Well, I don't know that every scientist has to have poetic dreams about snakes (Kekule) or apples (Newton).

Weirdly I often have dreams where I am 'solving' some problem, however on waking I realise the details are nonsense - not just the solution, but the problem itself. I think dreams are often a way to just exercise and sort memories and ideas. There is no reasoning involved.

A lot of science is just process, or multiple careful experimentation. Preparing for getting lucky is another matter.

When I first learned about recursion I could not understand it (specifically, I didn't understand how eventually you reached a termination point and popped back up the stack). I spent all day thinking and thinking and thinking. As I was falling asleep- in that sort of lucid awake-but-not-awake state, the entire program I wanted to write appeared before me in glowing letters. Some deep part of my brain woke me up, I wrote the program down, and went back to sleep.

When I woke up the next day, I compiled it, ran it, and it worked on the first try (up to the stack limit on my computer... 6 function calls deep).

Since then I've noticed that if I get super-duper obsessive about things I don't understand, insights will start to pop into my head, but at the same time, I can't really talk to other people because the human interaction part of my brain is "paged out". I kind of understand tormented geniuses better knowing that many of them also walk a fine line between human life and discovery.