In his chapter, Eric Turkheimer argues that psychiatric reduction is a doomed project because mental illnesses have properties that make them scientifically unruly; properties we might call, following the philosophical literature, “multiple realizability” and “social constructedness.” While it may be defensible to maintain that psychiatric syndromes are multiply realized social constructions, and thus not appropriate targets for reductive explanation, I argue that it is less obvious that all psychiatric entities share these properties. It may well be that reductionist projects will do better in psychiatry when they focus on explaining lower-level entities (for example, pathological mechanisms at the level of the cell or the neural circuit) that are only part of the puzzle of psychopathology. However, the promise of this sort of “patchy reductionism” needs to be evaluated in light of the ethical demands on psychiatric researchers not only to expand scientific knowledge but to improve clinical care and outcomes.