Contingency-based flexibility mechanisms through a reinforcement learning model in adults with Attention-Deficit/Hyperactivity Disorder and Obsessive-Compulsive Disorder

Abstract

AbtractThe probabilistic reversal learning paradigm is one of the most used to assess cognitive flexibility during a contingency-based learning process. Lack of cognitive flexibility is related to the symptomatology that characterizes disorders such as Attention-Deficit/Hyperactivity Disorder (ADHD) and Obsessive-Compulsive Disorder (OCD). Resting-state functional connectivity (rsFC) could be a specific predictor of performance in contingency-based cognitive flexibility. However, the mechanisms underlying learning and flexibility in an environment of uncertainty with OCD or ADHD adults have not been widely explored. Computational modelling may be helpful to separate components of the behavioural processes and identify the source of the disorders. In the present study, we aimed to identify the mechanisms underlying contingency-based cognitive flexibility in a sample of the impulsive-compulsive spectrum and healthy controls, and explore the rsFC between the frontoparietal networks (FPN) regions as a possible neuromarker. 148 Spanish-speaking participants (43 patients with OCD, 53 with ADHD, and 52 healthy controls) completed a probabilistic reversal learning task (PRLT). Previously, we obtained a record of FNP rsFC using functional near-infrared spectroscopy (fNIRS). For this purpose, we applied the reinforcement learning model in combination with Bayesian GLM. We found that groups showed an optimal performance in the acquisition phase of the PRLT and higher performance of HC compared to the diagnostic groups in the reversal block, although performance is still optimal in all groups. Likewise, we found that the parameters studied (reinforcement learning rate, punishment learning rate and inverse temperature) predict task performance differently by phase and group. Regarding FNP rsFC, we found that rsFC between left posterior parietal cortex (lpPC) and right posterior parietal cortex (rpPC) seems to credibly predict performance in the acquisition block in the healthy controls. These findings suggest that reducing the uncertainty between action-outcome may help to improve the adaptation of ADHD and OCD patients to changing environments. Thus, understanding sensitivity to punishment or reinforcement and its influence on the decision-making may be important for designing case-specific interventions. According to our data, rsFC between lpPC and rpPC could be important for optimal learning of our task.