CORDIS - EU research results

Children and social robots: An integrative framework

Periodic Reporting for period 4 - CHILDROBOT (Children and social robots: An integrative framework)

Reporting period: 2021-07-01 to 2022-12-31

Robots used to be made for labor; now, they are increasingly also made for social relationships. Social robots can learn from us, teach us, play with us, and assist us. With the market for social robots expected to grow substantially in the next 20 years, social robots may become a life-changing technology. At the outset of the project, however, we still knew little about one of the most intriguing, relevant, and timely issues in this process – children’s interaction with social robots. Children are not only increasingly recognized and targeted as early adopters of new technologies; they may also be more susceptible to potential effects of interacting with robots than are adults.

As research on child-robot interaction (CRI) has traditionally been a fragmented field, the main aim of the project was to develop an integrative framework of CRI by synthesizing theories and concepts from communication research, human-robot-interaction, as well as psychology in a new way. Focusing on eight-to-nine year old children, the project studied (1) the antecedents of children’s acceptance of social robots; (2) the consequences of CRI for children’s learning of prosocial skills from social robots and their relationship formation with them; and (3) the processes that explain why such effects emerge. The project also provided new standardized measures for CRI that facilitate the study of crucial aspects of CRI among children.

The project has generally advanced our understanding of the antecedents, consequences, and underlying mechanisms of CRI and systematized and integrated existing knowledge on the topic. We now better understand why children do – and do not – accept robots and that robots can be role models for children’s prosocial behavior. We recognize, more than before, the crucial role of communication for child-robot relationship formation. And we now grasp better than before the important role that anthropomorphism, perceived similarity, and perspective taking play when children form relationships with social robots. Overall, then, the project not only broadened and deepened, but also enhanced and integrated our sense of the characteristics, underlying processes, and consequences of children’s interaction with social robots.
Based on integrative theoretical work, survey research, and multiple experiments on interactions between children and social robots, the project has produced several important results. First, advancing existing research, our studies have shown children’s acceptance of social robots is largely hedonically oriented, driven by attitudes and norms, and declining over time. Personality characteristics, anxiety toward robots, and utilitarian views of a robot do not seem to play a role. Second, shifting the dominant focus of existing research from prosocial behavior toward a robot to prosocial behavior by a robot, we demonstrated that robots, as role models, can enhance children’s prosocial behavior. Non-human actors, such as social robots, may thus have social functions that have traditionally been fulfilled by humans. Third, communicative processes such as a robot’s question asking, self-description, and self-disclosure matter for child-robot relationship formation. Transparency about a robot’s machine status – an aspect often neglected in research – reduced children’s willingness to form a relationship with it. Fourth, when we try to understand why children from relationships with robots, it is important to consider the extent to which children attribute human characteristics to robot, the degree to which children perceive robots as similar to themselves, and whether children think that robots can read their thoughts. The more children entertain these ideas when interacting with social robots, the more likely they are to form relationships with them. Finally, we demonstrated that important aspects of child-robot interaction can be assessed, even among relatively young children, with self-report measures in a reliable, valid, and efficient way. These measures are now increasingly used also by other researchers and not only contribute to improved and more standardized measurement in the field but also to more cumulative insights.

Our results have been published in academic outlets and been presented at conferences, both to academic and non-academic audiences. Although the research of the project has dealt with fundamental, basic questions, the implications of the results may be of interest to robot companies and roboticists as well as to those interested in whether and how to use robots among children.
Within the project’s home discipline, communication science, interactions between robots and children received little attention when the project started. Our project has gone beyond the state of the art in communication science and put children’s interaction with machine-type communication partners, such as social robots, more strongly on the agenda of communication research and related disciplines. More specifically, our studies on the antecedents of children’s acceptance of social robots have extended existing research by showing the advantages of formal modeling and a long-term perspective: We not only revealed a relatively simple structure of predictors of children’s acceptance of robots, but also its decay over time. Our studies on social robots as role models of prosocial behavior merges with a broader ‘non-human turn’ in the social sciences. Our research contributes to this reorientation both in communication research and the social sciences more broadly. In addition, our studies on the impact of transparent communication are instrumental in emphasizing the importance of responsible robotics. Our findings suggest that robot companies, researchers, and the public should pay more attention to an open, honest, and transparent discussion about what robots are and can do as well as about what robots are not and cannot do. Finally, our measures of crucial concepts in CRI have enhanced the methodological toolkit of researchers who want to study children’s use of social robots and its effects, while our systematization of the existing literature has been crucial for integrating a field that is traditionally diverse and dispersed.
Child playing with social robot