What Science Actually Knows About Dreams

Search "What Do Dreams Mean" and Watch the Internet Fall Apart

Type that phrase in and you'll get dream dictionaries, symbol charts, and a lot of confident people telling you what teeth falling out signifies. Almost none of it is research. It's fortune-telling with a nicer font.

Which is a shame, because the real science of dreaming is far more interesting than the mysticism. Researchers have woken sleepers hundreds of times over to catch what they were experiencing. They've had a two-way conversation with someone who was asleep.

The honest headline: we know a lot about when and how dreams happen, and almost nothing about why. That gap is real and this post won't paper over it.

Key Facts

• Dreams happen in non-REM sleep too — up to 70% of non-REM awakenings produce a dream report (Siclari et al., Nature Neuroscience, 2017)
• Recall depends hugely on how you wake up: classic work found ~80% recall from REM awakenings vs ~6% from non-REM (Ruby, Frontiers in Psychology, 2011)
• Nightmares and night terrors are different states. Nightmares come from REM in the second half of the night, with full recall. Night terrors come from deep non-REM in the first third, with almost none (StatPearls)
• Lucid dreaming is lab-verified. Dreamers signalled with pre-agreed eye movements while polysomnography confirmed REM sleep (LaBerge et al., 1981)
• About 55% of people have had a lucid dream at least once, roughly 23% have them monthly (Saunders et al. meta-analysis, 2016)
• The function of dreaming is unsettled. Memory, emotion, threat rehearsal, and "no function at all" are all still live (Ruby, 2011)

"Dreams Only Happen in REM" Is Wrong

This one is worth killing off, because it's still repeated everywhere.

REM stands for rapid eye movement — the sleep stage where your eyes flick around under closed lids and your body goes temporarily paralyzed. When researchers first found it in the 1950s, waking people from REM produced vivid dream reports and waking them from other stages mostly didn't. So REM got labelled "dream sleep." Fair conclusion at the time.

It didn't survive better methods. In a 2017 study, researchers woke volunteers repeatedly through the night and asked what, if anything, they had been experiencing — 233 awakenings in one experiment and 815 in another with trained subjects ¹. Up to 70% of non-REM awakenings produced a report of a dream experience.

Non-REM dreams tend to be shorter, plainer, and less story-like. But they're there. The old finding wasn't wrong about REM being dream-rich. It was wrong about the rest of the night being empty. For the tour of the stages themselves, see what actually happens while you sleep.

The same 2017 work found something sharper. Dreaming, in both REM and non-REM, tracked with reduced low-frequency brain activity in a bilateral region at the back of the brain the researchers called the posterior hot zone — roughly the occipital lobe extending up into the precuneus and posterior cingulate ¹. Activity there predicted whether someone reported a dream. Higher-frequency activity in that zone tracked with what they dreamed about.

That's about as close as anyone has come to catching a dream in progress.

Why You Think You Don't Dream

You do. You just don't remember.

Everyone dreams. The variable is recall, and recall depends enormously on how you wake up. The classic Dement and Kleitman result: about 80% of awakenings out of REM sleep produced a dream report, versus about 6% out of non-REM ². Later work softened those numbers — between 5% and 30% of REM awakenings produce nothing even in healthy sleepers, and non-REM recall is much higher than 6% when you ask carefully ^{2 1}.

Practically, this means:

• Waking naturally at the end of a cycle, often out of REM, gives you the best shot at remembering.
• Alarms that yank you out of deep sleep hand you nothing. There was probably content. You just weren't given a chance to encode it.
• REM clusters in the second half of the night. Cut your sleep short and you cut the dream-rich part disproportionately.
• The memory is fragile. If you get up and start your day, it's gone in under a minute.

None of this makes you a "non-dreamer." It makes you someone whose alarm is well-timed for productivity and badly timed for dream recall. That grogginess in the same window has its own name and its own research — see sleep inertia.

Nightmares and Night Terrors Are Not the Same Thing

This distinction matters, especially for parents, and it gets blurred constantly.

Source: StatPearls, Parasomnias in Adults ³.

Both are parasomnias — unwanted events that happen during sleep. But they come from opposite ends of the night and opposite states of the brain ³.

A night terror is a partial arousal: part of the brain surges toward wakefulness while the rest stays in deep sleep. The person looks awake and terrified and is neither. They're not experiencing a scary story. There usually isn't a story. Which is why the standard advice is to keep them safe rather than wake them, and why they remember nothing in the morning while you're still rattled.

A nightmare is a dream you had and can describe. The distress carries over because you actually woke up.

If night-time episodes are frequent, injuring someone, or new in adulthood, that's worth taking to a doctor rather than reading about. First-episode sleepwalking in an adult, in particular, prompts investigation for other causes ³.

Lucid Dreaming: The Weirdest Verified Result in Sleep Science

For most of the 20th century, lucid dreaming — knowing you're dreaming while it's happening — was unfalsifiable. A story people told.

Then someone found the loophole. During REM sleep your body is paralyzed, but your eye muscles aren't. So a dreamer who becomes lucid could, in principle, send a message out.

In 1981, LaBerge and colleagues had five subjects agree on a specific eye-movement pattern before sleep. If they became lucid, they were to make that exact signal. The signals showed up on the recording, during confirmed REM sleep ⁴. Small study, enormous implication: it was now possible to timestamp events inside a dream from outside it.

Forty years later, four independent labs in France, Germany, the Netherlands, and the US pushed it further. Across 36 participants, researchers asked lucid dreamers questions mid-dream — spoken math problems, flashed lights, taps — and had them answer with eye movements or facial muscle twitches ⁵.

The honest numbers: across 158 attempts during signal-verified lucid dreams, 18.4% got a correct response (29 correct answers). About 17.7% were ambiguous and 60.1% got no response at all ⁵.

That's a low hit rate. It's also 29 more real-time conversations with a sleeping brain than anyone had before. Four labs, different methods, same result — that independent agreement is what makes it convincing rather than a fluke.

About 55% of people report having had a lucid dream at least once, and around 23% have them once a month or more, according to a meta-analysis of 50 years of studies ⁶. So it's common. It just isn't reliable.

What Are Dreams For? Nobody Knows

Here's where the field earns its honesty. There are several serious theories, none settled ².

Memory consolidation. Sleep helps move new learning into durable storage — that part is solid, and we covered it in sleep and memory. The dream-specific claim is that dream content reveals that process running.

The evidence: 99 people trained on a virtual maze, then either napped or stayed awake. Four nappers spontaneously mentioned the maze in their dream reports — and those four improved about tenfold more than nappers who didn't ⁷. Four people. That's a hint, not a finding.

It replicated, though. An overnight study of 39 students found the same pattern: those who incorporated the maze into dream reports improved far more on completion time (172 seconds versus 2) ⁸. Task content showed up in reports from every stage of sleep, not just REM ⁸. Still small. Still correlational — dreaming about the maze might be a symptom of consolidation rather than the mechanism.

Threat simulation. Revonsuo proposed in 2000 that dreaming evolved as a safe rehearsal space for danger, pointing to how overrepresented threats are in dream content compared to waking life ⁹. It explains the prevalence of being chased. It's very hard to test.

Emotional processing. Dream reports are loaded with strong emotion, especially fear, and the amygdala — the brain's threat-processing hub — is active during REM ².

No function at all. Also on the table. Dreaming may be a byproduct of a sleeping brain doing other necessary work, the way a running engine makes noise without the noise being the point ².

Pick a side? I'd bet on some mix of the first and third, and I'd bet against ever getting a clean answer. Dreams are subjective reports of an unconscious state. That's a brutal measurement problem, and no amount of imaging fixes it.

What This Rules Out

Since we're being rigorous about "we know" versus "we think," here's what the research does not support:

• Universal dream symbols. No dictionary. Dream content tracks your own waking life, not a shared codebook.
• Dreams as prediction. Nothing in the literature supports it.
• "I never dream." You do. You don't recall.
• Dreams only in REM. Dead, as of the 2017 work above ¹.

Where Komori Fits

Direct about the limit: Komori cannot see your dreams. Nothing can. Dream research requires waking people up and asking them, which is exactly why it's such slow, expensive science.

Komori is being built as a contactless system that senses from above the bed — no camera, nothing worn — with a base unit on the nightstand handling power, Wi-Fi, and local processing. When it ships, it's designed to log sleep position, movement and restlessness, bed-exit events, sound events, and room conditions like temperature, humidity, CO2, light, and noise. It's a general wellness device, not a medical or diagnostic one, and it doesn't stage your sleep.

Where it's relevant to this post is thinner and more honest: dream recall depends on when and how you wake up, and that's shaped by whether your night was continuous or chopped into pieces. Movement and restlessness data is a rough read on the second part, and we've written about what movement actually says about sleep. If your nights are consistently broken up, the second half — where REM concentrates — takes the hit.

That's the connection. It's a real one, and it's not a dream-reading feature.

The Bottom Line

Dreams are not messages, and they're not noise either. They're an experience your brain generates in several sleep stages, most richly in REM, from a specific region at the back of the head — and you can only report the ones you happened to wake up next to.

We can now, barely, talk to people inside them. We still can't say what they're for. Anyone who tells you otherwise with total confidence is selling something.

If nightmares are wrecking your sleep, or someone in your house is having violent episodes at night, skip the internet and talk to a doctor. That's the one part of this topic where the answers are clinical, not philosophical.

---

Footnotes

1. Siclari F, Baird B, Perogamvros L, et al. "The neural correlates of dreaming." Nature Neuroscience, 2017;20(6):872–878. ↩ ↩² ↩³ ↩⁴

2. Ruby PM. "Experimental Research on Dreaming: State of the Art and Neuropsychoanalytic Perspectives." Frontiers in Psychology, 2011;2:286. ↩ ↩² ↩³ ↩⁴ ↩⁵

3. Fariba KA, Tadi P. "Parasomnias in Adults." StatPearls Publishing, updated 2023. ↩ ↩² ↩³

4. LaBerge SP, Nagel LE, Dement WC, Zarcone VP. "Lucid dreaming verified by volitional communication during REM sleep." Perceptual and Motor Skills, 1981;52:727–732. ↩

5. Konkoly KR, Appel K, Chabani E, et al. "Real-time dialogue between experimenters and dreamers during REM sleep." Current Biology, 2021;31(7):1417–1427. ↩ ↩²

6. Saunders DT, Roe CA, Smith G, Clegg H. "Lucid dreaming incidence: A quality effects meta-analysis of 50 years of research." Consciousness and Cognition, 2016;43:197–215. ↩

7. Wamsley EJ, Tucker M, Payne JD, Benavides JA, Stickgold R. "Dreaming of a learning task is associated with enhanced sleep-dependent memory consolidation." Current Biology, 2010;20(9):850–855. ↩

8. Wamsley EJ, Stickgold R. "Dreaming of a learning task is associated with enhanced memory consolidation: Replication in an overnight sleep study." Journal of Sleep Research, 2019;28(1):e12749. ↩ ↩²

9. Revonsuo A. "The reinterpretation of dreams: an evolutionary hypothesis of the function of dreaming." Behavioral and Brain Sciences, 2000;23(6):877–901. ↩