HI-TOM: A Benchmark for Evaluating Higher-Order Theory of Mind Reasoning in Large Language Models

TL;DR

This paper introduces HI-TOM, a benchmark for higher-order Theory of Mind reasoning in large language models, revealing current models' limitations in recursive mental state reasoning.

Contribution

The paper presents HI-TOM, the first benchmark specifically designed to evaluate higher-order ToM in LLMs, and provides an analysis of their performance shortcomings.

Findings

LLMs show decreased performance on higher-order ToM tasks

Current LLMs struggle with recursive reasoning about beliefs

The study highlights the need for improved models for complex social reasoning

Abstract

Theory of Mind (ToM) is the ability to reason about one's own and others' mental states. ToM plays a critical role in the development of intelligence, language understanding, and cognitive processes. While previous work has primarily focused on first and second-order ToM, we explore higher-order ToM, which involves recursive reasoning on others' beliefs. We introduce HI-TOM, a Higher Order Theory of Mind benchmark. Our experimental evaluation using various Large Language Models (LLMs) indicates a decline in performance on higher-order ToM tasks, demonstrating the limitations of current LLMs. We conduct a thorough analysis of different failure cases of LLMs, and share our thoughts on the implications of our findings on the future of NLP.