Aishik Sanyal

Adding an affect proxy to a recurrent state machine makes the agent less sensitive to novelty and more likely to exhibit structured exploration and sustained caution, suggesting a pathway to more robust and human-like behavior in AI agents.