Understanding Stress Responses: Exploring Facial Expressions in the Context of Individual Performance and Automated Agents

Abstract

Understanding and effectively detecting stress is paramount across various domains, including healthcare and business, where individuals often operate under pressure. This paper proposes a novel approach that employs facial expression analysis to discern stress levels during both stressful and non-stressful phases. Our study aims to understand the impact of automated agents on individual performance and how performance, in turn, influences facial expression dynamics. Utilizing facial video data, we scrutinize the facial expressions exhibited by individuals under induced stress conditions, comparing them with relaxed states. Building upon prior research in stress detection, we conduct in-person experiments to study facial expressions indicative of stress responses, examining features such as facial muscle movements, microexpressions, and overall expression dynamics. Machine learning techniques are leveraged to classify stress levels based on facial expressions extracted from video data. Preliminary findings reveal distinct patterns of facial expressions associated with stress, encompassing heightened muscle tension, altered facial symmetry, and changes in expression intensity. By contrasting these patterns between stressful and non-stressful phases, our objective is to formulate a robust model for real-time stress detection using facial analysis. This research contributes to the advancement of stress detection methodologies, offering potential applications in healthcare, psychology, and human-computer interaction. Future directions include refining the classification model and exploring additional contextual factors that influence facial expressions during stress.

Keywords: Induced Stress, Individual Performance, Time Pressure, Performance Pressure, decision-making, human-agent, Stress, and training

DOI: 10.54941/ahfe1005666