Attributing Intentionality to Artificial Agents: Exposure Versus Interactive Scenarios

Recent studies suggest that people can interact with robots as social agents. However, it is still unclear what mental processes people rely on when interacting with robots. One core process in social cognition is the adoption of intentional stance, a strategy that humans use to interpret the behavior of others with reference to mental states. In this work, we sought to examine how the adoption of intentional stance may be modulated by the type of behaviors exhibited by a virtual robot and the context in which people are exposed to it. We developed an interactive virtual task and used the InStance Test to measure the attribution of intentionality to the robot. Our results show that participants attributed more intentionality to the virtual robot after interacting with it, independently of the type of behavior. Leveraging data from a previous study, we also show this increase is stronger than in a non-interactive, purely observational scenario. This study thus improves our understanding of how different contexts can affect the attribution of intentional stance and anthropomorphism in Human-Robot Interaction.