This. I know it has very little scientific merit, but Roger Penrose's argument that "consciousness is quantum" always rang that bell to me, the thought (or a solution to a problem) comes to me always entirely at once, not one word at a time.

The scientific method seems to me language-based. How to come up with hypotheses, however, seems to me exactly as you mention. They just pop out of nowhere, like a sudden inspiration that has been processing in the background and now just clicks.

I am not sure what you mean by the scientific method. The scientific method can be as simple as trial-and-error or observation, such as the idea that maybe you shouldn't eat a certain berry if you just watched someone eat one right before they passed out and started foaming at the mouth. Animals without highly developed language do this all the time.

There is no doubt that writing out or sounding out your reasoning can boost your ability to think and communicate. For example, it can promote a dialectic which can refine your ideas, and you can even have an inner dialogue using language. It can also help you form a representation of the thought process for communication.

But it can also muddle your thinking or leave you with blind spots if the language is not expressive enough or you rely on it too much. You do not need a linguistic representation of a concept in order to reason about it (well, I don't seem to and I doubt I'm special in that regard).

Just because an internally consistent language may be sufficient for basic reasoning doesn't mean it's necessary overall, or that it doesn't place a 'ceiling' on the types of thoughts you can think. I would argue that it cuts both ways.

It is definitely interesting that you seemingly get reasoning "for free" when you only optimize for language (e.g. large language models). But that is the way it would be if grammar was a more formal descendant of logic.

Maybe this is why authentic geniuses, or people capable of coming up with ideas not commonly in circulation, often have a hard time explaining their ideas (see https://kottke.org/13/05/the-three-types-of-specialist)

Thanks for the thoughtful post. I am a bit late, but I meant coming up with a Hypothesis and validating it through empirical evidence. Connecting these two explicitly needs logic (when I do x y happens) and, IMHO, logic needs language (or at least seems to have a relationship with language).

I guess if animals can do this, we have to options: either animals have language (even if basic) or the scientific method does not require language. I see merit in both arguments.

The working of a neural network makes total sense to me since I learnt it. Ideas emerging out of blue, etc. I believe it’s how brain works.

About thinking, I always think (hah) of this quote:

"Lo que solemos elevar a la categoría de meditaciones no es más que el ruido de un motor encendido." Ray Loriga

My idiomatic translation: "What we usually elevate to thinking/thoughts is only the noise of a running engine."

I also believe language is necessary for thinking. I would define thinking as following the scientific method, though. That is, thinking is not necessary to tie my shoelaces or go shopping or dance (that I consider more autopilot, maybe like system 1 and 2?). But to make a rigorous argument, IMO one needs language. Under this definition of thinking it seems logic is pretty important, and logic seems to me language-based.

> I also believe language is necessary for thinking.

I suspect that different people think verbally to different degrees. I mean, everyone uses words in their head to manipulate ideas, to some extent; but some people use pictures more than others, and some people are very abstract, and despite being super-bright, aren't very good at translating their ideas into clear language.

Do people not able to think without language normally? Does everyone have an internal sound for everything they think all the time? I don’t, I am bilingual but I don’t think this is true. I’m surprised to hear that.