Satisficing Mentalizing: Bayesian Models of Theory of Mind Reasoning in Scenarios with Different Uncertainties

TL;DR

This paper introduces Bayesian models for Theory of Mind reasoning that balance accuracy and computational efficiency, evaluated against human data under various uncertainties, with a novel switching approach for satisficing mentalizing.

Contribution

It presents a switching Bayesian approach that combines simplified models to improve mentalizing efficiency over traditional full Bayesian models.

Findings

Switching Bayesian models outperform full models in efficiency.

Models align well with human behavior under uncertainty.

Satisficing models provide a practical balance between accuracy and computational cost.

Abstract

The ability to interpret the mental state of another agent based on its behavior, also called Theory of Mind (ToM), is crucial for humans in any kind of social interaction. Artificial systems, such as intelligent assistants, would also greatly benefit from such mentalizing capabilities. However, humans and systems alike are bound by limitations in their available computational resources. This raises the need for satisficing mentalizing, reconciling accuracy and efficiency in mental state inference that is good enough for a given situation. In this paper, we present different Bayesian models of ToM reasoning and evaluate them based on actual human behavior data that were generated under different kinds of uncertainties. We propose a Switching approach that combines specialized models, embodying simplifying presumptions, in order to achieve a more statisficing mentalizing compared to a Full Bayesian ToM model.