For what it's worth, the "Super Base Rate Fallacy" seems to line up with my own experiences, except that there's sometimes an independent part of my mind that can go "Okay, I have 99.999% confidence that the floor will eat us. But what's the actual odds of that confidence, and what evidence did I use to reach it?". While I can't just dismiss the absurd confidence value as absurd, I can still (sometimes) do a meta-evaluation about the precise confidence.

It's sort of like how if a friend says that global warming is 99.99% likely to be true, I can't simply rewrite my friend to have 50% confidence. But I can question the evidence and see how he reached his conclusion, and if it's just "oh, I read a newspaper article that said it was real", my actual confidence will be vastly lower.

I only recently figured out this trick (and suspect LessWrong probably helped me develop it), so I couldn't say why it sometimes works and sometimes doesn't. I can say it's much harder to ignore paranoia about people, and much easier to ignore anything that would be easily objectively checked ("the floor will eat me", step on to floor, "the floor failed to eat me. Falsified!")