Just trying to think of counterexamples:
Oneboxing: Omega's decision is conditioned on its prediction of your mental state. Causing yourself to believe oneboxing is better than twoboxing changes reality, if Omega notices the change in belief.
Plausible deniability: You don't want to believe true things if such knowledge creates an obligation to report. You may even prefer to be misled to believe the opposite.
Lying: You need to pass a lie detector, so you cause yourself to believe the lie.
Optimism: You are more effective with a positive attitude, so you cause yourself to believe the optimistic side because the value of the truth difference is worth less than the value of the productivity increase.
Charisma: If other people think you agree with them, they're more likely to like you. It's easier to copy their speech patterns and mindset if you cause yourself to believe their views, even if momentarily.
Compression: The full state of every proton in the universe is impossible to keep in your head, so you believe a simpler version because it saves you bits. But "because it saves you bits" is a distinct reason from "because it's probably true"(unless the word 'probably' is meant to quantify bits).
Efficiency: A heuristic might be wrong, but following it leads to the correct place anyways for unrelated reasons. Ideally you'd understand why, but in the meantime people that believe it are more successful.
Some of them you might be able to reword: Believing the heuristic is an accurate model vs. believing that it is useful.