A MODEL OF OVERCONFIDENCE

Abstract People use information about their ability to choose tasks. If more challenging tasks provide more accurate information about ability, people who care about and who are risk averse over their perception of their ability will choose tasks that are not sufficiently challenging. Moderate overestimation of ability and overestimation of the precision of initial information leads people to choose tasks that raise expected output (and utility); however, extreme overconfidence leads people to undertake tasks that are excessively challenging. Consistent with our results, psychologists find that moderate overconfidence is both pervasive and advantageous.