Addressing the potential for training scars: investigating the risk of using Extended Reality in police training

Abstract This investigation explores the potential impact of Extended Reality (XR) training in police agencies by leveraging Bayesian Brain and Active Inference hypotheses. These theories suggest that the brain constructs predictive models in response to both internal and external sensory stimuli. XR offers simulated scenarios to enhance experiential training and operational skills, thereby generating particular types of sensory information. However, over-reliance on repetitive XR scenarios can lead to ‘Training Scars’ habituated behaviours from recurrent exposure to similar environments, negatively impacting real-life decision-making. To counteract this, we propose integrating artificial intelligence and large language models into XR training to generate varied scenarios with elements of randomness, enhancing officers’ capabilities by diversifying their experiences and thus improving their predictive models.