GPT-4o Lacks Core Features of Theory of Mind
This work addresses the problem of understanding the cognitive capabilities of LLMs for researchers and developers, revealing that their social proficiency is incremental and not due to genuine ToM.
The researchers tackled the question of whether Large Language Models (LLMs) possess a Theory of Mind (ToM) by developing a new evaluation framework based on a cognitively-grounded definition, and found that LLMs fail at logically equivalent tasks and exhibit low consistency, indicating they lack a domain-general or consistent ToM despite approximating human judgments in simple paradigms.
Do Large Language Models (LLMs) possess a Theory of Mind (ToM)? Research into this question has focused on evaluating LLMs against benchmarks and found success across a range of social tasks. However, these evaluations do not test for the actual representations posited by ToM: namely, a causal model of mental states and behavior. Here, we use a cognitively-grounded definition of ToM to develop and test a new evaluation framework. Specifically, our approach probes whether LLMs have a coherent, domain-general, and consistent model of how mental states cause behavior -- regardless of whether that model matches a human-like ToM. We find that even though LLMs succeed in approximating human judgments in a simple ToM paradigm, they fail at a logically equivalent task and exhibit low consistency between their action predictions and corresponding mental state inferences. As such, these findings suggest that the social proficiency exhibited by LLMs is not the result of a domain-general or consistent ToM.