Social Robots Are Like Real People: First Impressions, Attributes, and Stereotyping of Social Robots

Abstract

Research shows that particular social robots are effective in interactions with people, from teaching young children to supporting older adult’s healthcare. There is little research, however, that examines how people form the interpersonal impressions of social robots that influence whether these interactions are effective or not. We report two studies that look at the differences between social robots with the goal of identifying those impressions. In Study 1, we examined 10 years of published articles about social robots and identified 342 robots for coding on a variety of attributes, from perceived gender to surface textures and their range of motion. We asked participants (N = 4,415) to evaluate each robot along these attributes. A complete catalog of the robots is posted here, and an overall description of the sample is included in the article. In Study 2, we asked people (N = 3,920) to evaluate the 342 robots and found that two impressions of the robots, competence and warmth, were as important for the evaluation of robots as they have been for perceptions of people in prior literature. Using the 21 attributes identified in Study 1, we found that the best predictors of robot competence were a robot’s mobility (degrees of freedom) and surface textures, while age, the absence of mechanical features, and mobility were the best predictors of robot warmth. We end with comments about the design and success of social robots in assistive applications.