Emotion-Driven Interactive Storytelling: Let Me Tell You How to Feel

Abstract

Interactive storytelling is a form of digital entertainment that has gained attention with the development of creative computational methodologies. However, one of the main problems this field is facing is the poor control that the content creator (e.g. film director or game designer) has over the experience of the user (e.g. viewer or player) once the story starts. Hence, we leverage artificial intelligence to increase the creative control of the content creator by designing a system that guides the user’s emotions towards a particular state as the story unfolds. Specifically, we have developed an EEG-based emotion recognition system trained on EEG recordings acquired from 5 participants watching a selection of 384 videos. The system is able to operate a binary classification on both valence and arousal with an accuracy of 62% and 57%, respectively. A short film was then created, where each scene automatically adapts to the user’s emotion, based on a set of predefined interactions established by the content creator (i.e. the actual film director). The analysis shows that the system not only improves the engagement of the user, but also induces an emotion closer to the one intended and specified ahead of time by the content creator for the story. Our results indicate that there is a practical application of emotion-based studies for future content creators to better control an intended emotional response delivered and received by the audience.