Learning and Selfconfirming Long-Run Biases
Abstract
We consider an uncertainty averse, sophisticated decision maker facing a recurrentdecision problem where information is generated endogenously. In this context, we study
self-con rming strategies as the outcomes of a process of active experimentation. We
provide inter alia a learning foundation for self-con rming equilibrium with model un-
certainty (Battigalli et al., 2015). We also argue that ambiguity aversion tends to stie
experimentation, increasing the likelihood that decision maker gets stuck into suboptimal
certainty traps.