Pick the Best Weighted Average, Not the Best Story

Expected utility theory answers a narrow question: how should you choose when you don't know which outcome your act will produce? Its slogan is blunt — choose the act with the highest expected utility, where each possible outcome's utility is weighted by the probability that the act leads to it, and the weighted total decides the comparison. That is a normative theory: an account of how people should decide, not how they do. The same machinery has a second life as a predictor of actual choices, and there it performs badly. In many real-life choice situations the theory makes faulty predictions, the entry notes, citing Kahneman and Tversky. What follows is subtler than it looks. Bad predictions about our choices show that the ideal and the behavior come apart; they do not by themselves settle how we should choose.