Spontaneous coordination with self-commitment: How the presence of others alters the strength, goal and timing of commitment

Abstract

Commitment is a paradoxical feature of human behavior, often seen as both an irrational bias and a virtue for achieving goals. This study investigates its social roots, revealing how social contexts shape the strength, content, and timing of self-commitment, even in individual tasks. Through a series of game-like experiments, participants pursued one of two equally desirable goals via sequential actions under varied social conditions: alone in a private room (Experiment 1), alongside an optimal reinforcement learning (RL) agent (Experiment 2) or another human (Experiment 3) on a shared display, or alone with a mere passive observer present (Experiment 4). Our results demonstrate that (1) all social contexts consistently heightened self-commitment, underscoring its sensitivity to the public nature of tasks; (2) in parallel-play settings (Experiments 2 and 3), participants spontaneously inferred others' intentions and avoided selecting the same goal, despite instructions that such avoidance was unnecessary, suggesting that theory-of-mind (ToM) inference of another agent is spontaneously evoked to bias goal selection; and (3) Bayesian ToM modeling indicated that participants delayed revealing their intentions in parallel-play settings but not in the mere-presence condition, implying that spontaneous bargaining with a potential partner, rather than mere social presence, prompts more cautious commitment formation. These findings illuminate that, even in individual tasks, self-commitment is deeply intertwined with social context, influencing how people manage their goals and interactions with others.