Evaluating Emotional Responses to Experimental Video Art

Abstract

Video is a medium that conveys information capable of reflecting multiple emotions to the viewers. Experimental video, a new contemporary art form, is a complex stimulus that can induce diverse emotions in its audience. This study exposed 22 individuals to two experimental videos, each comprising multiple sequences of experimental images and sounds. We asked participants to record their facial expressions while watching the videos using web cameras. We analysed the recorded footage using DeepFace and Facial Emotion Recognition (FER) algorithms to estimate the seven universal emotions. The Deepface algorithm demonstrated higher sensitivity than FER. We performed time analysis to detect possible visual events that might trigger emotional alterations and Principal Component Analysis (PCA) to search for similarities in participants’ emotional profiles. The sudden voice of a child and the sudden appearance of a human figure, possibly triggered the increase of happiness as the time analysis exposed. PCA revealed high variability in neutral emotion across clusters, with notable increases in anger in two of the three clusters. The first cluster showed the most variability in sadness, while happiness slightly increased only in this cluster.