I remember when I was having dinner with a friend earlier this year and I was trying to express what my research interests were, and what I said is that I want to understand thought. what even is thought? what is it exactly? and perhaps from there we could also explore what “reason” and “knowledge” are.

now I’m less convinced that understanding thought itself is particularly fruitful. at least, the question as it’s framed is too broad. here’s what I think thought is: mental reconstructions and combinations of sensory experiences. this is admittedly speculative, but I here’s what I think is happening.

you are born and you are perceiving a world. there’s your input (sensory experience) and there’s also your output (motor commands). that’s pretty much it for a while, but at some point your brain develops the capacity to consciously simulate both the input and the output. you can imagine seeing a green turtle, or you can imagine saying really mean things to your boss. my theory is that thoughts begin as very simple simulations of sensory experiences/motor actions, and in time they become increasingly sophisticated.

there are some interesting questions here around how the brain makes a command to, e.g. simulate the image of a turtle. like in particular, before you start imagining a turtle, how is the pre-imagined “turtle” represented? how does the brain have the kind of precision required to tell its imaginative faculties to imagine a turtle and not a purple flag? I’m not sure about how this works, but my guess is that it’s always based on pre-existing memories. like, you have seen turtles and you have seen flags and you have seen the color purple, and each of those things is represented in some way in the brain (perhaps as a pattern of neural activation, or perhaps as something else), and the way that your brain reactivates the image of a turtle is that it says: “let me activate some of these neurons/primitive representations that are closely related to (or somehow are prerequisites for) the image of a turtle”, and this happens in a very fast, iterative, progressively detailed manner, until the image in your mind becomes sharp and detailed enough that you can now definitively say, “yes I am imagining a turtle.”1

so that’s one hypothesis for how the brain could “choose” what to imagine before it actually properly imagines that thing. but the broader point I was making is that this is all that thought is: simulating sensory experiences in the absence of external stimulus. or rather, intentionally simulating sensory experiences in the absence of external stimulus. initially all we can imagine are very simple, primitive experiences, perhaps “lightness” or “darkness”, or “mom” and “dad”, and increasingly we can simulate more and more fine-grained experiences, in fact we begin to imagine entire worlds, we begin to imagine hourlong back-and-forth conversations in our head, and we begin to imagine things that we have never seen or experienced before.

how do we do that? how did JK Rowling come up with Harry Potter, and imagine him in detail, when that character never existed before? well, Rowling didn’t create something entirely new: the concept of a wizard already existed, the concept of “Harry” already existed, the concept of fancy private schools already existed, all she did was combine these things in a novel way. that is how I think novelty occurs in thought: we combine and recombine sensory primitives in new configurations and combinations, which will sometimes generate things that are totally new. but importantly, I think we keep needing new stimulus from the world to keep coming up with new things. if your external world is always the same, I imagine there’s a limit to how many new ideas you could come up with. the world needs to change (even if very slightly and subtly) for you to keep being able to come up with novel thoughts.2

ok, so I think that addresses the question of what thought is, and the question of how we have new thoughts that didn’t exist before. this brings me back to the original problem of: is it worth trying to study the very nature of thought itself? to try to understand, on some fundamental level, what thought is? my current answer to this question is no. if we put aside the ever-perplexing problem of consciousness, the problem of “what thought is” feels similar to asking what “sight is” or what “hearing is”. there isn’t fundamentally anything special about them, some deep principle that we haven’t discovered yet. sight is generated by the way that light bends and reflects into your retina, and you have little receptors in your retina that respond to the presence of photons, and these signals bubble up to your visual cortex to generate vision. of course, there is a lot of detail missing there, and much more to understand. but my point is that the “big picture” is a very reasonable one. and just as we have a big picture understanding of what vision and hearing are, maybe the picture I’ve just described is a satisfactory model of thought, and what we really need to do is fill in the details.

there’s one other aspect of thought that I think is worth mentioning, and which I think contributed to my earlier sense that there is something fundamentally mysterious about it. the question is this: how is it that our thoughts “mean” things? how is it that I’m able to think about, say, my friend’s dog, and I’m able to imagine that dog (a shiba inu) in all her cuteness and lovability? like what is going on that enables my brain—which has no causal, physical relationship to this dog at the moment—to “refer” to this dog? and the same question applies for my ability to think about other physical objects, and abstract ideas like the number 7 or “freedom” or “truth”.

my answer to this is something that I derive loosely from people like David Chapman and Wittgenstein, which is that the correspondence between your thoughts and the outside world is always contingent. there is no special sauce in your thoughts that make them magically refer to the exact thing that you believe you’re thinking about. your thoughts are a hazy, loose approximation of the things outside your head, and your belief that your thoughts are somehow fundamentally real, that they’re inextricably linked to the thing they refer to, is actually just false. your thoughts do not intrinsically have meaning. they have meaning only by virtue of your ongoing interaction with and connection to the physical things in the world that you’re thinking about. this is why we’re capable of being delusional: our thoughts don’t need to correspond to anything real at all. the thread that is tying our thoughts to the real world is a tenuous one, a thread in constant need of reweaving.

once you recognize this, thought becomes even less mysterious than it was before. how am I able to think about brains, and galaxies, and historical figures? how am I able to think about ideas that have never been contemplated before, characters and worlds that have never existed? thought is indeed a magical thing, but the magic is less in the fact that it happens at all, and more in the richness of experiences it enables us to have.


  1. obviously the personification here is used for illustrative purposes, there is no part of your brain that’s going “let me imagine a turtle”. ↩︎

  2. I want to stress that by “novelty” I mean even the most subtle kind. you could live in a forest for the rest of your life and you’ll have more than enough novelty to come up with interesting ideas indefinitely. by virtue of even being able to manipulate your physical environment (e.g. re-arrange physical objects into new configurations; etch words onto a piece of paper) you can create physical patterns in the world that have never existed before, which then provides new sensory experiences that your brain can recombine into further new thoughts. there’s a beautiful feedback loop between new physical patterns in the world and new ideas in your head. ↩︎