A Systematic Review of Prosocial Decision-Making in Human-Embodied Agent Interaction

Abstract

Embodied agents are increasingly present in daily life, creating situations in which humans and agents compete for limited resources and humans may choose to benefit agents at a personal cost, referred to as prosocial behavior. However, research lacks a comprehensive synthesis of prosocial decision-making across embodied agent types. Following PRISMA guidelines, this systematic review analyzed 24 empirical studies published between 2010 and 2025. Studies were synthesized by agent type, conflict type, interaction scenario, and factors influencing prosocial behavior. Most examined conflicts over limited space in realistic settings, whereas fewer investigated resource allocation or game-theoretic scenarios. In Human–Social Robot Interaction, prosocial behavior was primarily shaped by agents’ social cues, suggesting that people treat social robots as social actors. In Human–Autonomous Vehicle Interaction, prosocial behavior was mainly driven by cost-benefit considerations, indicating more utilitarian decision-making. These findings inform the design of embodied agents that promote prosocial human–agent interactions.