Understanding stress-induced illegitimate aggression: the role of physiological and psychological factors in police cadets

To better understand the consequences of stress in realistic scenarios, police cadets were tasked with performing a police intervention under differing expectations. One group was led to anticipate a dangerous mission, while the other expected a routine event. In the field, however, both groups faced the same challenging situation. The warned group exhibited strong pre-intervention stress responses, which was minimal in the other group. By contrast, the unwarned group experienced a sudden surge in stress within the first minute of the intervention, as reality clashed with their expectations. A similar sudden stress response by the beginning of the intervention was missing from the warned group. A significant portion of cadets unlawfully attacked suspects, a behavior linked to intense stress displayed at the onset of the intervention. This emotional, illegitimate aggression was driven primarily by the noradrenergic stress response, with no indication of cortisol involvement. Traditional statistical methods (group comparisons, univariate, and multivariate regressions) suggested that psychological traits had little impact compared to acute stress effects. However, machine learning revealed that psychological characteristics-such as those assessed by the Reactive-Proactive Aggression Questionnaire, Buss-Perry Aggression Questionnaire, Big Five Personality Test, and Barratt Impulsiveness Scale-played a crucial role in conjunction with stress responses. Multivariate analyses yielded data similar to those obtained through machine learning, but only when the dependent variables were selected to match those identified as crucial by the latter. These findings highlight the power of machine learning in uncovering complex interactions that traditional methods might overlook.