Abstract
Emotional contagion is defined as” the tendency to automatically mimic and synchronize facial expressions, vocalizations, postures, and movements with those of another person and, consequently, to converge emotionally.” When exposed to people of specific emotions, people will experience the same ones. Hatfield(1994) indicated that gender differences might exist in emotional contagion, that is, females will be more susceptible to emotional contagion. On the other hand, computers are social actors paradigm（Nass,1995）proposed that people will treat computers as real people, and the effects discovered in interpersonal relationships can also be applied in human-computer interaction (HCI). Most of emotion-related researches in HCI focus on different effects of computers’ displayed emotions on users; however, there is little research on the effects of positive or negative emotions displayed by computers on users, and no research on emotional contagion in HCI so far. This study aims to investigate if users will be susceptible to interface characters’ emotions and to exam the effects of gender differences. The independent variables of this study were interface character’s emotions (positive and negative) and subject’s sex (male and female), and the dependent variables were subjects’ emotional states, their assessments of the interface character, the performance of pictures assessments and recall and analysis of facial expression recordings. The results demonstrated that users would be susceptible to interface character’s emotions during interaction; after interacting with the interface character, subjects in positive emotion groups became more positive and subjects in negative ones more negative. Otherwise, subjects in positive/negative emotion groups would have different assessments of the perception, subject’s assessment of likeability, pleasure of interaction, satisfaction, willing to interact in positive groups would be more positive than in negative ones. This could be explained by mood-congruent processing (Bower, 1982) and affect infusion model (Forgas, 1995). However the interface character’s emotional states had no significant effects on subjects’ trust toward the interface character, interface character’s decisions and the quality of arguments, it’s the similarity between users and the interface character that had impacts on those three variables. When it comes to emotional contagion and gender differences, although the interaction effects were not significant in MBS and DES, it’s significant in SAM. In positive emotion groups, females’ emotional states were more positive than males, but the difference was smaller; in negative ones, females’ emotional states were more negative than males and the difference huger. When it comes to the perception, the interface character’s emotional states had no siginificant effects n subjects’ trust toward the interface character, interface character’s decisions and the quality of arguments either, but the interaction effects were significant on likability, pleasure of interaction and the willing to interact. In positive groups, the likability, pleasure of interaction and the willing to interact of females were more positive than males; in negative ones, those thress variables of females more neagative than males. According to the results of this study, we provided some directions and design suggestions for future researchers.