Neuroscientists track the neural activity underlying an “aha”
quantamagazine.org
quantamagazine.org
This is the premise of a really good article I reccommend to anyone, the Seductions of Clarity by C. Thi Nguyen (https://philarchive.org/rec/NGUTSO-2)
It is quite literally the source of one of our most dangerous failure modes.
That aside, working with complex systems and constraints there often isn't an aha moment, there's just a decision to be made. As someone who loves that aha moment, I can get stuck trying to figure out perfect from good enough. Interesting to see there is indeed a positive emotion correlated with that aha moment that keeps people searching for solutions.
I wonder if there's a correlation between addiction and this aha moment. Like you get drunk and suddenly "aha!" those big unresolvable problems don't matter. The next morning they matter again until, aha, beer:30 hits.
Are you kidding? I've been chasing that epiphany dragon for decades and so has everyone else in the shop. Ever feel like you've got one foot out the door once you comprehend the systems you work with?
You only get the a-ha moment when there's essentially one discrete piece of information needed to decide between alternatives. That doesn't apply to most problems.
Your brain simultaneously assigns probabilities to possible solutions, and in certain cases there's an information update that sets one solution to probability 1 and the others to 0. If your brain is actively expending energy keeping these possibilities warm simultaneously, then this will naturally lead to a rapid change in energy which will feel like something because it's a change in the flow of neuro chemicals.
It's not obvious that it would feel pleasant. But since the nucleus accumbens is active during problems solving then it's not entirely surprising that the the NAc gets extra stimulated in the rush of energy as the probabilities collapse and weights get updated to the real solution.
But relatively few problems require you to simultaneously juggle multiple possible solutions and pieces of evidence that are brought together in a single instant. So chasing that feeling is generally a poor strategy.
One is a trickle, the other a rush.
Try pairing a feeling to: "So THAT'S how a mouse cursor moves"
"So THAT'S why revolving doors move clockwise!"
"So THAT'S why lights at night feel cosy!"
With a little practice, you can arbitrarily get 'aha' moments. I assume the good feeling is some sort of dopamine release where my brain is rewarding me for "figuring something out," even though I've kind if hijacked the mechanism.
I completely agree with you that fMRI along with EEG and fNIRS isn't giving us true insight into how thought works.
However, it doesn't mean they are completely invaluable when brain activity and stimulation can be manipulated, and that change can be measured through measures other than just the brain activity itself.
For example, our work at https://affectablesleep.com stimulates slow-wave activity during sleep, a core component of what we refer to as sleep's restorative function.
Though we measure and see the change in brain activity, research has also measured changes in cortisol, HRV, immune function, etc etc.
A friend is working on neurostimulation in depression and their background is in fMRI, though for accessibility they are working with EEG.
The point is, you're right, our understanding of the brain is not much more than medieval, but that doesn't mean it is completely invaluable.
I believe it is not valuable when we're trying to understand the "aha" moment, simply by looking at blood-flow in the brain, as this article suggests.
Seems like there would still be some use in looking at what's physically there. And, knowing that all humans have roughly the same activity patterns seems very important, in the sense that we're not some blank slate for concepts to be encoded.
https://www.youtube.com/watch?v=aB_4YU6UtCw
tldr: There is a background, non-verbal process in your brain that has the advantage of a larger working set size than your foreground verbal thinking. It is able to observe and consider more stuff at once and find associations better than your conscious thought process.
But, it has several disadvantages. It takes time to do its processing. You can't will it into action. It communicates non-verbally with your foreground process. It doesn't work under pressure (thus the need for relaxed, unfocused time). The non-verbal understanding is difficult to deconstruct, generalize and reapply. It can lead to you solving a problem, not understanding how and not being able to solve a variant of the same problem.
So, the general recommendation is: If you have a complex problem to solve, first absorb as much information about the problem as your brain can hold. But, do not try to solve anything. Then, go take a break. A walk in a natural environment is preferable. Don’t think about the problem. Relax in a low stress environment. Let your background brain have a chance to chew on it and maybe bubble up some suggestions.
Limbic: https://en.wikipedia.org/wiki/Limbic_system
Frontal Lobe: https://en.wikipedia.org/wiki/Frontal_lobe
Disclaimer: I only know this from armchair psychology books like Habit, Start with Why, etc.
By keeping problems in a dormant state, the brain can work on them in the background, leading to unexpected solutions when a new piece of knowledge is encountered
To me it felt more like a brute force search, or like solving a Wordle puzzle.
I consider "aha" more creative, like recognizing that key insight that crystalizes a solution to a problem you're working on. (Or maybe a pattern or analogy that cleanly collapses a swath of the complexity).
YXBwbGU= (Use Base64 Decoding) [/Spoiler]
Wild tangent incoming...
One instance that recently bothered me with an NYT puzzle was the crossword clue (3 letters): "Chromebooks, but not MacBooks". The answer was "PCs" which doesn't make sense to me under any level of categorization for PC.
If we go narrow/historic, then it means x86 IBM PC derivatives which eliminates a lot of chromebooks.
If we use the "home computer" interpretation, then I think it's unreasonable to except Macbooks from the PC umbrella.
If we go literal, well then everything is a PC, including smartphones, tablets, smart devices. The only reasonable test seems to be "Can it play Doom?". :D
Using PC in a "every consumer computing device but Mac" probably made sense in the 80s/90s, now it seems to dilute the term to the point of confusion. I have personally never thought of a Chromebook as a PC, given that it ships with an OS incapable of many things people generally associate with PC activities.
To guess it, I looked at 'crab' because it's a quite uncommon that has some deep relationship with a few words only. Then checked the most obvious one (which was the solution) against the other words, and determined that it didn't bear any significant relationship to the third word. So I checked the other (less obvious) potential solutions, and after a frustrating lack of match, I gave up. And then got annoyed that the first candidate was the right one. To be fair, I guess it's partly because I'm an ESL, as I think that solution/sauce can be used as a nominative locution enough to form a "special relationship".
To be a designer, you have to play with people's (as in general crowd, not individuals) general understanding of the subject. In particular, that means avoiding the curse of knowledge, and yes for normal people PC meant "not Apple consumer product". So ultimately, the search algorithm includes:
- categorize all relationships between words, ranked by strength
- compare with what is expected to be known in popular culture (adjust ranks)
- match against the designer's expectations of similar problems (look for clues to pick a best match)
It's a lot of words to say it's the opposite of a aha moment, the result of a pure computational problem, that is often quite frustrating. Thank you for coming to my TED talk.
And yeah that often results in mild disappointment or frustration instead of an "Aha!" moment. Actual puzzle video games fair better for me at that aspect, as they avoid the inevitable subjectivity of natural language.
If anyone wants an additional hint, the word you plug in here isn't put in the same spot for all three words.
->https://www.quantamagazine.org/how-your-brain-creates-aha-mo...
Transformers learn almost everything through language-like supervision. Wrong token = small penalty, right token = small reward. That’s great for pattern induction, but it means the model treats a correct chain-of-thought and a beautifully phrased but wrong chain-of-thought as almost the same kind of object—just sequences with slightly different likelihoods.
Human reasoning isn’t like that. When a logic chain closes cleanly, the brain fires a strong internal reward. That “Aha” isn’t just emotion; it’s an endogenous learning signal saying: this structure is valid, keep this, reuse this. It’s effectively a structural correctness reward, orthogonal to surface language.
If AI ever gets a similar mechanism — a way to mark “self-consistent causal closure” as positively rewarded — we might finally bridge the gap between language-trained reasoning and true general learning. It would matter for:
fast abstraction formation
reliable logical inference
discovering new concepts rather than remixing old ones
Backprop gives us gradient-based correction, but it’s mostly negative feedback. There’s no analogue of the brain’s “internal positive jolt” when a new idea snaps together.
If AGI needs general learning, maybe the missing piece isn’t more scale — it’s this reward for closure.
Nicholson, Hannele B. M. (2007) DISFLUENCY IN DIALOGUE: ATTENTION, STRUCTURE AND FUNCTION PhD. thesis, University of Edinburgh, Edinburgh, Scotland, UK, available online: https://era.ed.ac.uk/bitstream/handle/1842/1763/Nicholson%20...
But "um" may not be quite the same as "aha" for English native speakers (and Japanese native speakers may use both "um" and "ahm" as disagrement).
That also brings to mind that first exposure to this dataset affects the effectiveness of the rest of the dataset. If you're doing initial exposure, you'll definitely get the "aha" moment. But if all of the images in the dataset are of the same type, your brain quickly learns the pattern and the "aha" moment vanishes.
If they did their study on all of the images per test subject, the results after maybe the first 5 are basically useless for any definitive conclusions.
I haven't got any beyond my own working notes and some basic plots, but I've unceremoniously dumped them into a document here incase anyone else finds them interesting. If so I'd _love_ to chat with you. enjeyw @ google's email provder.
https://thealephengine.substack.com/p/67e3786f-8e84-41bd-888...
Serendipity is more like a fortunate accident.
> You might even say “Aha!” This kind of sudden realization is known as insight, and a research team recently uncovered how the brain produces it (opens a new tab), which suggests why insightful ideas tend to stick in our memory.