Rasmus Overmark

A growing body of work attempts to evaluate the theory of mind (ToM) abilities of humans and large language models (LLMs) using static, noninteractive question-and-answer benchmarks. However, theoretical work in the field suggests that first-personal interaction is a crucial part of ToM and that such predictive, spectatorial tasks may fail to evaluate it. We address this gap with a novel ToM task that requires an agent to persuade a target to choose one of three policy proposals by strategically revealing information. Success depends on a persuader's sensitivity to a given target's knowledge states (what the target knows about the policies) and motivational states (how much the target values different outcomes). We varied whether these states were Revealed to persuaders or Hidden, in which case persuaders had to inquire about or infer them. In Experiment 1, participants persuaded a bot programmed to make only rational inferences. OpenAI's reasoning model o3 excelled in the Revealed condition but performed no better than chance in the Hidden condition, suggesting difficulty with the multistep planning required to elicit and use mental state information. Humans performed moderately well in both conditions, indicating an ability to engage such planning. In Experiment 2, where a human target role-played the bot, and in Experiment 3, where we measured whether human targets' real beliefs changed, o3 outperformed human persuaders, significantly so in the Revealed conditions. These results suggest that effective persuasion can occur without explicit ToM reasoning (e.g., through rhetorical strategies) and that o3 excels at this form of persuasion. Overall, our results caution against attributing human-like PToM to LLMs based on predictive, spectatorial benchmarks while highlighting their potential to influence people's beliefs and behavior.

Cognitive development research distinguishes between what children know (competence) and their ability to demonstrate their knowledge (performance). An experiment can fail to reveal a child’s competence if its design limits the child’s performance. This distinction allows researchers to design experiments that limit the impact of performance factors, leading to observations of competence earlier in cognitive development. The distinction is often used in a deflationary way, where performance factors are taken to be extraneous to the competence of interest, so that they are irrelevant to a theory of competence. Using the role of children’s pragmatic inferences in mentalizing development as a case, I argue that this is sometimes an implausible use of the distinction. Performance factors do not necessarily support deflationary explanations and so are not always extraneous. I develop the concept of a learning procedure which specifies the attentional and inferential constraints that govern the construction of children’s developing competences. Changes in children’s pragmatic inferences can critically reshape learning procedures with developmental consequences for mentalizing. This case has the wider implication that performance factors split into two kinds. Synchronic performance factors are extraneous to competence and track competence at a specific time. Diachronic performance factors are developmentally entangled with competence and affect competence over developmental time. I outline how these two kinds of performance can be teased apart empirically.