Effects of Perspective Taking on Ratings of Human Likeness and Trust

Abstract

The effects of perspective taking on ratings of human-likeness and trust are investigated. Seventy-four participants were shown pictures of two agents (human and robot) and storytelling narratives, which they had to complete. Afterwards, participants completed augmented versions of the Trust Scale and Human-Likeness Posttask Survey. Half of the participants were given stories using the perspective of the agent (perspective taking condition) and the other half was given stories using a third-person perspective (non-perspective taking condition). It was hypothesized that participants in the perspective taking condition would rate the agent higher on human-likeness and trust compared to the non-perspective taking condition. Interestingly, the results support our hypothesis for human-likeness but not for trust. The findings have important implications for the design of social robots by demonstrating the importance of perspective taking exercises on perception of humanness. Future studies need to validate the effects of perspective taking on human-robot interaction in various contexts and with different robot agents.