Using computational methods to model neural recordings and behavioral data, the authors find that reinforcement learning rules, combined with a standard model of decision-making, can explain a ...