Modeling Behavioral Dynamics and Cognitive States in VR Safety Training: A Learning Analytics Approach

• In Event: AERA Poster Session Wednesday 9:45am
  In Poster Session: Instructional Technology SIG Poster Presentations

Abstract

This study captures dynamic learning processes in virtual reality (VR)-based safety training using behavioral data from 27 engineering students in a heat-stress scenario. A three-phase analysis—sequential behavior analysis, Hidden Markov Modeling (HMM), and behavior network analysis —revealed four latent cognitive states: information gathering, task execution, collaborative learning, and confusion. Learners frequently returned to instructional states, while some remained in confusion without progressing. Network density varied across states, with more integrated behaviors observed during collaboration and orientation. Findings support a state-based view of VR-based learning and demonstrate the value of combining HMM and network analysis to capture individuals’ real-time learning dynamics. The study informs the design of adaptive VR training by identifying moments of struggle and opportunity for instructional support.

Authors

• Stephen Emmanuel Abu, University of Alabama

• Jewoong Moon, University of Alabama

• Idowu David Awoyemi, University of Alabama

• Raissa Marchiori, Florida Gulf Coast University

• Siyuan Song, Arizona State University