Reinforcement learning models make use of reward prediction errors (RPEs), the difference between an expected and obtained reward. There is evidence that the brain computes RPEs, but an outstanding question is whether ...