Textmeddelande: Adaptive representations for reinforcement learning