Robot Agreeableness and User Engagement in Verbal Human-Robot Interaction

It is a well-established fact in the field of Human-Robot Interaction that the personality a social robot is endowed with (i.e., the one it expresses by its gestures and words) plays a key role towards the engagement experienced by any person interacting with it. In this article, we address the problem of designing the conversational style of a social robot to convey different degrees of Agreeableness, which is one of the Big Five personality traits, and investigate the relation between the person's engagement and the robot's degree of agreeableness. We propose to use subjective perception of time elapsed as an indicator for engagement and design two robot behaviours, one making the robot always agree with the person's opinion on topics of discussion (the compliant behaviour) and one making the robot always disagree (the contrasting behaviour). In an experiment involving 14 participants, we assess whether the robot adopting the contrasting behaviour is perceived as more engaging than the one adopting the compliant behaviour, and use participants' previous interactions with vocal assistants as a reference for estimating the naturalness, enjoyability and politeness of the conversation with the robot. Results suggest that designing social robots that are able to disagree with the person they're interacting with might be key to make them more engaging and entertaining.