When a feature becomes a bug: evolutionary mismatch in human–robot interaction

As robots become increasingly integrated into everyday life, evolved psychological mechanisms originally adapted for human social interaction may shape responses to artificial agents in ways that are no longer functionally appropriate. Drawing on an evolutionary psychology framework, the present study examined how anthropomorphism, empathy, and systemizing influence decisions involving a humanoid home robot. Participants ( N = 519) completed an online experiment in which they viewed one of six robot stimuli varying in appearance (human-like, ambiguous, or robotic) and gender (female or male). They then read a scenario in which the manufacturer instructed them to reset the malfunctioning robot, thereby erasing its memories, while the robot pleaded not to be deactivated. Participants rated their willingness to deactivate the robot and the anticipated discomfort associated with doing so, and subsequently completed measures of anthropomorphism, empathy, and systemizing. Two 2 × 3 general linear models showed that stronger anthropomorphic tendencies predicted lower willingness to deactivate the robot, whereas higher systemizing predicted greater willingness. Empathy did not predict the decision itself but was associated with greater anticipated discomfort, together with anthropomorphism, while systemizing predicted lower discomfort. Contrary to our expectations, the decision to deactivate the robot did not vary as a function of its appearance. However, robot appearance influenced anticipated discomfort through interaction effects. These findings suggest that distinct evolved psychological mechanisms differentially shape behavioral and emotional responses to humanoid robots, consistent with the notion that evolutionary mismatch is at play human–robot interactions.