Search

Computer-generated characters, commonly used as avatars and agents, present a broad range of appearance and types. The growing body of research on avatars and agents does not yet provide a categorical and dimensional framework for understanding people’s perception of computer characters. Two studies were conducted to explore people’s categorization of computer characters and dimensions used in the categorization. In Study 1, the participants (N = 40) sorted the images of 70 computer-generated characters into groups. Cluster analysis revealed that human and nonhuman emerged as two large types. Group labels provided by the participants were analyzed to help interpret subcategories within each of the two clusters. In Study 2, a different sample of participants (N = 273) rated the same 70 characters on a set of attributes. The ratings were used in regression analysis to help interpret the multidimensional scaling analysis of the sorting data in Study 1. Four dimensions were identified: humanness, graphic details and stylization, good-natured vs. bad-natured, and gender. The results provide an initial categorical and dimensional framework for further research on perception and effects of computer characters. ..PAT.-Conference Proceeding [ABSTRACT FROM AUTHOR]