An experiment in mapping reinforcement learning agent brain activity to sound. Output activations drive oscillators while hidden layers control formant filters, allowing users to "hear" the agent's decision-making process during simulation.