It is widely believed, both by proponents of the theory-theory and proponents of the simulation theory, that people can successfully predict one another's behavior on the basis of attributions of belief and desire.  The idea is that person A can observe person B's behavior, on that basis figure out that B has certain beliefs and desires, and on the basis of the attribution of those beliefs and desires, successfully predict what B will do.

My own opinion is that there is no reason to believe this.  Yes, we can often successfully predict what other people will do.  Sometimes we do it by straight induction.  (People look both ways before crossing a street.)  Sometimes we can predict what a person who has a certain skill will do as a consequence of having that skill.  (She's a good chess player; so she will take my rook.)  One of those skills is language.  We can often predict what people will do on the basis of what they have told us.  (He will meet me at my office at 10 tomorrow, because that's the time we agreed on.)

What I don't see is any evidence that we can reliably predict what people will do in a way in which attributions of belief and desire play an ineliminable role.  There are no experimental studies that even attempt to show that we can.  (What one kind finds lots of is studies where a researcher just assumes that we utilize folk psychology.)  Nor even have I ever heard of a single real-life example in which it was at least quite plausible that one person successfully predicted the behavior of another (and didn't  just make a lucky guess) and in which it was not evident that the same prediction could have been made in other ways (straight induction or inference from a skill possession).  

So here's my question:  Can any of you tell me an anecdote from REAL LIFE in which it was clear that one person successfully predicted another's behavior in a way that made an ineliminable use of attributions of beliefs and desires?  Fictional episodes, however realistic they may seem to you, do not fill the bill.

Please note that it is not enough to tell a story in which A predicts B's behavior and then rationalizes B's behavior in terms of beliefs and desires.  That will not show that the prediction actually depended on the attributions of belief and desire that provide the rationalization.

Please do not assume that I am an eliminativist or a behaviorist.  I am not.  I think that attributions of belief and desire play an important role in our lives, but facilitating prediction of people's behavior is not one of them.

Thank you in advance!