Why we dream and what sleep research says about the purpose of dreams
Researchers do not yet know why dreams happen, although sleeping brains reactivate memories and engage systems involved in emotion and motivation (Zhao et al., 2018; National Institute of Child Health and Human Development, 2019; Perogamvros et al., 2013). Dreams can accompany these processes without necessarily causing their effects. Their content gives researchers a record of mental experience during sleep, but a report after awakening is an imperfect measurement (Konkoly et al., 2021).
Sleep laboratories compare dream reports with brain recordings, learning tasks and assessments made after awakening (Wamsley et al., 2010; Siclari et al., 2017). The findings connect particular dreams with recent experiences, emotional concerns and subsequent memory performance (Wamsley, 2022; Hudachek and Wamsley, 2023; Barbeau et al., 2022). They also expose a central uncertainty: activity that accompanies a dream may be responsible for an observed result even if the conscious dream is not (Perogamvros et al., 2013).
Why do we dream according to sleep science
Dreaming is a form of conscious experience during sleep that researchers usually investigate through what sleepers report when awakened (Siclari et al., 2017). Dream imagery can draw on remembered events, anticipated situations and sensations beyond vision, while individual reported experiences may combine fragments rather than reproduce one waking episode exactly (Ilic et al., 2023; Wamsley, 2022). Scientists have linked it with activity in the sleeping brain, yet a precise biological purpose has not been established (Zhao et al., 2018).
An explanation of why dreams arise is different from an explanation of why they might be useful. Brain recordings can identify neural activity associated with a reported experience (Siclari et al., 2017). They do not establish that producing that experience improves memory, protects emotional health or serves a particular evolutionary purpose (Hudachek and Wamsley, 2023). The National Institute of Child Health and Human Development (2019) likewise identifies the purpose of dreaming as uncertain.
The brain remains engaged with information while asleep, and Perogamvros and colleagues (2013) described activity in emotional and reward systems alongside a proposal that salient memories may undergo further processing. Their account concerns sleep and dreaming together, leaving open whether the conscious experience itself is necessary. A dream can therefore reveal material the sleeping brain is handling without demonstrating why that handling occurs.
Dream research faces a direct measurement problem: investigators usually learn about dream experience after it has ended. Konkoly and colleagues (2021) describe retrospective accounts as vulnerable to forgetting and distortion. The wider issue resembles [errors in self-reported participation](/research/accuracy-of-self-reported-participation-rates-research), where recalled activity and independently recorded behaviour may differ. The two subjects are distinct, but in both cases the reported account is itself a measurement rather than a complete record of the original event.
The main theories about the purpose of dreams
Competing theories assign different possible functions to dreaming. Some connect its contents with memory processing, while others emphasise emotional experience or the construction of threatening scenarios (Zhao et al., 2018) (Perogamvros et al., 2013). The activation account begins with neural signals during sleep and describes how the forebrain may organise them into an experience. The brainstem account describes activation; threat simulation predicts subsequent defensive performance.
Memory consolidation and learning
The memory account links dreams with the reactivation and reorganisation of experiences, including the possible alteration of previously formed memories during their reactivation as discussed by Zhao and colleagues (2018). Their proposal does not establish that remembering a dream is required for either consolidation or reconsolidation. Laboratory results instead show associations between dream content and what participants learned before sleeping (Wamsley et al., 2010; Hudachek and Wamsley, 2023).
Exact reproduction is unnecessary. Wamsley (2022) examined whether fragments from different waking episodes appear together in reported dreams (Wamsley, 2022). Combining elements from past situations and possible future events is consistent with constructive simulation, but the content analysis did not test whether such combinations improved subsequent decisions. The existence of recombination is therefore a different finding from a demonstrated benefit of recombination.
Emotional processing
Emotional theories concentrate on the activation of circuits associated with motivation and feeling. Perogamvros and colleagues (2013) proposed that REM and non-REM sleep might selectively process emotionally significant experiences. Activity in the amygdala and reward-related systems fits that account, although their review presents improved waking emotion regulation as a possibility. It does not isolate an independent therapeutic function of a remembered dream.
Daily dream reports provide another way to investigate the connection. Barbeau and colleagues (2022) examined ratings of dreams, mood and stress and reported some support for a relationship with mood regulation. Associations between emotional dream content and a person's mood can arise without showing that the dream changed that mood. In the study, the strongest predictors of morning mood and stress were the corresponding presleep states.
Threat simulation and activation-synthesis
Threat simulation theory proposes that dreams supply imagined settings in which threatening situations can be experienced. Zhao and colleagues (2018) describe the idea as virtual rehearsal of avoidance or defensive responses. The hypothesis would require evidence that such nocturnal experiences produce a measurable improvement during later waking threats. Describing an alarming dream alone does not measure that proposed improvement.
Activation-synthesis begins from a different direction. In the account reviewed by Zhao and colleagues (2018), stimulation from the brainstem reaches forebrain systems that can construct visual and movement-related experiences. It offers a mechanism for dream-like imagery without requiring that every image fulfil a particular psychological need. The two theories address different questions, so evidence of brain activation does not by itself verify rehearsal of threats.
When dreams occur and how the sleeping brain generates them
Human sleep alternates between rapid eye movement sleep, usually called REM, and non-REM sleep. Dreams occur during both phases (Siclari et al., 2017). The National Institute of Child Health and Human Development (2019) describes dreams as generally most vivid during REM sleep while recognising that they occur during non-REM sleep.
REM and non-REM dreaming
REM activity was historically associated so closely with dreams that the two were sometimes treated as interchangeable, although Siclari and colleagues (2017) studied brain activity and dream reports across both REM and non-REM sleep. Their findings support the presence of subjective experiences in either state. Stage classification describes the physiological context of an experience, while an awakening report supplies the evidence that a particular experience occurred.
Picard-Deland and colleagues (2023) used repeated awakenings to examine how reported dream material varied through the night. Recent memory sources appeared more frequently in reports from N1 and REM sleep than from other stages. The relative contribution of recent and distant sources also shifted across the night, independently of stage. Dream content therefore varies with both the sleep state and the timing of collection.
An overnight learning experiment by Wamsley and Stickgold (2019) similarly identified task-related material in reports from different sleep phases (Wamsley and Stickgold, 2019). Finding recently learned material outside REM challenges the assumption that the incorporation of waking experiences into dreams requires REM specifically. It also complicates any theory that equates the entire function of dreaming with a process unique to REM sleep.
Brain activity linked to dream experiences
Siclari and colleagues (2017) compared EEG recordings with reports of whether a dream had been experienced. In both REM and non-REM sleep, dream reports correlated with lower low-frequency activity in posterior cortical regions. This associates a measurable pattern of brain activity with subjective experience rather than identifying the reason a dream is generated. The finding is strongest as evidence about the neural conditions accompanying dreaming.
The distinction between having a dream and later describing its contents is consequential. Siclari and colleagues (2017) investigated reports of experience and non-experience after awakening. A person who cannot supply a coherent narrative may have forgotten an experience rather than having had no experience at all. Picard-Deland and colleagues (2023) therefore distinguished absence of recalled content from a reported impression that dreaming had occurred.
Brain activity during sleep also supports processes that investigators do not define through dreams. The NICHD (2019) describes both major sleep phases as contributing to learning and memory (National Institute of Child Health and Human Development, 2019). Evidence that one sleep phase changes later performance cannot therefore establish a benefit from conscious dreaming in that phase. The distinction requires experiments capable of measuring the experience and the accompanying sleep physiology separately.
Why dreams can be strange, recurring or hard to remember
Dreams frequently draw upon waking experiences without presenting them as continuous, accurate replays (Wamsley, 2022). An imagined scene can combine memory fragments from several waking occasions, and the participant may identify elements of an invented setting as belonging to multiple past experiences. Reports gathered after sleep may also omit material that was experienced but not retained long enough to describe (Konkoly et al., 2021) (Picard-Deland et al., 2023).
Dream content and waking experiences
Wamsley (2022) asked 48 college students to identify waking-life sources for 469 reported dreams. Multiple waking sources were identified in 43.9% of dreams with content. Participants connected dreams with at least one past episodic source in 53.5% of dreams and with at least one anticipated event in 25.7%. The sources were identified by participants from their own recollections.
The combination of remembered and anticipated experiences gives dream narratives their apparent discontinuity. Wamsley (2022) connects that pattern with constructive episodic simulation, in which elements from different episodes contribute to imagined scenarios. Because the sources were identified retrospectively, a participant might also fail to recognise a genuine connection or infer one after waking. The method observes perceived links rather than directly tracking individual memories through the brain.
Picard-Deland and colleagues (2023) collected dream reports through awakenings across the night. Their results suggest that the age of material appearing in dreams depends partly on when the report is collected. A single morning account may consequently represent a narrower sample of that night's experiences than reports gathered repeatedly during sleep.
Nightmares and recurring dreams
Threat simulation theories address dreams with frightening scenarios by suggesting that they might provide imagined encounters with danger (Zhao et al., 2018). The proposal is a functional hypothesis, not evidence that a nightmare teaches effective avoidance. Emotional processing theories likewise associate dream content with salient concerns and motivational systems (Perogamvros et al., 2013). Neither account, on its own, identifies why a particular frightening theme would recur.
Barbeau and colleagues (2022) examined emotional ratings in ordinary dream reports rather than experimentally inducing recurring nightmares. Their findings concern associations between dreams and waking mood, and cannot establish a mechanism for a recurring clinical symptom. A pattern of negative dream content can be described without assuming that the sleeper is successfully processing a trauma or rehearsing a future threat.
Dream recall after awakening
Memory for a dream is vulnerable at the point of collection. Konkoly and colleagues (2021) identify forgetting and distortion as obstacles to studying dreams through retrospective reports. The words in a report describe what survived until awakening, not necessarily everything experienced during sleep. Their recollections can omit part of the original experience.
Picard-Deland and colleagues (2023) separated a report of no recalled dream from an impression of dreaming with no retrievable details. That distinction leaves room for experience without narrative recall. It also prevents researchers from treating every blank report as proof of an absence of dreaming. The number of dreams someone can recount therefore need not be the number experienced.
What studies reveal about dreams, memory and emotions
Experiments can compare dreams containing a recently learned task with performance on that task after sleep. Other studies examine emotional ratings before and after a night's sleep, or attempt communication with someone who is still dreaming (Konkoly et al., 2021), (Wamsley et al., 2010), (Barbeau et al., 2022). Each method answers a different question. A correlation with improved memory, a mood association and a successful exchange during a dream are separate observations.
Dream reports and learning experiments
Wamsley and colleagues (2010) trained 99 participants on a virtual navigation task and tested them again 5 hours later (Wamsley et al., 2010). Better performance at retest was associated with task-related dreaming during an intervening nap. Thoughts about the task during wakefulness did not predict the same improvement. The finding connects particular dream content with sleep-dependent memory processing without establishing that the conscious experience caused the improvement.
In a later overnight study, Wamsley and Stickgold (2019) reproduced the association between dreaming about a spatial task and memory performance (Wamsley and Stickgold, 2019). Reports included relevant imagery across sleep phases. The replication strengthens the observation that recently trained material can appear in dream experience during sleep. It does not remove the possibility that neural reactivation affects both dream content and subsequent task performance.
Hudachek and Wamsley (2023) combined sixteen qualifying studies reporting 45 effects in a meta-analysis. They found a statistically significant association between dreaming about a learning task and memory outcomes. In polysomnographic studies, the relationship was statistically significant for non-REM dream samples, while the relationship was not statistically significant for REM samples. The synthesis supports dream content as a possible marker of memory processing.
Dreams and emotional memory
Perogamvros and colleagues (2013) proposed that emotional and motivational circuits could select experiences for offline processing during sleep. This mechanism offers a possible connection between emotionally charged dreams and memory organisation. Its predicted benefits include waking emotion regulation, but those outcomes were advanced as hypotheses in the review. An activated circuit does not by itself reveal what its conscious experience accomplishes.
Barbeau and colleagues (2022) collected 345 dream reports from 188 participants who also rated stress and mood. Their results offered limited observational support for an association between dreaming and mood regulation. The strongest predictors of morning mood and stress were the corresponding measures taken before sleep.
[Learning and memory experiments in chickens](/animals/chicken-cognition-learning-memory-social-behaviour-and-navigation-animals) provide a related example of cognition measured through task performance. Dream experiments have the additional burden of collecting the experience itself, usually through a later verbal account (Konkoly et al., 2021). The memory studies therefore address human dream reports in their particular laboratory settings rather than a universal explanation of how all minds learn.
Separating dreaming from sleep effects
Konkoly and colleagues (2021) tested communication during verified REM sleep in 36 individuals. Correct answers were registered on 29 occasions across 6 participants, conveyed through eye movements and facial muscle responses. The experiment demonstrated that limited information could be exchanged with some sleeping dreamers. It did not test whether their dreams improved memory or regulated emotion.
This form of real-time interaction can provide evidence about dream cognition before a later waking account is affected by the forgetting and distortion that complicate retrospective reporting (Konkoly et al., 2021). Its successful use is still a narrow demonstration of communication during certain lucid dreams, rather than evidence that any theory of dream purpose is correct. Most studies of ordinary dreaming continue to face difficulty separating a person's conscious experience from the brain processes taking place at the same time.
Study | Observations | Finding | Interpretative boundary |
Wamsley et al., 2010 | 99 participants, virtual navigation task | Task dreams associated with better retest performance | Association during a nap, not proof of causation |
Siclari et al., 2017 | EEG and dream reports from REM and non-REM | Posterior cortical activity correlated with dreaming | Neural correlate, not evolutionary function |
Wamsley, 2022 | 48 students, 469 dreams | Multiple waking sources in 43.9% of dreams with content | Participants identified their own sources |
Hudachek and Wamsley, 2023 | sixteen studies, 45 effects | Significant dream-content and memory association | Dream content may indicate processing |
Konkoly et al., 2021 | 36 individuals | 29 correct answers across 6 participants | Communication, not a test of dream purpose |
What is proven about why we dream and what is not
Dreaming occurs in both REM and non-REM sleep, and particular patterns of cortical activity are associated with reports of having dreamed (Siclari et al., 2017). Dreams can incorporate fragments of remembered experience, and reports of recent learning can be associated with subsequent task performance (Wamsley, 2022; Hudachek and Wamsley, 2023). Neither observation establishes a single biological purpose for conscious dreaming (Zhao et al., 2018).
Dream content can reveal memories and emotions being processed during sleep. Whether experiencing the dream itself produces the measured benefits remains unproven.
Laboratory conditions define each relationship. Siclari and colleagues (2017) linked posterior cortical activity with reported experience. Wamsley and colleagues (2010) linked task imagery during a nap with later performance on virtual navigation (Wamsley et al., 2010). These results concern different outcomes, and neither provides a direct test of all proposed functions of dreams.
Dream-memory associations have since been examined across studies. Hudachek and Wamsley (2023) found a statistically significant pooled relationship, with a statistically significant result for NREM and a non-significant result for REM samples. The authors describe dream content as a potential indicator of consolidation. Dream content may indicate memory consolidation even if the remembered dream is not the process responsible for an observed change in learning-task performance.
The emotional explanation remains less firmly established as a dream-specific effect. Perogamvros and colleagues (2013) presented a possible pathway through emotional and motivational circuits. Barbeau and colleagues (2022) reported associations with mood, while identifying presleep mood and stress as strong predictors of morning states. Neither study removes the wider influence of sleep or prior emotional state.
The measurement of dreams itself is still incomplete. Laboratory awakenings can collect reports from different sleep phases, but the account remains susceptible to forgetting (Konkoly et al., 2021; Picard-Deland et al., 2023). Interactive lucid-dream experiments show that communication is possible under selected conditions, although those conditions do not establish why ordinary dreams occur. Research that measures brain activity, dream content and later outcomes while distinguishing their separate contributions would directly test the remaining causal question.
Frequently asked questions
Do we dream every night?
Dreaming occurs during REM and non-REM sleep (National Institute of Child Health and Human Development, 2019; Siclari et al., 2017). Studies distinguish reports of no experience from a lingering impression of dreaming without remembered content (Picard-Deland et al., 2023). Forgetting after awakening makes nightly dream frequency difficult to determine solely from morning recollection (Konkoly et al., 2021).
How long do dreams last?
Dream reports are collected after awakenings from different sleep phases, and their content varies with the point in the night when they are collected (Picard-Deland et al., 2023). A remembered narrative does not provide a direct measure of the duration of the original experience. The research considered here does not establish one reliable duration for every dream.
Do dreams have meaning?
Dreams may incorporate recognisable fragments of waking memories and anticipated events (Wamsley, 2022). In Wamsley's 2022 study, participants identified multiple waking sources in some dream reports. Multiple waking sources were identified in 43.9% of dreams with content. The study measured reported links to waking events, rather than universal symbolic meanings.
Do blind people dream?
Yes. Research reviewed by Ilic and colleagues (2023) describes dreams among congenitally and early blind people involving auditory, tactile, bodily and other nonvisual sensations. Whether people blind from birth experience visual imagery in dreams remains disputed. Dream experience is therefore not defined solely by the presence of visual scenes.
What is lucid dreaming?
Lucid dreaming involves awareness of dreaming while the experience is taking place. In experiments reported by Konkoly and colleagues (2021), some people in verified REM sleep responded to questions through eye movements or facial signals. This establishes limited communication during selected lucid dreams, rather than proving that all dreams are controllable.
Sources
- Zhao H, Li D, Li X. Relationship between Dreaming and Memory Reconsolidation. Brain Science Advances. 2018;4(2):118–130. DOI 10.26599/BSA.2018.9050005.
- Perogamvros L, Dang-Vu TT, Desseilles M, Schwartz S. Sleep and dreaming are for important matters. Frontiers in Psychology. 2013;4:474. DOI 10.3389/fpsyg.2013.00474.
- 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. DOI 10.1016/j.cub.2010.03.027.
- Siclari F, Baird B, Perogamvros L, et al. The neural correlates of dreaming. Nature Neuroscience. 2017;20(6):872–878. DOI 10.1038/nn.4545.
- 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. DOI 10.1111/jsr.12749.
- Wamsley, Erin J. Constructive episodic simulation in dreams. PLOS ONE. 2022;17(3):e0264574. DOI 10.1371/journal.pone.0264574.
- Hudachek L, Wamsley EJ. A meta-analysis of the relation between dream content and memory consolidation. Sleep. 2023;46(12):zsad111. DOI 10.1093/sleep/zsad111.
- Picard-Deland C, Konkoly K, Raider R, et al. The memory sources of dreams: serial awakenings across sleep stages and time of night. Sleep. 2023;46(4):zsac292. DOI 10.1093/sleep/zsac292.
- Barbeau K, Turpin C, Lafrenière A, Campbell E, De Koninck J. Dreamers’ evaluation of the emotional valence of their day-to-day dreams is indicative of some mood regulation function. Frontiers in Behavioral Neuroscience. 2022;16:947396. DOI 10.3389/fnbeh.2022.947396.
- 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.e6. DOI 10.1016/j.cub.2021.01.026.
- Ilic K, Bertani R, Lapteva N, et al. Visuo-spatial imagery in dreams of congenitally and early blind: a systematic review. Frontiers in Integrative Neuroscience. 2023;17:1204129. DOI 10.3389/fnint.2023.1204129.
- National Institute of Child Health and Human Development. What happens during sleep? Last reviewed 29 April 2019. https://www.nichd.nih.gov/health/topics/sleep/conditioninfo/what-happens
Related
How Accurate Are Self-Reported Recreation Participation Rates?
Chicken cognition: learning, memory, social behaviour, navigation
Travel in Later Life: What an Eight-Person Study Found
Circadian rhythm and how your body clock shapes sleep and alertness
What déjà vu reveals about how the brain processes memories
What causes hiccups and why your body has this reflex
How long you can survive without food and what happens to your body