An empirical study of players' emotions in VR racing games based on a dataset of physiological data

A video game is an interactive software able to arouse intense emotions in players. Consequentially, different theories have been proposed to understand which game aspects are able to affect the players' emotional state. However, only few works have tried to use empirical evidence to investigate the effects of different game aspects of the players' emotions. In this paper, we present the results of a set of experiments aimed at predicting the players' emotions during video games sessions using their physiological data. We have created a physiological dataset from the data acquired by 33 participants during video game fruition using a standard monitor and a Virtual Reality headset. The dataset contains information about electrocardiogram, 5 facials electromyographies, electrodermal activity, and respiration. Furthermore, we have asked the players to self-assess their emotional state on the Arousal and Valence space. We have then analyzed the contribution of each physiological signal to the overall definition of the players' mental state. Finally, we have applied Machine Learning techniques to predict the emotional state of players during game sessions at a precision of one second. The obtained results can contribute to define game devices and engines able to detect physiological data, as well to improve the game design process.