Expressions of Emotions in Virtual Agents: Empirical Evaluation

Human communication is mainly based on speech, gestures, facial and bodily expressions. Nonverbal expressions, as body posture and small movements, regulate social interactions. These results lead to an ever-growing effort from behalf of researchers in the human-computer interaction field to include emotions and emotional expressivity in their projects. Evaluation of emotion expressions recognition becomes, thus, an important step in the development of project involving interaction between humans and avatars. In the present paper we describe the results of an evaluation experiment made in two settings: discrimination between three videos looking for a specific emotion expression and identification task that asks for the selection of a list of a seen expression. The obtained results are similar to previous studies from the field and underline again the necessity of careful design of some particular emotions, especially those one that can be considered from the same family.