Robots are becoming increasingly prevalent in homes, schools, workplace, and in other public locations. However, despite the fact that people will soon likely interact with multiple robots at once, most research in human-robot interaction (HRI) focuses on dyadic interaction (one human and one robot). Research in social psychology finds that intergroup interaction (in which the perceiver is part of the group or the targets perceived are part of a group) differs from one-on-one interaction and is often more negative. In this dissertation, we sought to determine if these group effects extended to HRI.

In a first line of research (Studies 1-3), we manipulated Number of Robots (Single, Group) and Robot Appearance (Study 1) and Behavior (Studies 1-3). Participants observed robots in videos (Study 1) or took part in a lab (Study 3) or field (Study 2) experiment in the United States (Studies 1-3) and Japan (Studies 2-3). Results indicated that participants have different perceptions of robot groups compared with individual robots and that these perceived differences depend on robot appearance and behavior.

In a second line of research (Study 4), we examine how human groups, compared with individuals, interact with a humanoid robot in a public setting. The robot was set up in the mall and provided mall guidance directions to visitors who approached. Results indicated that human groups were more likely than individuals to approach the robot, especially if the group was cohesive.

In a third line of research (Study 5), we examine how people respond to ingroup and outgroup robots in comparison to ingroup and outgroup humans. Participants played a game in teams of two humans and two robots against two humans and two robots. The study included explicit measures of perceptions of players and a behavioral measure of aggression. Results indicated that participants favored ingroup members over outgroup members more strongly than they favored humans over robots.

Together, these studies show that group effects exist in HRI and often parallel those in social psychology. Future research should examine edge cases regarding when group effects occur (e.g., when are robots not social enough to cue group effects). Robot designers should take this into account as they create robots for everyday use.