This paper presents an experiment on learning in repeated games, which complements the analysis of players' actual choices with data on the information acquisition process they follow. Subjects play a repeated Cournot oligopoly, with limited a priori information. The econometrics hinges on a...