Reinforcement Learning and Insight in the Artificial Pigeon