Probably, but nobody can tell you what they dreamed. Not one dream from before writing survives. Dreams leave no bones, no tools, no paint. What we can do is work from two directions. Sleep science studies dreaming in living people and asks which animals share the sleep states tied to it. Early written records show that once people could write about dreams, they did: as messages, as misery, and as part of healing. In between lies a very long silence.
How to read the labels on this page: [Documented] means the evidence shows it directly. [Strongly supported] means strong sources agree. [Plausible] means the evidence allows it but doesn’t prove it. [Disputed] means qualified experts disagree. [Unknown] means the evidence can’t answer the question.
Can we know what prehistoric people dreamed?
No. [Unknown] This collection’s rule for material evidence is to climb from what is found to what it might mean, and to stop where the evidence stops. With dreams, the evidence stops before it starts.
What we have is an inference. Sleep labs collect reports of experiences recalled on awakening (Nir & Tononi 2010, Box 3). Sleep states linked to dreaming turn up in many mammals and birds (Blumberg et al. 2020). So people probably dreamed long before anyone wrote a dream down. [Plausible] That’s an inference from living brains, not an observation of ancient ones.
What people believed about dreams before writing is also [Unknown]. Many communities today carry their own knowledge about dreams, and some of it isn’t meant for outsiders. This page doesn’t summarize any community’s dream traditions. Those belong to their knowledge holders.
What happens in a sleeping brain?
In 1950, Eugene Aserinsky began graduate work at the University of Chicago under the sleep researcher Nathaniel Kleitman (Silber 2024). In a two-page paper published September 4, 1953, the two reported that sleeping adults go through periods of rapid eye movement (Aserinsky & Kleitman 1953). Their 20 adult sleepers had three or four such periods a night, and people woken during them reported dreaming (Silber 2024). [Strongly supported] That stage became known as REM sleep.
In early studies, waking people from REM sleep produced vivid dream recall 74–80% of the time, compared with 7–9% for the rest of sleep, called non-REM (Nir & Tononi 2010). Then researchers changed the question. Instead of asking sleepers whether they’d had a dream, they asked for “anything that was going through your mind just before you woke up.” Reports of experiences from non-REM sleep then ranged from 23% to 74%. Calling REM “the dream stage” was, in the same review’s words, an oversimplification. [Strongly supported in this review] Dream reports occur after awakenings from both kinds of sleep.
In 2017, a University of Wisconsin–Madison team recorded sleepers with 256 scalp electrodes. The researchers woke people many times a night and asked what they had just experienced. In both kinds of sleep, people who had less slow-wave activity in a region toward the back of the brain just before waking tended to report dreaming. People with more of that activity tended to report no experience. Watching that region live, the team predicted whether a non-REM sleeper would report a dream correctly 87% of the time (Siclari et al. 2017). [Documented, for these experiments] The authors suggest this “hot zone” may be a core part of dreaming. [Plausible]
Analogy: think of a video game engine that keeps building a whole world after the controller is unplugged. Nir and Tononi make the real point: a brain cut off from its surroundings can generate an entire world of experience on its own.
What are dreams for?
The sources checked here do not establish a single function. [Unknown] In the 2010 review’s table of theories, Hobson’s camp suggests dreaming builds a kind of virtual-reality model that may prepare the brain for learning. Another camp says the biological function is simply unknown. A third says dreaming may have no function at all: it’s a by-product of how sleep and consciousness evolved (Nir & Tononi 2010, Table 1).
A major research question concerns the relationship between sleep and memory (Rasch & Born 2013). [Strongly supported] Dreams are another matter. Reviewing the evidence in 2013, Rasch and Born found no convincing direct link between the brain’s memory “replay” during sleep and the dreams people report. Only 1–2% of dream reports include a genuine memory of a specific event from before sleep. [Strongly supported, as of 2013] So the popular line that “dreams are the brain’s nightly maintenance” goes further than the evidence.
Dreams do borrow from waking life. About half contain leftovers from waking experience, dropped into new and unrelated settings (Nir & Tononi 2010). [Strongly supported]
Who started studying dreams scientifically?
Watch out for “firsts.” Plenty of people wrote about dreams before the 1890s, as the section on early writing below shows. In 1893, Mary Whiton Calkins, an instructor in psychology at Wellesley College, complained that dreams had been endlessly theorized and almost never carefully observed. So she and a second observer, “S.,” kept paper, pencil, a candle, and matches close at hand and wrote down each dream on waking: 205 dreams over 55 nights for her, 170 over 46 nights for him. She reports that once, one of the two observers wrote out a dream in the dark, then found in the morning that the pencil was unsharpened and the page blank. In her own records, she could trace a link to waking life in all but 11% of dreams. For S., a link was traced in about half, partly because the two sorted their dreams differently (Calkins 1893, pp. 311–315). [Documented]
Sigmund Freud (1856–1939) trained in Vienna as a research scientist and then a doctor, and founded psychoanalysis (Freud Museum London). The Interpretation of Dreams came out in November 1899, though its title page says 1900 (Library of Congress; Lütkehaus 2000). [Strongly supported] Its core claim is that a dream fulfils a wish, in disguise (Freud Museum London).
In 1977, the Harvard psychiatrist J. Allan Hobson (1933–2021) and his colleague Robert McCarley proposed the “activation-synthesis” hypothesis. The brainstem switches the upper brain on during REM sleep, and the upper brain builds a dream by matching those signals against stored memories (Nir & Tononi 2010, historical discussion; World Sleep Society). [Documented, as their proposal]
Are dreams messages from the unconscious?
The 2010 review compares three approaches: a psychodynamic model associated with Freud, Hobson’s model, and a neurocognitive model that treats dreams as simulations built from memories and general knowledge (Nir & Tononi 2010, Table 1). [Documented as that review’s comparison] This is a dated map of competing explanations, not evidence that Freud’s interpretation was confirmed or that those three models exhaust present-day research.
A second question hides here. What dreaming does biologically is different from whether a dream means something to the person who had it. The second involves the dreamer, their language, and their culture.
Do animals dream? Can people steer their dreams?
There is broad agreement that mammals and birds have REM sleep. Two lizard species show something REM-like, but earlier studies of crocodilians and turtles didn’t find it. Dolphins and whales show no clear REM sleep, and the platypus mixes REM and non-REM features in one state. Mammals and birds haven’t shared an ancestor in more than 300 million years, and nobody yet knows whether REM sleep evolved once or more than once (Blumberg et al. 2020). [Disputed] Whether any animal experiences a dream is [Unknown], because animals can’t tell us.
In a lucid dream, the sleeper knows they’re dreaming. In 2021, four lab teams in France, Germany, the Netherlands, and the United States tried to talk with 36 volunteers during REM sleep. Lucid dreamers answered questions, including simple math, with planned eye movements or small face-muscle twitches. The teams got correct answers on 29 of 158 tries, from six people (Konkoly et al. 2021). [Documented] Most dreams still run without the dreamer steering (Nir & Tononi 2010).
When writing arrives, dreams show up on the page
Writing didn’t start dreaming. It changed what could survive. The three examples below aren’t the oldest dream records. They’re early ones checked against the sources cited here.
- Mesopotamia. The Babylonian poem Ludlul bēl nēmeqi (“the Righteous Sufferer”) was probably composed in the last centuries of the second millennium BCE (Hurowitz 2011). Its suffering speaker says he was just as miserable dreaming as awake (Tablet III, line 8). The following passages describe visitors in three dreams (eBL edition, trans. Benjamin R. Foster, III 1–48). [Documented, as literature] It’s a crafted poem, not a patient’s diary.
- Greece. At the healing sanctuary of the god Asklepios at Epidaurus, stone tablets inscribed in the second half of the 4th century BCE present the sanctuary’s accounts of visitors who slept there seeking divine help. The sanctuary set these stories up in the very room where visitors slept. One scholar argues that this shaped what visitors expected (Barrenechea 2016, pp. 255–258). [Documented, as the sanctuary’s own claims] The inscriptions can’t prove the cures happened.
- Greek medicine. The medical treatise On the Sacred Disease, probably from the late 5th or early 4th century BCE, argued that fears by night and sleeplessness come from the brain. It blamed cries in the night on sudden heating of the brain, and said the brain also heats up when blood rushes into it during a frightening dream (Jones translation, 1923, pp. 174–177). [Documented, as its author’s theory] It isn’t modern neuroscience.
These texts speak for their own times, not for people tens of thousands of years earlier.
How we know
What evidence exists: For prehistory, nothing direct. For the science: waking sleepers and asking them, plus brain recordings, patients with brain injuries, and animal sleep recordings. For early history: poems, inscriptions, and medical texts.
Whose voices are missing: Every dreamer before writing. Dream reports depend on memory and words. Calkins’s blank-page episode also shows how recording can fail: the pencil had not made marks. Many studies rely on small groups: two observers in 1893, 36 volunteers in 2021. Communities whose dream knowledge is private are absent by design. That’s a gap, not a verdict.
How power shaped the record: Credit is contested. In a 1996 account, Aserinsky wrote that Kleitman played only a peripheral role in the 1953 study. A 2024 issue of the journal SLEEP carried a translation of a 1926 Russian study that saw similar cycles in sleeping infants. In the same issue, Silber argued that Aserinsky and Kleitman still deserve the main credit. Among his reasons: they studied adults using electrical recordings, tied the cycles to dreaming, and followed up with more research (Silber 2024). [Disputed] The ancient records were shaped too. Ludlul is crafted poetry. The sanctuary’s compilers chose and shaped the Epidaurus stories. The Sacred Disease author was attacking rival healers (Jones, pp. 140–141).
Then vs. now
| What people believed or claimed then | What the sources checked here establish |
|---|---|
| In Ludlul, dreams bring divine messengers. | A poem praising a god. Science studies dreams as experiences the sleeping brain makes; it doesn’t test divine messages. |
| At Epidaurus, sleepers sought healing through dreams. | The cures are the sanctuary’s claims and can’t be checked. |
| On the Sacred Disease: in a frightening dream, blood rushes in and heats the brain. | The brain does matter. The 2010 and 2013 reviews used here explain dreaming through brain activity and brain chemicals, not heat. |
| 1899/1900, Freud: a dream is a disguised wish. | A 2010 review compared a Freud-based model with two alternatives. That historical comparison is not proof of the model. [Disputed] |
| 1950s: REM sleep is the dream stage. | Dreaming also happens in non-REM sleep. [Strongly supported] |
| Popular idea: dreams are the brain’s nightly maintenance. | The 2013 review did not establish a direct connection between reported dreams and memory replay. This dated finding does not settle every proposed function. |
| Popular idea: we know what prehistoric people dreamed. | We don’t. [Unknown] |