Secondary Reinforcement in a Discrimination Problem