Figure 1: The agent-experimenter loop and its formalization as POMDP together with the inference problems from the agent’s point of view and from the experimenter’s point of view. This conceptual framework is the basis of our inverse optimal control method.