Leveraging Neurophysiological Information to Augment Interpretation of Responses to Vulnerable Robot Behaviors

Abstract

Previous human-robot interaction (HRI) research has shown that trust, disclosure, and companionship may be influenced by a robot's verbal behavior. Measures used to interpret these key aspects of HRI commonly include surveys, observations, and user interviews. In this preliminary work, we aim to extend previous research by exploring the use of electroencephalography (EEG) to augment our understanding of participants' responses to vulnerable robot behaviors. We tested this method by obtaining EEG data from participants while they interacted with a robotic tutor. The robotic tutor was designed to exhibit high vulnerability (HV) or low vulnerability (LV) behaviors similar to a previous HRI study. Our preliminary results show that event-related potentials (ERPs) may provide insights into participants' early affective processing of vulnerable robot behaviors.