Abstract
Humanoid robots occupy an ambiguous position between social partners and artefacts, and people differ markedly in how they evaluate them. Research on individual differences in human–robot interaction has focused on personality, anthropomorphic tendencies and technology acceptance, leaving largely unexamined the processes by which people decide whether an artificial agent can be treated as a credible source of social information. The present study introduces epistemic stances, as conceptualised within the mentalizing tradition, to this literature. Seventy adults completed the Epistemic Trust, Mistrust and Credulity Questionnaire, along with measures of generalized attitudes toward robots (Frankenstein Syndrome Questionnaire, Negative Attitudes toward Robots Scale, Robotic Social Attributes Scale), and a morph-based task in which robot faces varying in human likeness were rated for likeability, eeriness and human likeness. Epistemic stances were selectively associated with generalized social evaluations: Epistemic Trust predicted higher Robot Anxiety and lower attributed Warmth, whereas Credulity predicted higher attributed Warmth, with no associations for perceived Competence. At the stimulus level, morph condition strongly shaped all three judgements, whereas epistemic stances added little. Rather than a failure to extend person-level effects, this dissociation follows from the claim that epistemic stances operate on socially communicated information rather than on perceptual appraisal. Epistemic functioning may therefore shape the social meaning attributed to artificial agents, offering a testable framework for the design of robot communication.