From Jon Elster, Ulysses Unbound, p. 72:
It is a conceptual truth that one cannot consciously decide to adopt a belief simply on the grounds that having it would be useful.
Not to get all personal here on sociology’s brand-new team blog, but can I just say: clearly Jon Elster has not dated some of the people I have dated. ‘Cause then he would have had conversations suggesting some deep empirical limitations to his conceptual truth. ‘Cause then he’d have ascertained that while, say, a person did not really, really believe in astrology, that, for immediate practical purposes they indeed did/would give all indications of believing in astrology, including reading charts and drawing conclusions about others on the basis thereof. (Various other things could be substituted for astrology here, depending on what conversation and with whom.) I guess “useful” in the quote isn’t really it so much as “adopting a belief on the grounds that having it would make life more interesting, meaningful or fun.”
Myself, I do not experience my beliefs as anything over which I have control. Of course, I engage in things that prove to be wishful thinking, etc., but I do not experience wishful thinking as I wish X were true, so I’m going to believe X is true. I experience beliefs instead as being caused by the world, and some abstract interpretations about the world, neither of which I feel like I’ve chosen even though, at a different level, I can appreciate the inky fingerprints of “the social” all over them.
I’ve wished on many occasions, in fact, that I could pull off the trick of consciously choosing beliefs, but I can’t seem to pull it off, the same way I can’t manage to see the pictures embedded in “Magic Eye” puzzles. I still recognize things I would choose to believe as chosen and as thus not really being beliefs.
And thus, I embody Elster’s conceptual truth. I experience it, however, as a kind of cognitive disability.
The closest I come, perhaps, is with beliefs about “internal locus of control,” as measured in surveys by agree/disagree items like “Becoming a success is a matter of hard work” or “What happens to me is my own doing”. My natural inclination–I am in sociology after all–is to be more skeptical toward such statements than what is optimal for actually accomplishing things in the world. So I make a conscious effort to have more of an internal locus of control, which may in fact have had some influence on how I would respond to such statements on a survey, but mostly manifests itself on me holding dear maxims like “One needs to focus on what one can control and make the best of it.” This is still not quite the same as suddenly resolving “What happens to me is my own doing” and really believing it.

2 Comments
Doesn’t Pascal have a discussion about this wrt the wager? Having argued that it’s rational to believe in God given the balance of expected payoffs, he goes on to discuss whether it’s possible to simply decide to believe in God because of the potential benefits. While acknowledging that some people think that such a decision is impossible, he doesn’t see it as much of a problem at all: hang out with believers, go through the rituals of religious worship, in general make a good faith effort to believe, and believe you shall.
Yes, my former colleague James Montgomery has a paper on the rational choice of Pascal’s wager where he makes use of this part of Pascal’s argument.