Pathological aversive learning as a mechanism in clinical anxiety (360G-Wellcome-206460_Z_17_Z)

Generalised anxiety disorder (GAD) is a common and debilitating condition. Its core symptoms are strongly related to subjective feelings evoked by uncertainty. This fellowship will build on advances in computational modelling to characterise the role of estimations of uncertainty about aversive events in GAD. The project has three key goals: 1) To develop behavioural tasks, and computational models of these tasks, that allow estimations of uncertainty during aversive reinforcement learning. I aim to achieve this by building on existing reinforcement learning tasks that manipulate uncertainty through changing stimulus-outcome contingencies, combined with models that allow explicit estimation of participants’ perceived uncertainty. 2) To determine whether individuals with GAD overestimate uncertainty, or overestimate the likelihood of negative outcomes in uncertain situations. To do this I will exploit tasks and models developed in phase 1 of the proposal to compare subjects with GAD and healthy individuals on their estimation of uncertainty in aversive environments. 3) In this phase I aim to identify the neural basis of these abnormalities using functional neuroimaging (fMRI), by combining these tasks and models with fMRI to identify neural systems underlying uncertainty estimation in aversive contexts, and examining the differential function of these systems in patients with GAD.