Impaired probabilistic classification learning with feedback in patients with major depression

• We examine feedback learning in patients with major depression.
• Patients are assessed with the weather prediction task.
• Patients are impaired in feedback based learning compared to healthy controls.
• Reduced learning on the FB task reflects disturbed reinforcement learning in MD.

BackgroundThe function of basal ganglia (BG) in the pathophysiology of major depression (MD) is still unclear. Recent research found changes in BG regarding size, structure and cerebral perfusion in patients with MD. Neuroimaging shows recruitment of the striatum during feedback (FB) based incidental learning of probabilistic classification learning, while the medial temporal lobe (MTL) is associated with paired associate (PA) based incidental learning. The purpose of this study was to evaluate whether FB-based incidental learning is affected in MD.MethodsThe FB and PA versions of the weather prediction task (WPT), a task of incidental probabilistic classification learning, were completed by patients with MD (n = 44) and healthy controls (n = 44). In FB-learning the participants received either a “thumbs-up” or “thumbs-down” message according to their right or wrong classification of cards to a certain kind of weather (either rainy or fine), while in PA learning no classification was required. Severity of MD was rated on the Beck Depression Inventory and Hamilton Rating Scale for depression.ResultsPatients with MD were selectively impaired on the FB task relative to controls (p < 0.05), while no significant difference was found for PA learning between the two groups. Furthermore, there were no significant differences between FB and PA-learning within the patient and control groups.ConclusionOur results indicate a distinct impairment on the FB-based version of the weather prediction task. These findings implicate disturbed reinforcement learning in this group of patients.