Learning About Unstable , Publicly Unobservable Payo s ∗

Neoclassical nance assumes that investors are Bayesian. In many realistic situations, Bayesian learning is challenging. Here, we consider investment opportunities that change randomly, while payo s are observable only when invested. In a stylized version of the task, we wondered whether performance would be a ected if one were to follow reinforcement learning principles instead. The answer is a de nite yes. When asked to perform our task, participants overwhelmingly learned in a Bayesian way. They stopped being Bayesians though when not nudged into paying attention to contingency shifts. This raises an issue for nancial markets: who has the incentive to nudge investors?