Memory retention of spatial knowledge in fire evacuation and safety training

This study investigated the retention of spatial knowledge in buildings following route-learning training in a virtual reality environment. A total of 121 participants were tested up to three months later on putting waypoints of the route in the correct order and recalling directions at waypoints. Memory accuracy declined over time, consistent with classic memory theory. Route knowledge was retained more robustly than sequential order, highlighting the importance of contextual retrieval cues. Landmark presence, decision-point complexity, and route features modulated recall, demonstrating that both task and environmental characteristics influence spatial memory. A hierarchical Bayesian regression model quantified forgetting with median memory accuracy, capturing uncertainty across individual variability of participants and the environment in the experiment. Predicted accuracy decreased from approximately 91% initially to 77% after 12 weeks, and to approximately 72-75% after 6-12 months indicating that a substantial portion of spatial knowledge is retained over long intervals. By applying memory theory to analyse retention data, this study addresses a gap in the safety training field by providing a theory-driven approach to quantifying training effectiveness, enabling evidence-based design and assessment of safety and evacuation training in practice.