Neurocomputation of hostile expectations

Hostile expectations can be defined as the assumption that aggression will occur in ambiguous or neutral situations. My aim in the present project is to formally define and test cognitive models of how hostile expectations are acquired and implemented at the neural and behavioral levels. Importantly, I will test whether hostile expectations also predict other domains of aggressive cognition and behavior, such as hostile attributions, i.e. the tendency to assign blame or harmful intent to ambiguous or neutral acts. To that end, I will develop a simple task to induce high- or low-threat expectations (i.e. a rival who draws a gun more or less often) which will be followed by aggressive or non-aggressive outcomes (i.e. the rival draws a gun or a phone). The probabilistic structure of the task will be manipulated, such that aggressive outcomes occur at varying frequencies across trials. Using a formal approach, I will inspect how well participants perform in tracking these changes and relate this to concurrent neural activity as well as to behavioral and self-report measures of aggression and hostility. I will use both electroencephalography and functional magnetic resonance imaging in pursuit of an in-depth characterization of how hostile expectations are implemented in the brain. This novel, multidisciplinary approach will allow to bridge predictive and retrospective processes of aggressive cognition, relate them to aggressive behavior, and map their neural representations in the temporal and spatial domains. The present project will provide a more precise understanding of the cognitive and neurobiological processes underlying aggression. This will contribute to refine existing theories and to improve the predictive validity of assessment instruments for aggressive behavior.

Aggressive individuals often “see red”: they assume that others want to harm them even in neutral or ambiguous situations. This makes them jump to conclusions and rashly act on their impulses. A better understanding of how these tendencies emerge might help us predict and manage antisocial behavior. Here, we crafted a virtual shooting task to measure how people develop hostile expectations. In the task, a sample of 256 healthy young adults had to either shoot or withhold their weapon depending on whether they expected another person to draw a gun or a phone. In this context, a tendency to shoot more often signals higher hostile expectations. Using computer algorithms, we were able to parse different aspects of hostile expectation learning and inspected whether they were linked with self-reported aggressive and psychopathic tendencies. As expected, persons with higher levels of aggressive and psychopathic traits developed more pronounced hostile expectations - i.e., they shot their gun more often. However, their hostile expectations were also more uncertain, meaning that they generally shot more indiscriminately. Surprisingly, persons with high levels of aggressiveness and psychopathy also showed more temporally stable hostility beliefs. This implies that their hostile expectations were more resistant to change once learned. As a result, they were more surprised when non-aggressive outcomes ensued (when their rival drew a phone instead of a gun). In sum, our study shows that persons with higher aggressiveness and psychopathy acquire strong and stable yet imprecise hostile expectations. Our findings thus provide fine-grained, mechanistic insight into the architecture of hostile thought and could potentially inform violence risk assessment and anger management therapy among others.