Training systems are a potential stress countermeasure by simulating high stress conditions in a safe and controlled environment. Training often involves increasing the complexity of scenarios over time until trainees can reliably execute the task. However, several limitations reduce the long-term retention and robustness of training during intense acute stress felt in-situ. These limitations include generalized training practices that are not tailored to the individual, unreliability of self-reported subjective stress, over-trained skills that are inflexible and not robust to novel stressors, training pedagogies that focus too much on task proficiency rather how the individual manages stress during task execution, and ambiguity for when trainers should increase/decrease training difficulty.
Adaptive systems are proposed as a supplement while training individuals to maintain task performance. An adaptive system is a joint human-computer system that is able to automate functions/tasks to varying degrees to help the user, often without explicit instruction. In the context of training for stressful tasks, an adaptive system could detect and monitor stress using physiological sensors and machine learning and use this information to modify scenarios to provide individualized training. This would allow coping skills to be practiced without overwhelming or under-stimulating the trainee’s stress tolerance, adapt training according to proficiency in both task execution and physiological stress, and offer clear benchmarks for when to increase/decrease training difficulty. Coupled with a simulated training environment, an adaptive system could adapt training by altering the task procedure and implicitly changing in the task environment to help the user build resilience to novel stressors.
This paper presents conceptual approaches and applications for training stressful operations using adaptive systems. A generic adaptive stress training system framework is described along with recommendations based on an experimental example in the spaceflight domain for training emergency fire procedures in a virtual reality International Space Station.