Player behaviour metrics for adjusting content in VR games: the case of fear

Video games are complex multidisciplinary artefacts, that more and more include sophisticated narrative content able to leverage emotions in players. Due to their interactive nature, they should meet the expectation of a diversified audience. Till now the problem of automatically adapting content to players have focused mainly on gameplay-related aspects, and too few attentions have been devoted to track players emotions to adjust coherently the narrative of the game. This issue becomes more evident in Virtual Reality (VR) based games, in which the immersivity is drastically emphasized, and for specific game-genres, such as horror games. We have tackled this issue by designing and developing a system that tracks the behaviour of the players to adapt the content of a horror VR game, increasing or decreasing drastically its frightening elements according to how much a specific player "likes to be scared". We have then tested our proposed solution with a sample of players, obtaining some preliminary encouraging results.