Theoretical Framework for Utilizing Eye-Tracking Data to Understand the Cognitive Mechanism of Situational Awareness in Construction Hazard Recognition

Comprehending the cognitive processes underlying hazard identification is crucial for enhancing worker safety behavior in construction. Recent studies have explored eye-tracking technology’s potential in understanding human cognition across contexts. However, limited research delves into the intricate cognitive processes linking eye movements and hazard recognition, particularly in the context of situational awareness (SA). Thus, this study investigates the relationship between eye movement data and SA’s cognitive processes in hazard recognition virtual reality (VR) scenarios at construction sites. The study employed experiments with 36 participants identifying construction hazards across six VR scenarios, yielding 216 trials. Eye movement data were collected via the VR headset’s eye-tracking device, concurrently recording hazard recognition performances. The results uncovered valuable insights into the correlation between eye movement patterns and global and local SA. In the context of global SA, time to and after first fixation elucidated the distinct variations among individuals in terms of perception (Global Level 1 SA) and comprehension times (Global Level 2 and Level 3 SA) across various hazard scenarios. In the realm of local SA, more fixations and saccades (Local Level 1 SA) were observed during the first dwell, underscoring the significance of the first encounter with a hazard. Additionally, pupil dilation, indicative of increased mental workload, occurred upon successful hazard recognition (Local Level 2 and Level 3 SA). These findings highlight the explanatory potential of various eye movement data types for diverse SA levels. They can serve as effective SA indicators in hazard recognition contexts, enhancing understanding of cognitive processes and refining assessment and training for SA in hazardous settings.