Implications of Human Irrationality for Reinforcement Learning