Effectiveness of Training under Stress in Immersive VR: An Investigation of Firefighter Performance, Gaze Entropy, and Pupillometry

Ranjana K Mehta ^1* John Kang ² Yangmin Shi ³ Jing Du ³

• ¹ University of Wisconsin-Madison, Madison, United States
• ² Texas A and M University, College Station, Texas, United States
• ³ University of Florida, Gainesville, Florida, United States

Training is essential for responder performance during emergencies, which are filled with uncertainties and stress. Virtual Reality (VR) offers a safe, repeatable, and cost-effective training tool for simulating stressful emergency scenarios. This study aimed to evaluate firefighters' performances in a VR-based emergency response training scenario by using a visuospatial sequence learning task to assess learning and retrieval effectiveness under stress in VR. Forty firefighters from the local community were randomly assigned to either a control group or a stress training group, and they completed the VR-based visuospatial learning task, followed by two sets of retrieval tasks (one in a routine condition and the other in an emergency situation). Eye-tracking measures (such as gaze behavior and pupillometry), perceptions of workload and anxiety, and task performance were collected from both groups. While the stress training group exhibited poorer performance scores and longer operation times than the control group, the retrieval of learned information was similar. These findings were associated with lower gaze entropy, larger pupil dilation, and constriction in the stress training groups, especially during the initial training trials, along with heightened perceptions of mental demand, effort, frustration, and lower perceived performance. Eye-tracking data obtained from VR headsets can provide insights into individual cognitive states under various environmental stressors that may be utilized to create more adaptive training paradigms.