgold and others have also confirmed, experimentally, the timing effect that the chemical account predicts. Wake a subject during REM and dream recall rises sharply. Let them sleep through, and it collapses. The dreams that people do remember are disproportionately the ones interrupted at the right moment, which means that the vast library of nightly dreaming remains, for nearly all of us, permanently sealed.

Why the forgetting protects you

It is tempting to read all of this as a story of loss, as though the brain were carelessly throwing away its own creative work every night. But there is a case to be made that the amnesia is not a bug at all. It is a safeguard, and it protects something we could not easily do without: the integrity of our memory of what actually happened.

Consider what it would mean to remember dreams with the same fidelity as waking events. The brain uses the same broad machinery to reconstruct a memory whether the original experience was real or imagined. If dreams were encoded and stored with full clarity, they would sit in the same archive as your real past, indexed alongside genuine events, with no reliable tag marking them as fiction. Over a lifetime of nights, you would accumulate thousands of vivid false histories: conversations that never occurred, places you never went, versions of people you love behaving in ways they never did. The line between what happened and what you merely dreamed would blur, and it would blur in a way that no amount of effort could later untangle.

Seen this way, the nightly erasing performs a kind of quality control. By declining to save dreams, the brain keeps its record of reality relatively honest. Imagination and memory are kept in separate drawers, and the separation is enforced not by willpower but by chemistry, by the simple expedient of switching off the molecule that writes memories during precisely the hours when the mind is most inventive. The forgetting is not memory failing. It is memory protecting itself from contamination.

This does not mean dreams are meaningless or that they leave no trace. The emotional residue of a dream can color a morning. The problem-solving that seems to happen overnight, the sense of waking with a decision already made, may owe something to the sorting that dreaming reflects. The point is subtler: the content of the dream, the specific narrative, does not need to be preserved for its work to be done. The dream is a process, not a document, and a process leaves its mark without needing to be filed away.

Catching what was never saved

There is, for the curious, a modest way to intervene. Because dream recall depends on that narrow window before norepinephrine and full waking chemistry return, the dreams can sometimes be caught if you move quickly. Keeping a notebook or a phone by the bed and recording whatever fragments remain in the very first moments after waking, before the day’s first thoughts crowd in, reliably increases the number of dreams people can retrieve. People who keep dream journals over weeks tend to report a steady rise in recall, not because they are dreaming more but because they are intercepting the record before it dissolves.¹¹

What this technique reveals is quietly profound. The dreams were never lost in the sense of being deleted from a place they once occupied. They were simply never saved. There was no file to recover, only an experience that flickered through consciousness on its way to nowhere, unless a waking mind happened to catch it in the act.

So the next time a dream slips away at dawn, leaving only the impression that something rich just occurred, it is worth resisting the small grief that comes with it. The forgetting is not the brain being wasteful with its best material. It is the brain doing exactly what it must to keep you anchored to your own life, sorting the real from the invented, the day that happened from the night that only seemed to. Every morning, you wake up as yourself, and part of the reason is that your brain, in the dark, chose to let the dreams go.

Sources

1. Aserinsky, E. and Kleitman, N., “Regularly Occurring Periods of Eye Motility, and Concomitant Phenomena, During Sleep,” Science, 1953. — https://www.science.org/doi/10.1126/science.118.3062.273
2. Brown, R. E. et al., “Control of Sleep and Wakefulness,” Physiological Reviews, 2012. — https://journals.physiology.org/doi/full/10.1152/physrev.00032.2011
3. Crick, F. and Mitchison, G., “The function of dream sleep,” Nature, 1983. — https://www.nature.com/articles/304111a0
4. Stickgold, R. and Walker, M. P., “Sleep-dependent memory triage: evolving generalization through selective processing,” Nature Neuroscience, 2013. — https://www.nature.com/articles/nn.3303
5. Hasselmo, M. E., “Neuromodulation: acetylcholine and memory consolidation,” Trends in Cognitive Sciences, 1999. — https://www.cell.com/trends/cognitive-sciences/fulltext/S1364-6613(99)01365-0
6. Aserinsky, E., “The Discovery of REM Sleep,” Journal of the History of the Neurosciences, 1996. — https://www.tandfonline.com/doi/abs/10.1080/09647049609525671
7. Diekelmann, S. and Born, J., “The memory function of sleep,” Nature Reviews Neuroscience, 2010. — https://www.nature.com/articles/nrn2762
8. Nir, Y. and Tononi, G., “Dreaming and the brain: from phenomenology to neurophysiology,” Trends in Cognitive Sciences, 2010. — https://www.cell.com/trends/cognitive-sciences/fulltext/S1364-6613(09)00248-9

Related reading

• The Bony Shrug at the Bottom of Your Face
• The Toe That Forgot Its Purpose

More from the Body edition →