26-06-2025
Eureka! Increasing the Odds of a Sudden Insight
This transcript has been edited for clarity.
Welcome to Impact Factor , your weekly dose of commentary on a new medical study. I'm Dr F. Perry Wilson from the Yale School of Medicine.
At a time when the capabilities of artificial intelligences seem to be growing by leaps and bounds at an incredible pace, it's comforting to remember that there are certain cognitive processes that continue to feel distinctly human. I want to talk about one of those processes today. It's the idea of 'eureka': a sudden flash of insight, where the solution to a puzzle or a problem or a mystery just sort of clicks.
When we think about these moments, several features come to mind. First is the suddenness. It's not that you slowly improve at solving a problem, you slowly improve and slowly improve and then — all of a sudden — you dramatically improve. Then there's the timing of the insight; it's variable. Given the same puzzle, some people 'get it' quickly, and some take longer. And, of course, some people never get it at all.
This process seems so singular, so qualitative, that it may appear to be impossible to study. But a study of sudden insight is exactly what we're going to be exploring today. Let's see if it clicks.
Abandoning stories of Greek philosophers and kings' crowns, researchers, led by Anika Löwe, refer to these sudden flashes of insight as 'aha moments.'
To me, an 'aha moment' is sitting in my girlfriend's basement watching the 'Take on Me' music video, but there's never a forever thing.
Here, we're talking about those sudden flashes of insight, and, in particular, whether a bit of sleep improves the chances of having them. The paper appears in PLOS Biology.
How do you test aha moments in the laboratory? The system is really clever. Participants (90 in this study) were exposed to a very simple computer game during which dots move around a circle in different directions, but one of four directions (northwest, northeast, southwest, and southeast) is more common than the others. The participants are asked to press one of two buttons when they see the set of moving dots and are then given feedback on whether they are right or wrong. There is a hidden rule here — for example, to be correct they need to hit the left button when the majority of dots are moving northwest or southeast, but the right button when the dots are moving northeast or southwest. They have to figure this out through trial and error.
It's not that easy, especially since there is a lot of noise in the direction of travel of the dots. But, as you might expect, task performance improves over time — people figure out the rule underlying the correct answers and do better and better over hundreds of iterated trials.
As looking at the moving dots, you notice that they change color as well, from orange to purple at random. For the first 80% of trials or so, these color changes are random; they have nothing to do with the correct answer. But here's the trick: Toward the end of the series of trials, the colors start to match up with the correct directions. In other words, all of a sudden, orange dots would indicate that the participant should press the left button, purple dots the right button. But the participants have to realize this is happening; no one tells them in advance.
That requires a sudden flash of insight — an 'aha' moment. But once a participant has that moment, their performance skyrockets. It's way easier to match a color to a button than two of four noisy directional trends to a button.
I find this super clever. Major props to whatever neuroscientist had the flash of insight to design this system.
Now we have a framework for measuring aha moments. The next step is to perturb the system.
The researchers wanted to determine whether sleep (and which stage of sleep) would improve the aha phenomenon.
After an initial round of testing, where the 'color rule' appeared only at the very end, the participants went into a quiet room to nap.
An EEG was used to determine whether they merely rested or they had reached stage 1 or stage 2 of sleep. Stage 1 sleep is pretty light — almost conscious. Stage 2 is slightly deeper but not to the level of deep sleep. Only one participant reached stage 3 during the short nap.
Some people figured out the secret rule before naptime; they were excluded from further study. So now we have a group of people who had not yet had an aha moment. Would the achieved sleep stage matter when they were tested again?
It did. Pretty profoundly, too.
About 50% of the group who did not sleep at all during the rest had the color insight during the second round of testing. About 60% of those who hit stage 1 sleep had the aha moment, but 80% of people who reached stage 2 sleep had that insight — a statistically significant improvement.
Interestingly, sleep only seemed to affect the proportion of people who had the insight. It didn't affect the time to the insight. And napping had no effect on task performance prior to the insight. Rather, it seemed that napping just allowed the magic to happen.
But how does the magic happen? The researchers interrogated the EEGs to see whether something deeper was going on besides stage 2 sleep. They found that aperiodic activity in the brain captured all the information that sleep stage did. In fact, it correlated more strongly with insight than sleep stage itself. So…what is aperiodic activity, and what is happening when aperiodic activity in your brain is higher?
Aperiodic activity is electrical signals in the brain that don't have a nice repeating pattern. They seem more random and have been described as the 'background noise' of the brain. But what is interesting is what is happening to neural connections as aperiodic activity is increasing. They get weaker.
This sounds bad, but weakening connections between neurons is actually a critical thing. It allows connections that are already weak and of poor quality to break entirely, allowing the stronger connections more exclusivity. Those of you in the machine learning space will recognize the similarity to regularization methods. But basically what is happening is that the 'noise' of the complex brain is being turned down, so the signal can come through that much louder.
And maybe that is what allows the insight to happen. Perhaps a brain with a bit less noise and a bit less clutter can more clearly see the underlying structure being presented to it and make those cognitive leaps that (for now) separate humans from the neural networks we create in silico.
Of course, looking at this study I'm sure you're thinking the same thing I am: How can I leverage this knowledge to increase my likelihood of having eureka moments? Well, better sleep seems like the obvious answer here, but I think we all knew that already.
Maybe we can take a lesson from this idea of signal and noise. If the key to aha moments is allowing the stronger connections in our brain to function more cleanly, without the distraction of those weaker connections, maybe that means we need to avoid those weaker connections if possible — those distractions. Does scrolling TikTok reduce your chance of having sudden insights? Does meditating in a quiet room increase them? I'm not sure, but at least now we have a mechanism to study those questions.
In closing, let me express my wish that you have many aha moments, and that they won't be gone in a day or two.
