Non-monotone social learning

We revisit the canonical binary-state model of social learning to investigate possibility non-monotone learning: ceteris paribus, with some predecessor(s) switching actions that reveal greater confidence in one state world, agents nevertheless become more confident other. A necessary and sufficient condition for is provided an illustrative setting where are either uninformed or partially informed by binary private signals. In a general continuous signals, we obtain does not rely on shape underlying information structures apart from couple simple boundary requirements. As result, process completes eventually can be well approximated way such anti-imitation bound happen.