Why do we dream? The science of dreaming, explained

Nobody has proven why we dream. The leading scientific accounts agree on more than their names suggest: a dream is an experience the sleeping brain builds from memory and current concerns, with its emotional systems active and its self-monitoring regions turned down. Where they part is on whether dreaming does a job of its own, such as strengthening memory or softening emotion, or is a by-product of a brain that keeps thinking while it sleeps.

What are dreams?

A dream is the experience of a world while you are asleep: places, people, movement, speech and feeling, usually lived as real while it lasts. Sleep laboratories use a wider term, sleep mentation, for anything a sleeper reports on waking, from a full story to a single drifting thought. Short definitions of these and other terms are in the glossary.

REM sleep and the first laboratory dreams

Modern dream science starts in Chicago in 1953, when Eugene Aserinsky and Nathaniel Kleitman described regular periods of rapid eye movement during sleep and the changes that came with them¹. In 1957 William Dement and Kleitman woke sleepers at different points in the night. Woken during rapid eye movement (REM) sleep, they usually reported a dream; woken at other times, they rarely did². For a decade REM sleep and dreaming were treated as nearly the same thing.

Dreaming outside REM sleep

That equation did not hold. In 1962 David Foulkes asked people woken from non-REM (NREM) sleep whether anything at all had been passing through their minds, instead of asking whether they had been dreaming, and many described experiences, some of them fully dreamlike³. Domhoff counts this work as the point at which NREM dreaming was established in detail. NREM reports are on average more thought-like, and dreamlike ones are more common late in the night⁶.

The clearest modern test came in 2017 from Francesca Siclari, Giulio Tononi and colleagues in Wisconsin. Across three experiments they recorded sleepers with high-density EEG, woke them more than a thousand times from NREM and REM sleep, and asked each time whether they had been experiencing anything. In both kinds of sleep, a dream report went with a local drop in slow, low-frequency activity in a region at the back of the brain, which they called a posterior hot zone, and faster activity in that zone tracked what the dream contained. By watching the zone in real time, the team could predict whether a person woken from NREM sleep would report a dream or nothing at all⁴. Earlier studies, as their introduction notes, found dream reports after up to 70 percent of NREM awakenings⁴.

How much of the night do we dream, and does everyone dream?

Sleep runs in cycles of about 90 minutes, NREM followed by REM, and REM sleep takes up more of the second half of the night than the first⁵. Because dreaming happens in both states, a sleeper may be dreaming for a large part of the night. Nobody can measure the exact share, since the only evidence of a dream is a report given after waking.

For the same reason it is hard to prove that everyone dreams, but the evidence points that way. People who say they hardly ever dream mostly differ in recall: in a study in Lyon, low recallers remembered about 0.5 dreams a week against 5.2 for high recallers, and their brains differed even at rest¹⁷. There are real exceptions. Some patients with damage to particular forebrain regions stop dreaming entirely¹⁰, and young children dream far less often than adults, as the section on children below shows⁶.

How long do dreams last?

Dreams seem to take about as long as they feel. Dement and Kleitman woke sleepers either 5 or 15 minutes into a REM period and asked which it had been; most judged correctly, and longer REM periods produced longer dream reports². Domhoff describes ordinary dreams as unfolding over a few minutes to half an hour or more⁶. The popular idea that a whole dream flashes past in the second before you wake has little laboratory support.

What happens in the brain when you dream?

Brain scans of REM sleep, when vivid dreams are most likely, show a consistent pattern. In 1996 Pierre Maquet's team in Liège scanned seven people who stayed in steady REM sleep inside a PET scanner and recalled dreams on waking. Blood flow rose in the pons, the thalamus, both amygdalae and the anterior cingulate cortex, and fell across a large area of the dorsolateral prefrontal cortex and parts of the parietal cortex⁷. A year later Allen Braun and colleagues at the US National Institutes of Health scanned 37 men across the whole sleep-wake cycle. In REM sleep, limbic and paralimbic regions, including the insula, cingulate and medial temporal cortex, became active again, while the frontal and parietal association areas that carry out the highest-order integration of information stayed depressed⁸.

Why dreams feel so emotional, and so odd

The amygdala and its neighbours are central to emotion and emotional memory, while the dorsolateral prefrontal cortex supports planning, judgement and the sense that something cannot be right. That combination fits the two features people notice most in dreams: strong feeling, and an uncritical acceptance of scenes that would stop you short when awake. Kieran Fox and colleagues compared numerous imaging studies and found that REM sleep activates much of the default network used in daydreaming while quieting executive prefrontal regions even more deeply than waking mind wandering does. They describe dreaming as an intensified form of mind wandering with little insight into its own state⁹. A lucid dream, in which you know you are dreaming, is the uncommon case where that insight returns.

Dreams are also less bizarre than their reputation. In a 1971 laboratory study of REM dreams summarised by Domhoff, only 8.9 percent of dreams were rated highly improbable by waking standards; most recombined familiar settings, people and activities⁶. The strange dreams are the ones we remember and retell.

Solms and the forebrain

Mark Solms, a neuropsychologist, argued in 2000 that REM sleep and dreaming are controlled by different brain mechanisms. Patients with focal damage along a particular forebrain pathway, which he took to be dopaminergic, stopped dreaming while their REM sleep carried on unchanged. Dreaming could also be triggered in NREM sleep by forebrain seizures, and altered by drugs acting on dopamine without any change in REM sleep¹⁰. In Solms's earlier survey of 361 neurological patients, complete loss of dreaming went with lesions near the junction of the parietal, temporal and occipital lobes, or in the ventromedial region of both frontal lobes⁶. On his account, the brainstem machinery of REM sleep is one of several triggers, and the forebrain does the dreaming.

Why do we dream? The main theories

There are many theories, and each explains some findings better than others. Seven have shaped the field, and they are not all rivals: some are about mechanism, some about function, some only about content.

Activation-synthesis

In 1977 Allan Hobson and Robert McCarley proposed that dreaming is driven by a brainstem mechanism that switches on at regular intervals in REM sleep. Bursts of activity from brainstem circuits, including those for eye movement and balance, reach the forebrain, which synthesises a dream by comparing these signals with information stored in memory; shifts in the balance of brain chemicals might explain why dreams are forgotten¹¹. Its weak points are that dreaming is not tied to REM sleep as tightly as it assumed^4,10, and that dream content is more organised and personal than random activation would predict^13,6.

Memory consolidation

That sleep helps memory is well established; whether dreams play a part is a separate question. In a 2010 study from Robert Stickgold's laboratory at Harvard, Erin Wamsley and colleagues trained 99 people on a virtual maze and retested them five hours later. Half took an afternoon nap in between and the rest stayed awake. Of the 50 nappers, four reported dreams that referred to the maze, and those four improved about ten times more than nappers who did not. Thinking about the maze while awake predicted no improvement at all¹². The authors concluded that dreams may reflect the brain reprocessing recent memories offline. The limits are plain: the dreaming group was tiny, the result is a correlation, and the dreams came from NREM sleep during a nap.

Emotional regulation: overnight therapy

Matthew Walker and Els van der Helm reviewed the research on sleep and emotion in 2009 under the title "Overnight therapy?". In REM sleep the amygdala and hippocampus are active while noradrenaline and serotonin, chemicals tied to arousal and stress, are strongly suppressed. Their model proposes that this lets the brain reprocess emotional experiences so that the memory is kept while its emotional charge fades, leaving next-day reactions better calibrated⁵. It is a model of REM sleep more than of dream content, and it links disturbed sleep to the start and persistence of mood problems.

Threat simulation

Antti Revonsuo, of the University of Turku, argued in 2000 that dreaming evolved to simulate threatening events, so that our ancestors could rehearse perceiving and avoiding danger. He pointed out that dream content is organised and selective, not random, and drew support from typical dream content, children's dreams, recurrent dreams, nightmares, post-traumatic dreams and the dreams of hunter-gatherers¹³. Familiar dreams of being chased or falling fit the picture. Critics reply that many dreams contain no threat, and nobody has shown that dreamed rehearsal improves waking survival.

The continuity hypothesis

The continuity hypothesis holds that dreams mostly carry the same people, activities and concerns as waking life. Domhoff, who has done much of the quantitative work, reports that a majority of dreams draw on concerns about the past, present or future, often dramatised as worst-case scenarios⁶. Even the dreams of blind adults show it: in one study their dreams about getting from place to place contained a misfortune more often than the norms, matching a real daily difficulty, and the authors read this as evidence of continuity¹⁹. The hypothesis describes what dreams contain, not what they are for.

Dreaming as the default network at night

Domhoff's 2011 proposal goes a step further. The brain regions that support dreaming may form a subsystem of the default network, the set of regions active when the mind wanders, daydreams or imagines the future. On this view dreaming is what that network produces when it runs during sleep with enough activation, little input from the senses and the self-monitoring regions turned down. He calls dreaming the quintessential cognitive simulation⁶. Fox's comparison of imaging studies supports the overlap⁹, and so does the way children's dreams develop.

No function: a by-product of sleep

The philosopher Owen Flanagan argued in 1995 that dreams are "spandrels of sleep", borrowing a term from evolutionary biology for features that come along with an adaptation without having been selected for themselves. Sleep has functions and so does consciousness; dreams, on this view, are what a conscious brain produces while asleep, and need no function of their own¹⁴. Flanagan did not conclude that dreams are meaningless, since a by-product can still express what matters to the dreamer. Many cognitive dream researchers take a similar line, treating dreaming as a by-product of waking abilities such as imagery and autobiographical memory⁶.

What the theories share

Set side by side, the accounts overlap more than their debates suggest. Nearly all accept that dreams are built from memory, that emotion is central, and that the dreaming brain runs with its self-critical regions turned down. They differ on whether that activity does something for us, and on whether the content matters or only the process beneath it. Even if dreaming has no evolved purpose, the continuity findings say dreams reflect your concerns, which is what makes them worth attention. How the main schools of interpretation use that content is the subject of the dream interpretation guide.

Why can't I remember my dreams?

Most dreams are gone within minutes of waking, and some people recall many times more than others.

In 2014 Jean-Baptiste Eichenlaub, Perrine Ruby and colleagues in Lyon compared high recallers, who remembered about 5.2 dreams a week, with low recallers, who remembered about 0.5. Using PET scans, they found that high recallers had more activity in the temporoparietal junction during REM sleep, deep NREM sleep and quiet wakefulness, and more activity in the medial prefrontal cortex during REM sleep and wakefulness¹⁷. Both regions belong to the default network, and damage to them can end dream reports altogether. The authors suggest that the extra activity may help produce dreams or lay them down in memory, and that a more active temporoparietal junction may make the sleeper more responsive to the outside world, and so more likely to wake briefly in the night.

Raphael Vallat and the same group then tested that last idea. They played sounds to 18 high and 18 low recallers through a night in the laboratory. High recallers' brains responded more strongly, and they had more long awakenings, though not more brief arousals. The authors proposed that a dream needs an awakening of roughly two minutes to be stored in long-term memory¹⁸. Waking often enough, and long enough, may be part of why some people remember so much.

Personality matters less than people expect. In a study of 444 people, Michael Schredl and colleagues found that openness to experience, creativity, waking during the night and attitude towards dreams were all linked to how often people recalled dreams, yet together they explained only 8.4 percent of the differences between people¹⁶. Much of the usual advice follows from these findings: lie still for a moment on waking, keep your attention on the dream and record it before anything else. The guide on how to remember your dreams sets out the techniques, and a dream journal is the simplest of them.

What do people dream about?

The best-known reference point is the set of norms that Calvin Hall and Robert Van de Castle built from 500 dreams of American college men and 491 of college women, collected around 1950. In that sample, 45 percent of dreams contained at least one act of aggression, 40 percent at least one friendly interaction, 35 percent a misfortune and 8 percent something sexual. About half of all human characters were people the dreamer knew. Of the emotions coded, 80 percent were negative, in men and in women, with apprehension the most frequent¹⁵. Domhoff adds that 73.6 percent of men's and 68.2 percent of women's dreams contained at least one aggression, misfortune or failure, against roughly half with at least one friendly interaction, piece of good fortune or success⁶.

Emotion is common but often mild. In laboratory work summarised by Domhoff, about 70 percent of REM dream reports carried at least some emotion, most of it not very strong, and in 60 percent the emotion fitted the situation in the dream⁶. Some dreams recur across many people and cultures, and each has its own page here with what research says about it: teeth falling out, exams you are not ready for and flying are among the most searched.

Nightmares

A nightmare is a disturbing dream that wakes you or stays vivid after waking. Occasional nightmares sit within the normal range described above. Frequent nightmares that make you dread going to sleep, or that replay a traumatic event, are a reason to talk to a doctor or therapist, because effective treatments exist, among them imagery rehearsal therapy, explained in the guide to nightmares. A dream that keeps returning in the same form is covered under recurring dreams.

Do children, blind people and animals dream?

Children

David Foulkes followed children aged 3 to 16 in the sleep laboratory over several years, waking them thousands of times. Children aged 3 to 5 reported a dream on only 27 percent of REM awakenings, and their dreams were mostly static and bland. Between 5 and 7, dreams gained characters, action and the child's own self, but only at 11 to 13 did they come to resemble adult dreams in frequency, length, emotion and structure. The strongest correlate of this development was the waking ability to form mental images⁶. On this evidence, dreaming is a cognitive achievement that matures with the mind.

People who are blind

People born blind dream, but not in pictures. In a study of 372 dreams from 15 blind adults, those blind since birth or very early childhood had no visual imagery in their dreams and a very high share of taste, smell and touch¹⁹. A Danish study in 2014 collected four weeks of dream reports from 11 congenitally blind, 14 late-blind and 25 sighted adults. All the blind participants had fewer visual impressions than the sighted. The congenitally blind reported more sound, touch, taste and smell; among the late blind, the longer they had been blind, the shorter, less clear and less colourful their visual dream images were. Emotional and thematic content did not differ between the groups, but the congenitally blind reported more nightmares and more aggressive encounters, which the authors related to more threatening experiences in their daily lives²⁰.

Animals

REM-like sleep has been recorded in many mammals and in birds, and animal studies show that patterns of brain activity from recent learning are replayed during sleep¹². This shows that the brain state most linked to vivid dreaming in humans is old and widespread. It cannot show what, if anything, an animal experiences, because dreams are known only through reports, and an animal cannot give one. A dog twitching in its sleep may be dreaming; the evidence cannot yet say.

What this means if you want to work with your own dreams

Dreams draw on your current concerns, so the place to look for their meaning is your own life, not a fixed list of symbols; the comparison of a dream journal and a dream dictionary explains why. Dreams fade within minutes unless you hold on to them, so recording on waking matters more than any technique. And because the dreaming brain amplifies emotion while turning down judgement, a frightening dream is better treated as a feeling to understand than as a warning.

A run of dreams read together tends to say more than any single one, since repeated people, places and moods show which concerns keep returning. Recurring figures are where the Jungian tradition looks for patterns, and Jungian archetypes in dreams explains that approach and where its evidence is thin. None of the theories above supports reading a dream as a prediction of the future.

Read next

• Dream interpretation: what psychology actually knows
• How to remember your dreams
• What a dream journal is and how to keep one
• Lucid dreaming: the basics

Sources

1. Aserinsky E, Kleitman N (1953). Regularly occurring periods of eye motility, and concomitant phenomena, during sleep. Science, 118(3062), 273-274. doi.org/10.1126/science.118.3062.273
2. Dement W, Kleitman N (1957). The relation of eye movements during sleep to dream activity: an objective method for the study of dreaming. Journal of Experimental Psychology, 53(5), 339-346. doi.org/10.1037/h0048189
3. Foulkes WD (1962). Dream reports from different stages of sleep. Journal of Abnormal and Social Psychology, 65(1), 14-25. doi.org/10.1037/h0040431
4. Siclari F, Baird B, Perogamvros L, Bernardi G, LaRocque JJ, Riedner B, Boly M, Postle BR, Tononi G (2017). The neural correlates of dreaming. Nature Neuroscience, 20(6), 872-878. doi.org/10.1038/nn.4545
5. Walker MP, van der Helm E (2009). Overnight therapy? The role of sleep in emotional brain processing. Psychological Bulletin, 135(5), 731-748. doi.org/10.1037/a0016570
6. Domhoff GW (2011). The neural substrate for dreaming: is it a subsystem of the default network? Consciousness and Cognition, 20(4), 1163-1174. doi.org/10.1016/j.concog.2011.03.001
7. Maquet P, Péters J, Aerts J, Delfiore G, Degueldre C, Luxen A, Franck G (1996). Functional neuroanatomy of human rapid-eye-movement sleep and dreaming. Nature, 383(6596), 163-166. doi.org/10.1038/383163a0
8. Braun AR, Balkin TJ, Wesenten NJ, Carson RE, Varga M, Baldwin P, Selbie S, Belenky G, Herscovitch P (1997). Regional cerebral blood flow throughout the sleep-wake cycle: an H2(15)O PET study. Brain, 120(7), 1173-1197. doi.org/10.1093/brain/120.7.1173
9. Fox KCR, Nijeboer S, Solomonova E, Domhoff GW, Christoff K (2013). Dreaming as mind wandering: evidence from functional neuroimaging and first-person content reports. Frontiers in Human Neuroscience, 7, 412. doi.org/10.3389/fnhum.2013.00412
10. Solms M (2000). Dreaming and REM sleep are controlled by different brain mechanisms. Behavioral and Brain Sciences, 23(6), 843-850. doi.org/10.1017/S0140525X00003988
11. Hobson JA, McCarley RW (1977). The brain as a dream state generator: an activation-synthesis hypothesis of the dream process. American Journal of Psychiatry, 134(12), 1335-1348. doi.org/10.1176/ajp.134.12.1335
12. Wamsley EJ, Tucker M, Payne JD, Benavides JA, Stickgold R (2010). Dreaming of a learning task is associated with enhanced sleep-dependent memory consolidation. Current Biology, 20(9), 850-855. doi.org/10.1016/j.cub.2010.03.027
13. Revonsuo A (2000). The reinterpretation of dreams: an evolutionary hypothesis of the function of dreaming. Behavioral and Brain Sciences, 23(6), 877-901. doi.org/10.1017/S0140525X00004015
14. Flanagan O (1995). Deconstructing dreams: the spandrels of sleep. Journal of Philosophy, 92(1), 5-27. doi.org/10.2307/2940806
15. Hall CS, Van de Castle RL (1966). The Content Analysis of Dreams. New York: Appleton-Century-Crofts. Normative tables as published by DreamResearch.net, University of California, Santa Cruz. dreams.ucsc.edu/Norms/
16. Schredl M, Wittmann L, Ciric P, Götz S (2003). Factors of home dream recall: a structural equation model. Journal of Sleep Research, 12(2), 133-141. doi.org/10.1046/j.1365-2869.2003.00344.x
17. Eichenlaub JB, Nicolas A, Daltrozzo J, Redouté J, Costes N, Ruby P (2014). Resting brain activity varies with dream recall frequency between subjects. Neuropsychopharmacology, 39(7), 1594-1602. doi.org/10.1038/npp.2014.6
18. Vallat R, Lajnef T, Eichenlaub JB, Berthomier C, Jerbi K, Morlet D, Ruby PM (2017). Increased evoked potentials to arousing auditory stimuli during sleep: implication for the understanding of dream recall. Frontiers in Human Neuroscience, 11, 132. doi.org/10.3389/fnhum.2017.00132
19. Hurovitz CS, Dunn S, Domhoff GW, Fiss H (1999). The dreams of blind men and women: a replication and extension of previous findings. Dreaming, 9(2-3), 183-193. doi.org/10.1023/A:1021397817164
20. Meaidi A, Jennum P, Ptito M, Kupers R (2014). The sensory construction of dreams and nightmare frequency in congenitally blind and late blind individuals. Sleep Medicine, 15(5), 586-595. doi.org/10.1016/j.sleep.2013.12.008