Understanding the neural basis of reinforcement learning