Modelling environmental and cognitive factors to predict performance in a stressful training scenario on a naval ship simulator

Professionals working in risky or emergency situations have to make very accurate decisions, while the quality of the decisions might be affected by the stress that these situations bring about. Integrating task feedback and biofeedback into computer-based training environments could improve trainees’ stress-coping behaviour. This paper presents and assesses a refined version of the cognitive performance and error (COPE) model that describes the effects of stressful events on decisions as a foundation for such a support tool. Within a high-fidelity simulator of a ship’s bridge at the Royal Netherlands Naval College, students of the naval college ( n = 10) were observed while completing a 2-h-long shadowing and boarding operation combined with a search-and-rescue operation. For every action, variables were measured: objective and subjective task demand, challenge and threat appraisal, and arousal based on heart rate and heart rate variability. The data supported the COPE model and were used to create predictive models. The variables could provide minute-by-minute predictions of performance that can be divided into performance rated by experts and errors. The predictions for performance rated by experts correlated with the observed data ( r = 0.77), and 68.3 % of the predicted errors were correct. The error predictions concern the chances of making specific errors of communication , planning , speed , and task allocation . These models will be implemented into a real-time feedback system for trainees performing in stressful simulated training tasks.