Physiological detection of flow states across varied gameplay workloads

Abstract

Flow is a cognitive state associated with heightened and seemingly effortless behavioral performance. Pioneering research suggests flow is an optimal cognitive mode for demanding tasks but the physiological underpinnings and identification on acute timescales, particularly across task contexts, remains an active area of research. Studies have demonstrated flow experiences are ubiquitous, in that they can arise during nearly any type of activity, suggesting flow represents a fundamental mechanism of skill mastery under which sensory inputs are seamlessly coupled to the necessary motor outputs. Cognitively, flow states are accompanied by a sense of complete task immersion, feelings of control, automatic behavior, decreased self-referential thinking and anxiety, changes in the perception of time, and improved affect to the point that arduous tasks can become autotelic, or self-rewarding. As wearable sensor technology continues to advance, the identification and manipulation of cognitive states in real-time is increasingly becoming a reality, with the potential to revolutionize learning, treatment of neurological disorders, and optimize task performance in domains ranging from professional sports to industry. This study aims to capture flow states from physiological measures so that they might be identified through wearable sensors coupled to machine-learning pipelines. In this work, models are made to predict high behavioral performance across disparately varied task conditions using multimodal physiological data, and these behavioral bouts are loosely correlated to participant responses to traditional flow state survey questions.