Occam's razor is insufficient to infer the preferences of irrational agents
This work addresses a fundamental ambiguity in inverse reinforcement learning for irrational agents, highlighting the need for normative assumptions beyond observational data, which is incremental as it builds on existing IRL limitations.
The paper tackles the problem of inferring reward functions from observed behavior when the agent's rationality is unknown, showing that even with simplicity priors like Occam's razor, it is impossible to uniquely decompose a policy into planning and reward components, leading to high regret in distinguishing true decompositions.
Inverse reinforcement learning (IRL) attempts to infer human rewards or preferences from observed behavior. Since human planning systematically deviates from rationality, several approaches have been tried to account for specific human shortcomings. However, the general problem of inferring the reward function of an agent of unknown rationality has received little attention. Unlike the well-known ambiguity problems in IRL, this one is practically relevant but cannot be resolved by observing the agent's policy in enough environments. This paper shows (1) that a No Free Lunch result implies it is impossible to uniquely decompose a policy into a planning algorithm and reward function, and (2) that even with a reasonable simplicity prior/Occam's razor on the set of decompositions, we cannot distinguish between the true decomposition and others that lead to high regret. To address this, we need simple `normative' assumptions, which cannot be deduced exclusively from observations.