Math creativity has similar behavior. I was the best student in my computer theory class in IU grad school, for various reasons, but one of them was that I seemed to be the only one who started the weekly homework earlier than the night before it was due. The homeworks consisted mostly if not entirely of math-logic proofs, and I started early enough that (a) I wasn't stressed (stress is the mind-killer, at least for creativity) and (b) if I got stumped I could relax and sleep on it, solutions often coming in the morning or shower.

(Also (c) if I was *still* stumped, I could email the TA two nights before the due date, and get a relaxed reply from him, vs. everyone else bugging him the night before. Between that and my concise answers, I'm sure he loved me.)

I really hate job interviews that throw a hard algorithmic problem at me in an hour. Something like that, I want to be able to turn it around in my head, play with it, chew on it, put it aside -- as I would on the job -- not be frantically trying to solve it in minutes.