Does Embodiment and Interaction Affect the Adoption of the Intentional Stance Towards a Humanoid Robot?

Abstract

Humanoid robots are a useful research tool to understand which factors trigger the adoption of the Intentional Stance (i.e., the attribution of mental states). The InStance Test (IST), assesses this with the isolated robot subscale and the social robot subscale - where higher scores reflect a greater tendency to adopt the Intentional Stance. Previous work found that when a human interacts with an embodied robot exhibiting human-like behaviour, their IST scores in-crease. However, it is not yet known whether this effect is robust to changes in embodiment or interaction. Subsequently, we administered both subscales before and after participants witnessed a humanoid robot exhibiting human-like behav-iour. There was a significant increase in both scales after participants (1) inter-acted with the embodied robot, (2) watched a video of the robot interacting with a human and (3) watched a video of the robot without interaction. Secondly, to check there were no differences across the human-like experiments, we compared the differences of the magnitude of the IST score increase. We found no signifi-cant differences across the experiments. Additionally, we conducted follow-up control analyses’ with the robot exhibiting machine-like behaviour, to ensure the increases were not due to exposure effects. We found no significant difference in the IST after participants (1) interacted with the embodied robot or (2) watched a video of the robot interacting with a human. These results show that the in-creased IST scores after witnessing iCub behaving in a human-like way, is not due to mere exposure and is robust to changes in embodiment and interaction.