To reduce the burden of mental disorders it is a formidable aim to identify widely applicable disease markers based on neural processes, which predict psychopathology and allow for targeted interventions. We will generate a neurobehavioural framework for stratification of psychopathology by characterising links between network properties of brain function and structure and reinforcement–related behaviours, which are fundamental components of some of the most prevalent mental disorders, major depression, alcohol use disorder and ADHD. We will assess if network configurations define subtypes within and if they correspond to comorbidity across these diagnoses. We will identify discriminative data modalities and characterize predictors of future psychopathology.To identify specific neurobehavioural clusters we will carry out precision phenotyping of 900 patients with major depression, ADHD and alcohol use disorders and 300 controls, which we will investigate with innovative deep machine learning methods derived from artifical intelligence research. Development of these methods will optimize exploitation of a wide range of assessment modalities, including functional and structural neuroimaging, cognitive, emotional as well as environmental measures. The neurobehavioural clusters resulting from this analysis will be validated in a longitudinal population-based imaging genomics cohort, the IMAGEN sample of over 2000 participants spanning the period from adolescence to adulthood and integrated with information generated from genomic and imaging-genomic meta-analyses of >300.000 individuals.By targeting specific neural processes the resulting stratification markers will serve as paradigmatic examples for a diagnostic classification, which is based upon quantifiable neurobiological measures, thus enabling targetted early intervention, identification of novel pharmaceutical targets and the establishment of neurobehaviourally informed endpoints for clinical trials.