The process by which individuals learn from feedback when making recurrent choices among ambiguous alternatives is explored. We describe an experiment in which subjects solve a variant of the classic armed-bandit problem of dynamic decision theory, set in the context of airline choice. Subjects...