What tugs at your heartstrings? Exploring flow, and affect recognition through HRV, while video gaming

This study explores the detection of flow states through affective computing, focusing on physiological measures such as Heart Rate Variability (HRV) during Tetris gameplay. Flow, characterized by a balance between challenge and skill, is a state of complete absorption that is complex to detect. We experimented with different difficulty levels of Tetris to elicit flow states and recorded interbeat intervals using a Polar H10 chest band. We refined the experimental protocol through a pilot study: 5 Tetris levels were chosen so that participants reported a suitable variance in difficulty. During the study, 53 participants' HRV metrics were analyzed alongside self-reported flow and affective states using questionnaires. The data analysis involved classical statistical methods and machine learning algorithms to classify flow and affective states based on HRV features. SVC and RandomForest algorithms achieved high accuracy in predicting flow-related categories (48% for balance, 52% for skill, 49% for challenge), while affective states results were more various (44% for dominance, 44% for valence, 43% for arousal, using respectively BernoulliNB, GaussianNB, and ExtraTrees). The results indicate distinct HRV patterns associated with flow and affective states, suggesting that HRV is a viable indicator for flow detection. Further research is needed to validate these findings with more comprehensive experimental designs and diverse realworld applications.