17There has been increasing interest in using neuroimaging measures to predict psychiatric 18 disorders. However, predictions usually rely on large numbers of brain connections and large 19 disorder heterogeneity, thus lacking both anatomical and behavioural specificity, preventing 20 the advancement of targeted interventions. Here, we address both challenges. First, using 21 resting-state functional MRI, we parcellated the amygdala, a region implicated in mood 22 disorders but difficult to image with high fidelity, into seven nuclei. Next, a questionnaire 23 factor analysis provided four sub-clinical latent behaviours frequently found in anxious- 24 depressive individuals, such as negative emotions and sleep problems. Finally, for each latent 25 behaviour, we identified the most predictive connections between individual amygdala nuclei 26 and highly specific regions of interest e.g. dorsal raphe nucleus in the brainstem or medial 27 prefrontal cortical regions. A small number of distinct connections predicted behaviours, 28 providing unprecedented levels of specificity, in humans, for relating mental well-being to 29 precise anatomical connections. 30 time in animal models including primates 19 and there is increasing knowledge of the 62 behaviours mediated by amygdala interactions 11,13,20,21 . 63 If, however, a decision is taken to focus on a brain region such as the amygdala then 64 a third problem arises. Many of the key brain areas with which it interacts are in the brainstem 65 where it has been difficult to image activity. Moreover, such regions have very specific 66 connections to particular sub-nuclei within the amygdala. Therefore, our first step was to 67 parcellate the human amygdala into constituent functional sub-units. We took advantage of 68 the high-quality data acquired as part of the human connectome project (HCP; 22 ). Using 69 resting-state measures from 200 healthy participants, we reliably identified seven amygdala 70 nuclei within each hemisphere. We also invested considerable effort in developing a refined 71 data pre-processing pathway that focused on the removal of breathing related artefacts that 72 allowed us to examine activity even in brainstem regions, several of which exhibit very specific 73 interactions with particular amygdala subnuclei. 74 In tandem with improving anatomical specificity we also aimed to tackle another 75 major problem in relating baseline neural measures to mental well-being. Namely, the 76 disorders themselves are ill-defined and span a broad range of impairments which are not 77 consistently present in all patients diagnosed with the same disorder 23 and which are partly 78 overlapping between disorders. This may be another reason why a classifier trained to 79 distinguish a depressed from a non-depressed person is likely to reveal a broad network of 80 regions instead of mapping onto well-defined and anatomically interpretable brain circuits. If 81 we are able to focus on specific rather than broad symptom categories, we may...