simplified reinforcement learning (RL)