Papex

Open problem · Neuroscience · Open

Function of dreaming

Why do we dream?

Summary

Dreams have been linked to REM sleep since the 1950s, and proposed roles include learning, emotional processing and rehearsing responses to threats, but there is no agreed account of what dreaming is for, or whether it serves a purpose at all.

Why it is open

Dreaming has traditionally been identified with rapid-eye-movement (REM) sleep, a stage with wake-like brain activity, but it also occurs in non-REM (NREM) sleep [10]. Proposals include threat rehearsal [2], emotional processing [7], memory consolidation and generalisation; Erik Hoel notes that dreams are also treated as by-products of sleep [12].

Three features of the evidence keep the question open. Most evidence depends on waking reports, vulnerable to forgetting and distortion [10, 13]. Much of the physiological evidence concerns REM sleep rather than dreaming, and the two can come apart: Mark Solms argued from patients with brain lesions that focal forebrain damage can abolish dreaming without appreciably changing REM sleep [3]. Karen Konkoly and colleagues emphasise that correlational studies cannot establish a causal function [15]. Neural replay of waking experience can be recorded in sleeping rats [4], but rats cannot report dreams, and a 2023 review judged such memory reactivations "not likely to be the primary neural basis of dreaming" [14].

The sources point to what would help: methods that change dream content directly and then measure a waking effect, which Konkoly and colleagues say are still needed [15], and measures that do not rest only on recall after waking, such as the EEG marker and the real-time communication described below [10, 13]. Whether sleep itself has a core function is a separate Papex problem, the function of sleep.

What is known

Where dreams come from (1977–2017). In their 1977 activation-synthesis hypothesis, J. Allan Hobson and Robert W. McCarley proposed that the forebrain constructs dreams from brainstem activation and stored memories [1]. In 2000 Solms used lesion and pharmacological evidence to argue that dreaming and REM sleep have distinct mechanisms [3]. In 2017 Francesca Siclari and colleagues woke 39 people repeatedly through the night while recording high-density EEG, in two experiments with 1,048 awakenings [10]. In both NREM and REM sleep, reports of dreaming were preceded by less low-frequency activity in a posterior "hot zone" of the cortex, and the location of high-frequency activity tracked contents such as faces [10]. In 7 further participants, monitoring the zone in real time correctly predicted a dream report in 33 of 36 awakenings and its absence in 21 of 26 [10].

Threat simulation (2000–2008). Antti Revonsuo proposed that dreams evolved to rehearse detection and avoidance of threats [2]. In a test using dream reports, severely traumatised Kurdish children reported more dreams, with more threatening events and more severe threats, than less traumatised Kurdish or non-traumatised Finnish children [5]. In a test of the same prediction, participants from a high-crime area of South Africa (208) had more exposure to life-threatening events than participants from Wales (116) but reported fewer threat dreams; fewer than 20% of all dreams contained realistic survival threats, and escape occurred in fewer than 2% [6].

Memory (2001–2023). In rats, Kenway Louie and Matthew Wilson found sequences of hippocampal activity from waking behaviour, lasting tens of seconds to minutes, replayed during REM sleep [4]. In people, Erin Wamsley and colleagues trained participants on a virtual maze; the 4 of 50 nappers (8.0%) who mentioned the maze in their dream reports ranked among those who improved most [8]. The authors did not claim that the dream caused the improvement; they proposed that dream content and improvement both reflect memory processing during sleep [8]. A 2023 review found "both strong parallels and disparities" between memory reactivation and dreaming [14].

Emotion (2009–2020). Matthew Walker and Els van der Helm proposed that REM sleep reprocesses emotional memories, keeping their content while reducing their emotional charge [7]. In a human sleep-and-wake comparison, the overnight fall in amygdala responses to emotional pictures seen before correlated, among those who slept, with lower prefrontal gamma activity during REM sleep, which the authors take as a marker of reduced adrenergic activity [9]. Virginie Sterpenich and colleagues found that fear in dreams engaged the insula and midcingulate cortex (18 participants woken repeatedly), and that people who reported more fear in their dreams responded less to frightening pictures when awake (89 participants) [11].

Recent results. Konkoly and colleagues showed in 2021 that some people in a lucid dream, aware that they are dreaming, can answer an experimenter's questions with eye movements and facial muscle signals during REM sleep: correct answers came on 29 occasions from 6 of 36 people, in four laboratories [13]. In February 2026 Konkoly and colleagues reported 20 participants, all but one of them frequent lucid dreamers, who slept after failing to solve puzzles that each had a soundtrack [15]. Playing half of the soundtracks during REM sleep "reliably increased dreaming about the associated puzzles", but cueing did not raise next-morning solving overall; puzzles that appeared in dreams showed a non-significant trend towards being solved more often, and in a post-hoc analysis of the 12 participants whose dreams followed the cues, cued puzzles were solved more often [15]. The authors note that processing between the dream and the test, and demand characteristics, could not be ruled out [15].

Main approaches

By-product. Dreaming is a by-product of sleep physiology with no function of its own, a view Revonsuo attributes to several theories and Hoel describes as common [2, 12]. Strongest evidence: where dream content predicts a later benefit, the data fit a shared cause as well as an effect of the dream, as Wamsley and colleagues note for their own results [8], and in the puzzle study steering dreams with sound cues did not raise solving overall [15]. Main difficulty: Revonsuo argues that dream content is "organized and selective" rather than random [2], and Solms's evidence that dreaming depends on forebrain rather than brainstem REM mechanisms challenges the activation-synthesis account [1, 3].

Memory consolidation. Dreams reflect, and might aid, the reactivation of recent memories during sleep [8, 14]. Strongest evidence: task-related dreams predicted improvement in one human study [8], and replay during REM sleep in rats [4]. Main difficulty: the human study is correlational and rests on 4 dreamers [8]; the animal evidence concerns neural replay, not reported dreams [4, 14].

Threat simulation. Dreams rehearse threat perception and avoidance [2]. Strongest evidence: more frequent and more severe threats in the dreams of severely traumatised children, from dream reports compared between groups [5]. Main difficulty: the South African comparison, also based on dream reports, found fewer threat dreams despite more exposure to threat [6].

Emotional regulation. REM sleep, and perhaps the dreams within it, reduces the emotional charge of recent experiences [7, 11]. Strongest evidence: the correlation between REM gamma activity and later emotional reactivity [9], and lower waking fear responses in people who report more fear in their dreams [11]; both are correlational, and the second rests on dream reports. Main difficulty: the main experiment measured REM physiology, not dream content [9], and Hoel considers the evidence for dreams resolving emotional conflicts limited [12].

Generalisation. In Hoel's overfitted brain hypothesis, the strange, sparse content of dreams works like the noise injected into artificial neural networks to stop them memorising their training data instead of generalising [12]. Strongest evidence: the hypothesis draws on regularisation in artificial networks and features of dream reports [12]. Main difficulty: it rests on an analogy so far; Hoel wrote in 2021 that no animal study had separated the effects of sleep loss on memorisation and on generalisation, a test he proposed [12].

Open sub-questions

  1. Steering dreams towards a task. If cues during sleep increase task-related dream content, as in the puzzle study [15], does that increase predict next-day improvement on a hippocampus-dependent task beyond the cueing itself [8]?
  2. Threat dreams and behaviour. Do people who report more threat dreams respond faster or more accurately to threats in a behavioural test the next day, beyond the reduced fear responses measured by imaging [6, 11]?
  3. Dreams or REM. When NREM dreaming is detected with the posterior hot-zone marker [10], does the overnight change in amygdala reactivity follow dreaming or REM gamma activity [9]?
  4. Memorisation versus generalisation. In mice, does sleep deprivation impair the generalisation of a learned fear context more than its memorisation, as the overfitted brain hypothesis predicts, and does selective REM deprivation do so more than NREM deprivation [12]? Such a test would concern sleep stages, not reported dreams.
  5. Rehearsal inside a dream. Using real-time communication with lucid dreamers [13], does practising a task within a verified dream improve it as much as the same practice while awake?
  6. Replay and reports. How often does decoded memory reactivation during human sleep coincide with reported dream content about the same memory [4, 14]?

Work on this in Papex

If you have a testable answer to one of the sub-questions above, you can draft it as a private hypothesis in Papex, or prepare the draft from Claude or ChatGPT with the Papex connector. You review the exact version, sign it and decide whether to publish it; a public record that names this problem is listed on this page. A Papex record shows who signed which version and when. It is not peer review and does not mean the hypothesis is correct.

Key references

  1. Hobson, J. A. & McCarley, R. W. (1977). The brain as a dream state generator: an activation-synthesis hypothesis of the dream process. American Journal of Psychiatry 134(12), 1335–1348. 10.1176/ajp.134.12.1335 · Checked 15 Oct 2026
  2. Revonsuo, A. (2000). The reinterpretation of dreams: an evolutionary hypothesis of the function of dreaming. Behavioral and Brain Sciences 23(6), 877–901. 10.1017/S0140525X00004015 · Checked 15 Oct 2026
  3. Solms, M. (2000). Dreaming and REM sleep are controlled by different brain mechanisms. Behavioral and Brain Sciences 23(6), 843–850. 10.1017/S0140525X00003988 · Checked 15 Oct 2026
  4. Louie, K. & Wilson, M. A. (2001). Temporally structured replay of awake hippocampal ensemble activity during rapid eye movement sleep. Neuron 29(1), 145–156. 10.1016/S0896-6273(01)00186-6 · Checked 15 Oct 2026
  5. Valli, K., Revonsuo, A., Pälkäs, O., Ismail, K. H., Ali, K. J. & Punamäki, R.-L. (2005). The threat simulation theory of the evolutionary function of dreaming: evidence from dreams of traumatized children. Consciousness and Cognition 14(1), 188–218. 10.1016/S1053-8100(03)00019-9 · Checked 15 Oct 2026
  6. Malcolm-Smith, S., Solms, M., Turnbull, O. & Tredoux, C. (2008). Threat in dreams: an adaptation? Consciousness and Cognition 17(4), 1281–1291. 10.1016/j.concog.2007.07.002 · Checked 15 Oct 2026
  7. Walker, M. P. & van der Helm, E. (2009). Overnight therapy? The role of sleep in emotional brain processing. Psychological Bulletin 135(5), 731–748. 10.1037/a0016570 · Checked 15 Oct 2026
  8. Wamsley, E. J., Tucker, M., Payne, J. D., Benavides, J. A. & Stickgold, R. (2010). Dreaming of a learning task is associated with enhanced sleep-dependent memory consolidation. Current Biology 20(9), 850–855. 10.1016/j.cub.2010.03.027 · Checked 15 Oct 2026
  9. van der Helm, E., Yao, J., Dutt, S., Rao, V., Saletin, J. M. & Walker, M. P. (2011). REM sleep depotentiates amygdala activity to previous emotional experiences. Current Biology 21(23), 2029–2032. 10.1016/j.cub.2011.10.052 · Checked 15 Oct 2026
  10. Siclari, F., Baird, B., Perogamvros, L., Bernardi, G., LaRocque, J. J., Riedner, B., Boly, M., Postle, B. R. & Tononi, G. (2017). The neural correlates of dreaming. Nature Neuroscience 20(6), 872–878. 10.1038/nn.4545 · Checked 15 Oct 2026
  11. Sterpenich, V., Perogamvros, L., Tononi, G. & Schwartz, S. (2020). Fear in dreams and in wakefulness: evidence for day/night affective homeostasis. Human Brain Mapping 41(3), 840–850 (published online 2019). 10.1002/hbm.24843 · Checked 15 Oct 2026
  12. Hoel, E. (2021). The overfitted brain: dreams evolved to assist generalization. Patterns 2(5), 100244. 10.1016/j.patter.2021.100244 · Checked 15 Oct 2026
  13. Konkoly, K. R., Appel, K., Chabani, E., Mangiaruga, A., Gott, J., Mallett, R. et al. (2021). Real-time dialogue between experimenters and dreamers during REM sleep. Current Biology 31(7), 1417–1427.e6. 10.1016/j.cub.2021.01.026 · Checked 15 Oct 2026
  14. Picard-Deland, C., Bernardi, G., Genzel, L., Dresler, M. & Schoch, S. F. (2023). Memory reactivations during sleep: a neural basis of dream experiences? Trends in Cognitive Sciences 27(6), 568–582. 10.1016/j.tics.2023.02.006 · Checked 15 Oct 2026
  15. Konkoly, K. R., Morris, D. J., Hurka, K., Martinez, A. M., Sanders, K. E. G. & Paller, K. A. (2026). Creative problem-solving after experimentally provoking dreams of unsolved puzzles during REM sleep. Neuroscience of Consciousness 2026(1), niaf067 (published online 5 February 2026). 10.1093/nc/niaf067 · Checked 15 Oct 2026

Page reviewed 15 Oct 2026

Source

Open, as listed by Wikipedia · Unsolved problems in neuroscience. Checked 08 Oct 2026.

The status is the source’s. Papex does not decide whether a problem is solved.

Records on this problem

No public record addresses this problem yet.

More in Life sciences