Abstract
AbstractBackgroundCompared to healthy individuals, patients with obsessive-compulsive disorder (OCD) are found to rely more on model-free decision-making strategies which may underlie symptom expression. It is, however, unclear whether these behavioural differences are represented in neural alterations of model-free and model-based decision-making when tested simultaneously.MethodsWe investigated the neural signatures of 22 OCD patients and 22 matched controls who completed a two-step Markov decision-making task during functional MRI scanning. We used hierarchical Bayesian modelling and Bayesian statistics to examine model-based and model-free decision-making behaviours. Parametric regressors were employed for model-free and model-based reward prediction errors to inform neural reward presentation, which we analysed using a Bayesian Multilevel Modeling (BML) approach. Associations between significant activations and symptoms as well as cognitive scores were explored using Bayesian linear regression.ResultsWhile controls received significantly more rewards and were significantly less stochastic compared to patients, both groups similarly relied on model-free decision-making strategies. Importantly, our group comparison of neural reward prediction error responses showed greater activation for model-based reward prediction error in the medial prefrontal cortex (mPFC) in OCD patients compared to controls, but no differences for model-free reward prediction error processing. Increased mPFC activity was associated with lower depressive symptoms but not with other clinical or cognitive scores.ConclusionThese findings support the notion that OCD is associated with a bias towards habitual behaviour, which may be expressed through altered mPFC activity underlying goal-directed behaviour. Tackling the hyperactivity on the mPFC may provide a new target for optimised treatment and interventions.
Publisher
Cold Spring Harbor Laboratory