Social robots with a theory of mind (ToM): Are we threatened when they can read our emotions?

How would human users react to social robots that possess a theory of mind (ToM)? Would robots that can infer their users’ cognitions and emotions threaten their sense of uniqueness and evoke other negative reactions because ToM is a uniquely human trait? If so, can we alleviate these negative user reactions by framing robots as members of our ingroup? We addressed these questions with a 3 (robot’s affective ToM: correct vs. incorrect vs. control) × 2 (robot’s group membership: ingroup vs. outgroup) × 2 (user gender: female vs. male) between-subjects online experiment. Participants were asked to complete an online task with a robot named Pepper that was identified as an ingroup member or outgroup member. They first read a passage describing a past user’s interaction with Pepper, in which the user expressed sarcasm and Pepper correctly or incorrectly identified the user’s sarcasm or made a neutral comment. Males reacted more negatively to Pepper that correctly identified sarcasm and reported lower expected enjoyment with Pepper than females. Ingroup Pepper made participants feel closer to the robot but also threatened their sense of uniqueness than the outgroup Pepper. Design implications for fostering better human-robot interaction (HRI) are discussed.