Rating of Dog Breed Differences

摘要

Dogs have attracted the attention of biomimetic roboticists, for example, Sony’s Aibo. Properly designed robots may be able to elicit similar perceptions as dogs. It is important then to analyze the attitudes that people have towards dogs, in particular salient features like breed-membership. In the present paper we developed an electronic questionnaire to assess the existing mental models and categorizations of dog breeds. Subjects were asked to rate representatives of 48 different dog breeds with a 7-point Likert Scale for specific behavioral traits. Based on exploratory data analysis techniques, we found four clusters of breeds: Cluster 1 (Higher Drive, Useful, and Nice), Cluster 2 (Higher Drive, Useful, and not so Nice), Cluster 3 (Higher Drive, not so Useful, and not so Nice) and Cluster 4 (Lower Drive, not so Useful, and Nice). Implications for quadropedal robot design based on pertinent existing perceptions of dog breeds and their traits are discussed.