← Return to Preprint

Download

Bioelectric Integrity as an Intrinsic Reward Signal: Eliminating Reward Hacking via Physiological Constraints in Reinforcement Learning