Wall-E vs. Terminator: The relationship between physical appearance and dimensions of mind perception

Social robots are increasingly integrated into everyday environments, yet effective human-robot interaction remains challenging as robots often fail to engage social cognition the same way human partners do. Here, we examined the degree to which physical features impact the degree to which robots are being perceived as agents ‘with a mind’. For that purpose, we took all robot stimuli currently contained in the Anthropomorphic Robot Database (ABOT; n = 251) and examined the degree to which they trigger mind perception (N = 300). Consistent with prior findings, robots were generally attributed more agency (capacity to act and plan) than experience (capacity to sense and feel). Body components significantly explained variance in the perception of agency, especially in interaction with facial features; for experience, on the other hand, variance was significantly explained by body components, face and surface details. Increasing human-likeness boosted perceptions of both dimensions, but the trajectories differed: experience followed a cubic function, plateauing at medium levels of human-likeness, while agency followed a quartic function with a dip around 75% human-likeness—a pattern resembling the un-canny valley. Our results indicate that in order to create specific levels of agency (e.g., for surgery robots) and experience (e.g., for companion robots), a general increase of human-likeness might not suffice; instead, targeted design of individual physical features is crucial for convey-ing specific mental capacities. Our ratings linking physical robot features to dimensions of mind perception are made available, offering a comprehensive and accessible resource for researchers to experimentally manipulate perceptions of agency and experience.