Automatic processing of irrelevant co-speech gestures with human but not robot actors

Abstract

Non-verbal, or visual, communication is an important factor of daily human-to-human interaction. Gestures make up one mode of visual communication, where movement of the body is used to convey a message either alone or in conjunction with speech. The purpose of this experiment is to explore how humans perceive gestures made by a humanoid robot compared to the same gestures made by a human. We do this by adapting and replicating a human perceptual experiment by Kelly et al., where a Stroop-like task was used to demonstrate the automatic processing of gesture and speech together. 59 college students participated in our experiment. Our results support the notion that automatic gesture processing occurs when interacting with human actors, but not robot actors. We discuss the implications of these findings for the HRI community.