Low self-worth has been identified as a core symptom of depression (Zahn et al., 2015) and prospectively predicts risk of relapse at a group level. It has been shown to be present at least a year prior to the onset of a depressive episode (Ormel et al. 2004), and, as such, is a prime target for prevention of depression recurrence. Cognitive models of depression explain low self-worth as a result of overgeneral self-blame resulting in distortions in the individual's concept of self (Abramson et al., 1978). We have identified a brain network associated with overgeneral self-blame in patients recovered from depression and shown that remitted MDD patients undergoing one session of active MRI neurofeedback were able to modulate this network compared with a control intervention group, and that this correlated with improvements in self-esteem (Zahn et al 2019). This project will build on our earlier work and will seek to develop a machine learning approach to identify brain networks associated with self-blame. The identified networks will then be used as a target for fMRI neurofeedback in patients with recovered depression. The project will be in two parts: 1) The student will analyse fMRI imaging data of a self-blame task (Lythe et al. JAMA Psych 2015; n=64 remitted MDD, n=50 healthy controls) and will develop an SVM-based fMRI signature of MDD recurrence risk in response to self-blaming thoughts. 2) This signature will then be used as a neurofeedback target in a trial of patients with partially recovered depression to test the hypothesis that neurofeedback training will lead to improvements in self-esteem. Briefly, this intervention will involve participants viewing a vertical scale during an fMRI task of self-blaming thoughts. The vertical scale will increase the closer they get to the target brain network (which will either be the SVM network, or a control - pretreatment network). The primary outcome will be improvement in self-esteem as measured by the Rosenberg Self-Esteem Scale. The student will have the opportunity to learn key techniques for neuroscience research as well as transferrable skills. Their contribution will include study design, patient recruitment, and data analysis.