How rewards teach reinforcement learning agents to behave
Briefly