Following Donald Davidson, we suppose that cause produce their effects in accord with principles, at least wherever we can warrant causal claims. But we differ from Davidson in our view of what such principles are like. These need not be deterministic laws but, following J. S. Mill, ‘tendency principles’. Tendency principles are usually expressed as generics: they may only apply in appropriate contexts, they may not operate every time they apply (e.g. they may need triggering or act in a chancy way) and they don’t describe what actually happens since other tendencies may contribute as well. They can be found during research but can also be general knowledge. Knowing the principle and activity by which a cause produces its effect helps identify other features that can evidence causation.

We suppose that, stage by stage in causal processes, causes engage in activities that bring about their effects. These should be maximally vividly described when representing causal processes. We thus follow G. E. M. Anscombe, recent work in philosophy of biology on mechanisms, process tracing and realist evaluation in rejecting the Humean approach to causation, which eschews activities. Activities are interpreted as continuous processes afforded by the arrangements of the underlying system in which they take place. We illustrate here how to include activities in our original ‘boxes-and-arrows’ diagrams.

Very few, if any, causal processes are self-standing – rather, they depend on features of ‘underlying systems’: sufficiently stable physical, socio-economic, legal and cultural conditions that afford such processes (‘mechanisms’ in some New Mechanists’ terms). The underlying system is whatever it is in the world that affords the causal process – a structure exhibiting sufficient stability that affords the causal activities that occur within the process and the principles that govern them. This is illustrated by Rube Goldberg machines and examples from biochemistry, economics and deskilling in child protection. Underlying systems are helpfully understood through assessing their components and features, and the affordances associated with them, with the boundaries of the underlying system considered pragmatically.

This chapter addresses three questions posed for the 2018 Copenhagen conference. We argue that reduction in the widely assumed sense of eliminating psychological constructs is not a feasible option in psychopathology research. We argue that the popular “levels of analysis” metaphor is more problematic than helpful. We argue that a recent movement in philosophy of science, known as the new mechanists, offers a promising alternative to the naïve biological reductionism that has driven much thinking and research on psychopathology in the Decades of the Brain. Finally, we evaluate the NIMH Research Domain Criteria (RDoC) initiative in the context of these three questions, citing key features that facilitate moving clinical research forward more quickly and more effectively than what has characterized the field for decades. RDoC avoids reductionism, fosters integration of psychological and biological constructs, methods, and data, and is well suited to the emerging research agenda in the psychopathology literature.