Theory of Mind for Explainable Human-Robot Interaction

Theory of Mind can make robots more user-friendly by helping them understand human mental states: The paper argues that when robots are equipped with Theory of Mind capabilities, they can better infer what users are thinking and adapt their behavior accordingly, making interactions more natural and predictable.

Current explainability methods focus too much on the AI system rather than what users actually need to know: The research identifies that existing Explainable AI (XAI) approaches often explain how the AI works internally, but don't prioritize explaining things in ways that match what individual users need or want to understand about the robot's decisions.

There's a major gap in verifying whether robot explanations match their actual reasoning: The paper highlights a critical problem—most robots using Theory of Mind don't actually verify whether their explanations to humans correspond to what's really happening inside the robot's decision-making process, which could undermine trust.

Combining Theory of Mind with explainability frameworks could shift focus to users: By integrating Theory of Mind principles into explainability standards (specifically the VXAI framework's seven criteria), robots could provide explanations tailored to each user's perspective and informational needs rather than generic technical descriptions.

Better human-centered explanations could improve how people interact with and trust robots: When robots understand users' mental states and explain themselves accordingly, interactions become more interpretable and predictable, potentially increasing user confidence and effective collaboration.