You have probably heard it before: sleep on it, and suddenly everything will fall into place. It is the idea that dreams can be a source of insight and help you solve problems. Or at least that a good night’s sleep can. A recent study by Konkoly and colleagues tries to capture that idea experimentally. When I read the study and its accompanying press release, I quickly got the sense that people might run away with this. So here is a brief critical look.
The design is genuinely clever. Before going to sleep, the researchers gave participants a set of creative puzzles. During REM sleep, some of these puzzles are subtly reactivated through sound cues. The aim is to test whether you can not only steer dreams, but also make them more productive.
That first part works quite convincingly. Participants dreamed more often about the cued puzzles than about the others. That is not a trivial finding. Experimentally influencing dream content is methodologically difficult, so this part of the story is relatively solid. The researchers could, to some extent, determine what people dream about. But that is also where the certainty largely ends.
When you look at actually solving the puzzles the next morning, the picture becomes much less clear. Across the full group, there is no significant advantage for the cued puzzles. In other words, dreaming about something does not automatically mean you will solve it better. That does not strike me as surprising. After all, the solutions themselves were not cued. But it does go against the hopeful narrative.
The authors go one step further and focus on a subgroup of participants who actually dreamed more about the cued puzzles. In that group, there does seem to be an effect. But this is a post hoc analysis. You select participants based on how strongly they respond to your manipulation and then test whether the manipulation works. That can generate interesting leads, but it is weaker evidence. The risk of overestimating effects is always present here.
There is more. Participants knew that the study was problem-solving, and they were even given strategies to work on the puzzles in their dreams. That makes it difficult to fully rule out demand effects. People might dream more about certain puzzles because they think that is what is expected. Or they might report it differently afterwards.
The measurement itself is not straightforward either. What counts as “dreaming about a puzzle”? That is determined partly through retrospective dream reports and partly through signals during sleep. Both are interesting, but also open to interpretation and noise.
And then there is the size of the study. Twenty participants, a binary outcome (solved or not), and limited room to model differences between puzzles. That makes the results fragile. Small shifts can quickly appear to be meaningful effects.
Perhaps the most striking detail: lucid dreams, often presented as the key mechanism for creative problem solving during sleep, do not seem to perform better here. If anything, the numbers suggest the opposite, although they are far too small to draw firm conclusions.
So what does this study actually leave us with?
Mainly this: you can increase the likelihood that people will dream about a problem. That is interesting and raises questions about how memory and sleep interact. There is also a tentative indication that when a problem appears conceptually in a dream, the chance of solving it afterwards may be slightly higher.
But that is not the same as saying that dreams solve problems. Or that you can “optimise” your sleep for creativity with the right technique. It is tempting to turn an elegant experiment into a strong story, especially when it aligns so neatly with an intuitive idea. But that is precisely why caution here is not a weakness, but a strength.
It may help to look at it this way. This study shows that we can influence a small part of the dreaming process. What those dreams do to our thinking, then, is a lot more complex. And for now, still not entirely clear.