Modeling Social Cues: Effective Features for Predicting Listener Nods

Human communication involves a number of verbal, vocal, and non-verbal cues that facilitate face-to-face communication. Building a computational understanding of how these cues facilitate social and cognitive processes enables new research directions in social and cognitive sciences and the design of socially interactive systems such as agents and robots. In this paper, we present preliminary work in modeling a particular communicative mechanism—listener nods—toward designing humanlike agents and robots that show appropriate social behavior.∗