You, robot? The role of anthropomorphic emotion attributions in children’s sharing with a robot

• In Event: Do young children and non-human primates care about other people’s needs?

Abstract

Introduction. Sharing helps children form and maintain relationships with other children. A central aspect of sharing concerns the extent to which children anticipate the recipient to be happy with the shared resources (Decety & Svetlova, 2012; Eisenberg & Mussen, 1989). Yet, children born today interact not only with other children, but increasingly with robots as well. Moreover, while robots cannot experience happiness or gratitude in the same way humans do, children’s tendency to anthropomorphize and attribute such feelings to robots might open up the possibility that affective state attributions moderate children’s sharing towards robots as well. To investigate this postulate, we assessed children’s feelings about sharing with robots as well as their actual sharing behavior. Anthropomorphic appearance of the robots was varied, and we manipulated the extent to which children attributed the robots with affective states.
Hypotheses. It was hypothesized that more human-looking robots as well as robots that were described as having affective states would elicit more sharing behavior as well as more positive feelings about sharing.
Population and methods. Children (4-9 years old; n = 120) from the southeast of the Netherlands were recruited to participate. Participants were introduced to robots that varied in the extent to which they looked human-like. Children’s perceptions of the robot’s affective states were manipulated by explicitly demonstrating one robot as having feelings and the other one not using vignette videos. Subsequently, children’s sharing behavior towards both robots was measured in a mini dictator game. Further, children’s feelings about sharing with both robots were measured in an emotional judgment task.
Results. Children felt more positive about sharing with a robot that they ascribed with affective states than one without affective states, F(1,119) = 10.01, p = .002, η² = .08, 90% CI [.02, .16]. Children also shared more with the feeling rather than the unfeeling robot, F(1,119) = 4.58, p = .035, η² = .04, 90% CI [.00; .11]. These effects were the same across ages. An exploratory mediation analysis suggests that the increase in shared resources was due to children’s increased positive feelings about sharing. Importantly, no effect of anthropomorphic appearance on sharing or feelings about sharing was found.
Conclusion. In sum, across tasks, our results clearly indicate that children show an increase in other-oriented behavior towards robots when they believe it has affective states – i.e., when it can be happy with the shared resources. This novel finding contributes to a vivid discussion on the psychological mechanisms that underlie young children’s prosocial behavior (Dunfield, 2014; Eisenberg, 2000; Martin & Olson, 2015) as it provides fundamental evidence for the role of recipient affect in children’s sharing behavior in an experimentally controlled design. Furthermore, it significantly extends upon the literature on child-robot interactions.

Authors

• Sari RR Nijssen, Radboud Universiteit Nijmegen
  Presenting Author

• Barbara Müller, Radboud Universiteit
  Non-Presenting Author

• Tibor Bosse
  Non-Presenting Author

• Markus Paulus, Ludwig-Maximilians-Universität München
  Non-Presenting Author