For you to rule out a belief (e.g. geocentrism) as totally unbelievable, not only does it have to be less likely than insanity, it has to be less likely than insanity that looks like rational evidence for geocentrism.

You can test yourself for other symptoms of delusions - and one might think "but I can be deluded about those too," but you can think of it like requiring your insanity to be more and more specific and complicated, and therefore less likely.