<p>Cognitive impairments have been observed in patients with depression. These include deficits in inhibition, shifting, and updating; cognitive processes that are critical for goal-directed control over behavior (‘model-based planning’). Nevertheless, results of model-based planning in depression have been mixed. We aimed to address this by taking a within-person approach, examining model-based planning before and after a range of effective treatments for depression. Across two parallel studies, participants completed a two-step reinforcement learning paradigm before and after antidepressant medication, internet-based cognitive behavioral therapy (iCBT) or intravenous (IV) ketamine infusion. In experiment 1, 93 patients with treatment-resistant depression were randomized to a single dose of IV ketamine (0.5 mg/kg) or IV saline (50 mL 0.9% NaCl). In Experiment 2, 781 participants were followed for four weeks of antidepressant (N = 83), or iCBT (N = 611) treatment. N = 87 participants without any psychiatric diagnosis were followed as a control group. In both experiments, depressive symptoms significantly improved in treatment groups compared to their corresponding control groups, but we did not find evidence of changes in model-based planning. Moreover, we failed to find associations between individual differences in model-based planning and differential response to ketamine, iCBT or antidepressant treatments. Individual differences in model-based planning at baseline were associated with compulsivity, but not with depression symptoms. These findings suggest that model-based planning is not necessarily compromised in depression and does not improve following treatments. This result provides evidence for the trait-like nature of model-based planning and underscores the specificity of its relation to disorders of compulsivity.</p>