Dreams, creativity and problem solving: what experiments show

Inspiration, insight and creative ratings are different outcomes

An image that interests you, the correct answer to a puzzle and a story rated as original are all worth discussing. They are not the same outcome. Inspiration describes what gives you a starting point. Task insight concerns finding a solution or a useful rule. A creativity rating depends on what judges were asked to value.

As an invented example, a dream of a staircase might prompt a drawing. That establishes a personal connection between the dream and the drawing. It does not show that the drawing is more original than one you would have made after an ordinary break. A comparison is needed for that stronger claim.

The sleep-onset result concerned a hidden numerical rule

Lacaux and colleagues tested 103 participants in 2021. After excluding 16 who found the rule before a rest period, they grouped the remaining participants by their recorded state: wakefulness, the first sleep stage (N1), or the next stage (N2). These were observed sleep trajectories, not randomly assigned stages.

The hidden rule was later found by 20 of 24 in the N1 group, 15 of 49 who stayed awake and two of 14 who reached N2. Insight generally emerged during further task practice, not immediately on waking. Reported dreamlike experiences at sleep onset, including task-related ones, were not significantly associated with insight. Participants were selected for easy sleep onset and had reduced sleep beforehand. These conditions and this one task limit generalisation; the result does not establish that a dream supplied the answer.

A later comparison found a different sleep-stage pattern

Löwe and colleagues' 2025 study used a perceptual task: participants could discover that colour offered an easier rule than judging motion. Its sleep analysis included 68 participants after exclusions. The preregistered N1-over-wake hypothesis was not supported. An exploratory comparison found more insight after N2 than wakefulness.

This was a conceptual replication using a different task, not an exact repeat of the numerical experiment. Sleep groups again reflected what happened during rest. The authors note the absence of random assignment to sleep, waking rest and no rest. The disagreement limits a universal N1 claim without proving that the earlier result was false.

Dream incubation tested creativity tied to one theme

Haar Horowitz and colleagues randomly assigned participants to sleep or wake conditions, with or without prompts about a tree. After one exclusion, 49 remained. Participants then completed storytelling, alternative-use and word-association tasks related to the theme.

The sleep-plus-incubation group scored above the other groups on a combined creativity measure. Story and alternative-use judges did not know participants' groups. However, individual task comparisons varied, and the experiment lacked simultaneous brain-wave (EEG) confirmation of its device-defined sleep-onset periods. The manipulation combined prompts, reports and sleep opportunity; a result on these tasks is not a measure of general creative achievement. The 2023 paper also reused participants' incubation data reported earlier, so those publications are not independent replications.

Puzzle cueing tested later solving, not a remembered solution

Sanders and colleagues' 2019 experiment analysed 57 of 61 enrolled participants. Sounds linked to previously unsolved puzzles were selectively presented during sleep. Participants later solved a higher proportion of cued than uncued puzzles. This within-person comparison tested cueing; it did not compare sleep with an equivalent waking interval or establish a dream's contribution.

The procedure relied on an automated home sleep monitor, with exclusions for recognised sounds and some unusable nights. A 2020 correction fixes the placement of individual points in Figure 2. It should accompany the original paper rather than be mistaken for a separate experiment. The result does not tell us whether cueing helped selected puzzles, disadvantaged others or both.

Another task found no additional cueing advantage

Beijamini and colleagues' 2021 video-game study provides a different comparison. In its first experiment, 13 of 21 participants solved the previously unsolved challenge after overnight sleep, compared with five of 21 after daytime wakefulness. Time of day and sleep were therefore intertwined.

A second experiment analysed 58 participants receiving sounds during slow-wave sleep, rapid eye movement (REM) sleep or wakefulness, with no significant solving difference between those groups. Comparisons with the first experiment were exploratory, not a single randomised trial containing every condition. This does not prove cueing is ineffective in all settings, nor is it a direct replication of the earlier puzzle experiment. Different tasks, cues and comparisons can yield different findings.

Read the comparison before the headline

Start with the actual question. Sleep versus wakefulness asks about an interval containing sleep. Cued versus uncued problems asks about selective reminders. Prompted versus unprompted dream reports asks another question. A headline about “creativity” may hide any of these designs.

Then check the task, starting ability, exclusions and follow-up. Who had already solved the problem? Was the outcome correct solving, speed, the number of ideas or a judgement of originality? Were the groups formed by assignment or by what participants happened to do? These questions help you see why an encouraging result can be informative without applying to every creative activity.

A causal role for dreaming needs its own test

Suppose an invented intervention changes both what someone thinks about before sleep and what they later report dreaming. If the person performs differently afterwards, there is still a question about which part of the intervention mattered. Randomising the intervention helps test its overall effect; it does not automatically isolate the conscious dream experience.

A useful next study would address competing explanations, report unsuccessful as well as successful attempts and test whether the result extends beyond the prompted material. Learning, memory and dreaming examines related evidence about task reports and later memory. Why we dream addresses the broader question of possible functions, which no single creativity experiment settles.

Keep an idea without making it a productivity test

After ordinary waking, you can record an image, phrase or possible solution if you want to. Separate the remembered fragment from your additions: “I remember the curved doorway; the colour is my choice now.” This is an optional way of keeping a creative record, without a promise of improved performance.

Later, assess the work by the task itself. Does the proposed solution work? Does the sketch interest you? Does the sentence fit what you want to say? You can revise freely without claiming that the finished result reproduces a dream. The painting and music guides explore those particular experiences. None of this requires a prescribed sleep stage, a recall score or a finished product every morning.

Sources & their limits

These sources support the statements indicated above. Their authors do not endorse this website or provide an assessment of your experience.

1. Lacaux et al. (2021) — Sleep onset is a creative sweet spot

   Original paper's methods and results inspected through an archived PDF; observed groups and task-specific outcome.

   Cited in this guide · 2 places

   • The sleep-onset result concerned a hidden numerical rule
   • Did the sleep-onset study show a threefold increase in creativity?
2. Löwe et al. (2025) — N2 sleep and insight in a perceptual task

   Original methods, results and discussion inspected; conceptual comparison using a different task.

   Cited in this guide · 2 places

   • A later comparison found a different sleep-stage pattern
   • Is N1 the best stage for insight?
3. Haar Horowitz et al. (2023) — Targeted dream incubation and post-sleep creative performance

   Indexed original methods, ratings, results and discussion inspected after direct-access failures; no concurrent EEG sleep staging.

   Cited in this guide · 2 places

   • Dream incubation tested creativity tied to one theme
   • Does targeted dream incubation prove that any chosen dream will improve my work?
4. Sanders et al. (2019) — Targeted memory reactivation during sleep and next-day problem solving

   Indexed original methods, exclusions and discussion inspected; within-person cueing comparison, not demonstrated dream causation.

   Cited in this guide · 1 place

   • Puzzle cueing tested later solving, not a remembered solution
5. 2020 corrigendum to Sanders et al. — Figure 2

   Original indexed correction text and caption inspected; individual plotted points corrected.

   Cited in this guide · 1 place

   • Puzzle cueing tested later solving, not a remembered solution
6. Beijamini et al. (2021) — Sleep facilitates problem solving with no additional gain through targeted memory reactivation

   Original methods, group counts, results and exploratory-comparison limits inspected; different task and design from Sanders et al.

   Cited in this guide · 1 place

   • Another task found no additional cueing advantage

Put the context beside an experience

Continue through the science library