Between the Buttons: Stress Assessment in Video Games using Players’ Behavioural Data

Flow lies at the heart of the interaction between players and video games. It is usually regarded as the optimal experience blooming in the fragile equilibrium that floats between boredom and anxiety. Under such circumstances, stress assessment can be a crucial experiential marker. In this preliminary study, we propose a computational approach to characterise the stress level of video game players, suitable to be exploited in the development of adaptive video games while enhancing players’ experience. To such purpose, a Virtual Reality (VR)-based video game has been created to gather data from participants. The information collected includes both physiological data and motion behavioural data (from game controllers), as well as the subjects’ self-reports of perceived stress. Behavioural data are specifically considered in the work presented here. We characterize the stress level evolution in terms of state-space dynamics, which is suitable for either discrete (classification) and continuous stress level assessment. Different experiments have been performed and results so far obtained are encouraging. In particular, along the stress vs. no-stress classification test, an accuracy of up to 84.4% is achieved by using VR-based data.