Examining the effects of social comparison by assimilation and contrast of agents

Abstract

This study aimed to investigate the effects of social comparison and the effects of assimilation and contrast in human-agent interaction. The experiment was conducted in four conditions: two conditions that induced assimilation and contrast, and two conditions in which the agent was more or less capable than the human. Assimilation and contrast were manipulated by assigning participants a task and a chat with an agent. Social comparison effect was assessed using a typing task and a questionnaire. We confirmed that social comparison was performed, and that assimilation and contrast were performed in the agents. Self-evaluations differed between conditions in which the agents had higher or lower abilities than humans. In the condition where the agent induced assimilation and had lower ability than humans, the typing miss rate was lower than in the condition where the agent had higher ability.