Do infants perceive moral structure in geometric-agent interactions?

From a developmental perspective, infants’ understanding of agents and their goals progresses from a teleological stance, in which actions are interpreted as goal-directed and efficient within situational constraints, to a mentalistic stance that involves attributing beliefs and intentions (e.g., Gergely & Csibra, 2003). If geometric agents are to serve as valid proxies for social actors, including those involved in moral interactions, infants must first perceive their movements as structured interactions rather than purely physical motion. Rochat et al. (1997) showed that 3- to 6-month-old infants differentiate between interdependent and independent motion between disc shapes. This finding established a foundational understanding of what Premack (1990) called the theory of self-propelled objects and, by extension, the basis for interpreting geometric-agent interactions. Premack and Premack (1997) demonstrated that preverbal infants distinguish prosocial from antisocial interactions, and Geraci et al. (2025) found that even 5-day-old newborns preferentially look toward prosocial over antisocial interactions between abstract geometric agents. Within this line of research, nonsocial geometric motion has often been used as a control or baseline (e.g., Kanakogi et al., 2017, 2022; Biondi et al., 2021), yet few have directly compared such motion with moral events within the same preferential-looking paradigm. To address this methodological and theoretical gap, the present study aims to examine whether preverbal infants represent geometric-agent interactions in terms of an underlying moral structure, extending beyond what can be explained by purely physical (or non-social) interpretations. Specifically, we will use a preferential-looking paradigm: infants will view paired animations where an agent will hit or caress either another agent or a non-agent ring shape. Differential looking towards the two displays will indicate whether infants treat the former as morally meaningful interactions rather than merely physical actions. In the social animation, one geometric agent will either hit (Antisocial) or caress (Prosocial) another agent. In the neutral animation, the same actions will be directed toward a non-agent ring shape. We predict that infants will preferentially look longer at morally structured interactions, suggesting that they perceive these events as meaningful beyond their physical properties.