For questions about rewards functions (e.g. in the context of reinforcement learning, which may be denoted as $R(s, a)$).