Doubts about the descriptive validity of the expected utility theory

This paper summarizes the major events in the recent history of modelling human decisions under risk. After presenting the basics of expected utility theory, the key pieces of evidence are described which showed that under certain circumstances, this theory is not descriptively valid. The most promising alternative, cumulative prospect theory, is then presented in some detail, including a brief discussion of how it avoids violations of stochastic dominance and how it explains the above evidence. It is pointed out that there are other empirical observations which cannot be explained by cumulative prospect theory either, so that a model which would explain all evidence on decisions under risk is still to be found.