I often look away when I order food from a place where I can see it being prepared, because I expect to see things that will make me doubt my safety dining there. Similarly I prefer to sleep when I am sick, watch loud tv while on airplanes, and buy foods and drinks rather than make them myself where I can see myself making them.

This is all, of course, irrational. If I expect that opening my eyes will show me evidence that will make me believe X, then I already believe X. Or I should, if I am rational and expect to remain so. In these cases I don’t expect to remain rational. I quite reasonably expect that if I receive particular pieces of evidence I will update too much, so I should not believe what I expect to believe in the future, conditional on collecting evidence. I should avoid the evidence.

The ‘avoid evidence’ solution doesn’t seem like a very good one though. If I recognize that updating so much is irrational in time to avoid the evidence, why don’t I just recognize it when I get the evidence, and not update so much?

Perhaps I am just full of irrational fears that I can’t control by my mere will and reasoning. I don’t think that’s quite it though. Intuitively it seems the problem is that while I believe that I will vastly overweight any evidence I get to the effect that my sandwich is dangerous, when I actually see the rashy hand go into the lettuce or whatever it’s hard to judge whether this isn’t perhaps one of the rare occasions when I should be concerned. The specific piece of evidence looks different every time, so it’s hard to convince myself that a novel particular event that looks bad at the moment really fits into the reference class of other evidence that looks bad and isn’t dangerous.

Do other people behave this way? How should they behave instead? How do you fix this?