For many psychiatric disorders, neurobiological findings do not help to diagnose a specific disease or to predict its outcome. This book suggests to take a new look at mental disorders by using computational models to better understand human decision making. It shows how such models can be applied to basic learning mechanisms that cut across established nosological boundaries of mental disorders. Such a computational and dimensional approach focuses on the malleability of human behavior and its biological underpinnings. The book argues that this computational and dimensional approach can help to promote and focus neurobiological research, however, it does not replace an anthropological understanding of clinical questions including the definition of mental disorders and ethical considerations. This is illustrated by describing the new understanding of mental disorders with respect to clinical and neuro-computational aspects of psychosis, affective and addictive disorders.