Internal State Estimation via Physiological Data and Their Modulation by Environmental Context during Social Activity.

Abstract

Understanding individuals' internal cognitive states during group interactions is crucial for enhancing group dynamics and communication. This study investigated internal states by analyzing physiological data-EEG, electrocardiography, and pupil size-collected from high school students during group discussions. Using a data-driven clustering method, we identified four distinct internal states, each corresponding to the different power distributions in the four frequency bands of EEG activity. These states were associated with specific behaviors such as gazing at faces, speaking, and specific body language, as well as physiological metrics such as heart rate variability and pupil size. We also examined the influence of environmental factors on internal states, including the presence of a facilitator and the group size. The presence of a facilitator significantly increased the probability of participants remaining in the high alpha-power state, possibly reflecting a relaxed or moderately aroused state. This study provides insights into the physiological underpinnings of group interactions, which can be leveraged to improve educational settings and other group-based activities.