A Systematic Review on the Evaluation of Large Language Models in Theory of Mind Tasks

TL;DR

This systematic review assesses how large language models (LLMs) are evaluated on Theory of Mind (ToM) tasks, clarifying the taxonomy, benchmarks, and prompting strategies used to probe mental-state reasoning. It synthesizes benchmarks anchored in the ATOMS framework, discusses situatedness and multimodal evaluations, and analyzes metrics from accuracy to human-in-the-loop assessments. The authors highlight emergent ToM in LLMs but emphasize persistent gaps, data biases, and evaluation challenges that can inflate or obscure true capabilities. Overall, the work provides a critical map of current methods and pragmatic guidance for building robust, multimodal ToM benchmarks and evaluations in AI systems.

Abstract

In recent years, evaluating the Theory of Mind (ToM) capabilities of large language models (LLMs) has received significant attention within the research community. As the field rapidly evolves, navigating the diverse approaches and methodologies has become increasingly complex. This systematic review synthesizes current efforts to assess LLMs' ability to perform ToM tasks, an essential aspect of human cognition involving the attribution of mental states to oneself and others. Despite notable advancements, the proficiency of LLMs in ToM remains a contentious issue. By categorizing benchmarks and tasks through a taxonomy rooted in cognitive science, this review critically examines evaluation techniques, prompting strategies, and the inherent limitations of LLMs in replicating human-like mental state reasoning. A recurring theme in the literature reveals that while LLMs demonstrate emerging competence in ToM tasks, significant gaps persist in their emulation of human cognitive abilities.

Figures (1)

• Figure 1: Comparison of the number of papers inspecting different LLMs (left) and modalities (right).