We explored extracting users' true latent and implicit perceptions of robots by means of “Paul A. Weiss' Thought Experiment,” which considers the relationship between the elements, organization, and function of an object. We focused on three target objects appearing in a story: “computer;” “human being;” and “robot.” Participants were asked to answer the question “What was ‘lost’ as a result of the target object being homogenized/pulverized?” The results of a correspondence analysis on the answers for the three objects clarified the participants' ambiguous and contradictory perceptions of robots, that is, “robots are neither humans nor computers” but “robots are somehow similar to humans and computers.”