Basics of Human Reinforcement — LessWrong