Which ostensive stimuli can be used for a robot to detect and maintain tutoring situations?

Abstract

In developmental research, tutoring behavior has been identified as scaffolding infants' learning processes. Infants seem sensitive to tutoring situations and they detect these by ostensive cues [4]. Some social signals such as eye-gaze, child-directed speech (Motherese), child-directed motion (Motionese), and contingency have been shown to serve as ostensive cues. The concept of contingency describes exchanges in which two agents interact with each other reciprocally. Csibra and Gergely argued that contingency is a characteristic ostensive stimulus of a tutoring situation [4]. In order for a robot to be treated similar to an infant, it has to both, be sensitive to the ostensive stimuli on the one hand and induce tutoring behavior by its feedback about its capabilities on the other hand. In this paper, we raise the question whether a robot can be treated similar to an infant in an interaction. We present results concerning the acceptance of a robotic agent in a social learning scenario, which we obtained via comparison to interactions with 8–11 months old infants and adults in equal conditions. We applied measurements for motion modifications (Motionese) and eye-gaze behavior. Our results reveal significant differences between Adult-Child Interaction (ACI), Adult-Adult Interaction (AAI) and Adult-Robot Interaction (ARI) suggesting that in ARI, robot-directed tutoring behavior is even more accentuated in terms of Motionese, but contingent responsivity is impaired. Our results confirm previous findings [14] concerning the differences between ACI, AAI, and ARI and constitute an important empirical basis for making use of ostensive stimuli as social signals for tutoring behavior in social robotics.