With the biannual conference on Affective Computing and Intelligent Interaction that aims at the intersection of research on human emotion and computational models of emotion, research on emotion both in humans and virtual entities found an important platform for interdisciplinary exchange. This year the 5th conference was held in Geneva, Switzerland, on September 2–5.
On Thursday, September 4, CrossWorlds fellow Benny Liebold gave a talk on “Multimodal emotion expression of virtual agents. Mimic and vocal emotion expressions and their effects on emotion recognition”. The talk addressed human recognition of emotion expressions of virtual agents. It was found that our ability to recognize emotional states from virtual agents is worse when virtual agents use unimodal expressions of emotions compared to multimodal expressions. Even more interestingly, unimodal expressions of emotions together with neutral nonverbal cues are recognized worse than isolated unimodal expressions of emotions. The latter might be a result of users perceiving the neutral expression of virtual agents as a relevant component of the virtual agent’s emotional state. Further results can be obtained from the conference proceedings, which are going to be indexed in IEEE Xplore.
A total number of 55 papers covered a vast area of research from psychological research in human emotions to computational models of emotional processes, such as empathy and automated emotion recognition. We thank the Computer Vision and Multimedia Laboratory and the Swiss Center of Affective Sciences from Geneva University, who did a great job in organizing the conference, putting together a great conference program including the very interesting keynote speakers.
Comments