Contemporary cognitive models describe that biased information processing as a vulnerability factor in the development and maintenance of depressive episodes. Moreover, residual symptoms have shown to be among the strongest predictors for relapse in recurrent depression. In this doctoral dissertation various neuroimaging approaches and analytic strategies were used, with the overarching aim to investigate neural mechanisms related to two central risk factors in depression: attentional biases and residual symptoms. The studies were conducted as part of a preregistered randomized controlled trial, where 134 participants with a history of depression were randomized to two weeks of Attentional Bias Modification (ABM) training, or a closely matched control condition. An emotion regulation fMRI experiment and a resting state fMRI protocol was conducted after the intervention to investigate the neural effects of ABM training. The fMRI studies demonstrate that ABM training has an effect on brain function within circuitry associated with cognition and emotion, and indicate an effect of ABM on depressive symptoms. In the dissertation, network psychometry was used to model a network of depression symptoms and regional brain measures, in a mixed group of 268 individual with a history of depression and considerable variation in symptom severity, and never previously depressed individuals. The symptom network analysis revealed associations between unique depressive symptoms and brain structure, which may offer important cues to underlying mechanisms of depression. Taken together, this thesis supports the notion that depression is unlikely to stem from dysfunction of any specific biological, cognitive, or environmental factor, and that the clinical phenotype (the symptoms) can be seen as the end point of underlying dysregulation of distributed neural networks and cognitive-emotional control processes.