Theory of Mind (ToM) and motivation in Human-AI interaction
Skip to main content
eScholarship
Open Access Publications from the University of California

Theory of Mind (ToM) and motivation in Human-AI interaction

Creative Commons 'BY' version 4.0 license
Abstract

A previous study (Silacci et al. 2026) investigated the effectiveness of Simulated Exercising Peers (SEPs)-AI-powered virtual agents designed to provide social support in promoting physical activity. Participants have a weekly goal of 10.000 steps and their behavior is tracked using a custom mobile app exercise across different conditions: alone with no peer, with another human peer, with a human-like virtual peer or with a cyborg-like virtual peer. Results showed that AI simulated peers offered more consistent encouragement and stronger working alliances, whereas human peers were perceived as more authentic, with both supporting physical activity through complementary motivational pathways. In the present investigation, we assess Theory of Mind (ToM) through post-hoc semi-structured, validated interview (Bosco et al. 2016) administered to participants from the previous experiment. Our preliminary results show that greater performance in ToM correlates with the perceived relatedness experienced during the intervention across the conditions.