LLMs achieve adult human performance on higher-order theory of mind tasks

Journal article

This paper examines the extent to which large language models (LLMs) are able to perform tasks which require higher-order theory of mind (ToM)—the human ability to reason about multiple mental and emotional states in a recursive manner (e.g., I think that you believe that she knows). This paper builds on prior work by introducing a handwritten test suite—Multi-Order Theory of Mind Q&A—and using it to compare the performance of five LLMs of varying sizes and training paradigms to a newly gathered adult human benchmark. We find that GPT-4 and Flan-PaLM reach adult-level and near adult-level performance on our ToM tasks overall, and that GPT-4 exceeds adult performance on 6th order inferences. Our results suggest that there is an interplay between model size and finetuning for higher-order ToM performance, and that the linguistic abilities of large models may support more complex ToM inferences. Given the important role that higher-order ToM plays in group social interaction and relationships, these findings have significant implications for the development of a broad range of social, educational and assistive LLM applications.

Authors: Street, W; Siy, JO; Keeling, G; Baranes, A; Barnett, B

• Publication status: Published
• Peer review status: Peer reviewed
• Publisher: Frontiers Media
• Journal: Frontiers in Human Neuroscience
• Volume: 19
• Article number: 1633272
• Publication date: 2026-01-02
• Acceptance date: 2025-10-21
• DOI: 10.3389/fnhum.2025.1633272
• EISSN: 1662-5161
• ISSN: 1662-5161
• Language: English
• Keywords: mentalizing; social AI; large language models; AI; social cognition; theory of mind