Abstract — In this paper we propose to explore how a group of psychology students perceives two different instances of social robots (mechanic vs human) and how this affects their intention to work with a social robots in the future. Two studies were conducted. A multivariate analysis of variance and multiple regression analysis were employed. Results show no effects for robots appearance. Previous attitudes towards robots predict anthropomorphism. Anthropomorphism and desire show up as predictors of intention to work with robots. Keywords – Social robots, anthropomorphism, attitudes, desire, intentions. I. INTRODUCTION Robots are starting to be perceived not only as precise hardworking tools, but also as partners that can provide us with company, education and health care [1]. This acts of caring are generally seen as something “special”, and as social robots enter this field, it is vital to develop methodologies that help us understand how they will be perceived by all the actors involved. With this in mind, the present study proposes to explore how a group of psychology students perceives social robots, given different appearances (mechanic vs human) and qualities (anthropomorphism, likeability and animacy), and how this will influence their intention, in the future, work with a social robot. The results obtained from this sample can be informative about how a professional class foresees the future of its profession. A. Definition of social robots For the purposes of this study, it will be used the definition of social robot given by Fong, Nourbakhsh & Dautenhahn [2]. A social robot is a robot that expresses and perceives emotions, communicates with high-level dialogue, learns and recognizes models of other agents (humans or robots), establishes and maintains a social relationship, uses natural cues (gaze, gestures, etc.), exhibits distinctive personality and character and may learn and develop social skills. B. Robot appearance There is a generalized confidence that building robots with human features and behaviors will enhance their interactions with humans. In fact various studies have identified that variables like the robots voice [3 - 5], smile N. Piçarra, is with the FCHS of the University of the Algarve, Portugal (e-mail: a2657@ualg.pt). J. C. Giger, is with the FCHS of the University of the Algarve, Portugal (e-mail: jhgiger@ualg.pt). G. Pochatk, is with the Institute of Psychology, Polish Academy of Sciences, Poland (e-mail: grzegorz.pochwatko@psych.pan.pl). G. Gonçalves, is with the FCHS of the University of the Algarve, Portugal (e-mail: ggoncalves@ ualg.pt). [6], face traits [5, 7], gestures [8], presence of limbs [9], height [10, 11], whole body shape (mechanical vs humanoid) [9, 12, 13] and gender [14] have an influence in the perception of the robots intelligence, competence and friendliness. People also seem to take into account robot appearance when asked what jobs a robot should perform [15 - 18]. Nevertheless the question of appearance is not so straight forward, as some research results point to some drawbacks in using humanoid robots. For example, people who thought of robots as having human form, showed greater increases in blood pressure readings and negative emotions, after having their pressure measured by a robot, than those who thought of mechanical robots [19]. And although people generally see humanoid robots as more acceptable for house chores, they feel uncomfortable with the idea of social interactions with them [20, 21]. One possible explanation for this discomfort with humanoid robots has been termed the uncanny valley effect. Presented by Masahiro Mori in the early 70’s, it stated that as a robot increases in human appearance so does likeability, until a point where its similarity induces feelings of eeriness and dread. This lead authors to question if robots really need to look like humans [22, 23] and opened research on the causes underlying the uncanny valley effect [24 - 27], with research results suggesting a complex link between human appearance and eeriness. Recent research has framed the uncanny valley effect in evolutionary [28], and developmental terms [29]. Undoubtedly appearing and behaving in a human like fashion is an important factor for robots in HRI, but a factor that is mediated by the socio-psychological characteristics of the human user. C. Anthropomorphism Although frequently used in research as a synonymous of human resemblance, anthropomorphism is not a measure of appearance, but instead, a natural and automatic process of inductive inference, of reasoning about nonhuman agents, attributing them humanlike psychological traits, emotions, intentions, motivations and goals [30]. According to Epley, Waytz and Cacioppo, this attribution of human character and intentions to nonhuman agents is the result of the interplay of three factors: sociality (desire for social contact and affiliation), effectance (need to effectively interact and make sense of one’s environment) and elicited agent knowledge (accessibility and applicability of anthropocentric knowledge). Variations on the level of these three factors are the result of dispositional, situational, developmental and cultural variables [30]. When reasoning about non human agents, the level of anthropomorphism will be a function of, the available homocentric knowledge, its applicability and The Robot Got the Looks. Social Robots Appearance as Predictor of Anthrpomorphism and Intention to Work. N. Piçarra, J. C. Giger, G. Pochwatko and G. Gonçalves Workshop on Motivational Aspects of Robotics in Physical Therapy. IROS 2012, October 7-12, 2012, Vilamoura, Algarve [Portugal]