I disagree - humans often reason in a series of steps, and can write these down before they've reached an answer. They don't always wait till they reach a conclusion (with no self-insight into how they did so) and then retrospectively generate a plausible answer as LLMs do.
In mathematical proofs they may guess and answer and then work out a proof, but that is a different process.
