Planning for Mistakes

How should you decide what to do? Orthodox decision theory offers attractively simple advice: do whatever has the most expected value. In what follows, I argue that this is not always good advice for agents like us. Agents like us are susceptible to error: Sometimes, we do not succeed at doing what is recommended by the advice we are trying to follow. For agents susceptible to error, there can be situations in which trying to do what they expect to produce the most value will be expected to leave them worse off. I'll discuss why situations like this arise and give some concrete examples. I'll also identify a condition on agents' credences which is both necessary and sufficient for maximizing expected value to be the best advice for them to try to follow. However, crucially, this condition will not always be satisfied by agents like us.