Players with observable, continuous payoff types choose actions sequentially. Players update beliefs about the payoff-relevant state of nature using private signals and information about previous signals in the history of types and actions. Heterogeneity in preferences ensures complete learning.