Towards the Design of Adaptive Virtual Reality Horror Games: A Model of Players' Fears Using Machine Learning and Player Modeling

Abstract

Horror games are designed to induce fear in players. Although fundamental fears, such as the unknown, are inherent to the human being, more specific fears, such as darkness and apparitions, are individual and can vary from person to person. When a game aims at intensifying the fear evoked in individual players, having useful information about the fears of the current player is vital to promote more frightening experiences. This paper explores fear modeling and presents a new method to identify what players fear in a virtual reality horror game. The proposed method uses machine learning and player modeling techniques to create a model of players' fears, which can be used to adapt in-game horror elements to intensify the fear evoked in players. The paper presents the proposed method and evaluates its accuracy and real-time performance in a virtual reality horror game.