This is totally anecdotal, but I suspect this is a skill that varies across the population. A buddy in high school was uncannily accurate when guessing the time without access to a watch or clock. His error bar would increase with the amount of time past between "experiments", but was never more than ~5 minutes out, at a three or four hour interval. Inside an hour he was always +/- 1 minute. He thought it was unremarkable, and that we were weirdos for always quizzing him about the time.

Feinman writes somewhere about training himself to accurately perceive a minute, with some success.

I'm curious whether there are any human studies about this.

I wonder if that idea is compatible with something like caesium standard time. Is causality at work in the frequency of atomic resonance? (Separately, I would be inclined to disagree that we directly perceive causality.)