Levels of explanation in psychiatric and substance use disorders: implications for the development of an etiologically based nosology

The soft medical model for psychiatric illness, which was operationalized in DSM-III, defines psychiatric disorders as syndromes with shared symptoms, signs, course of illness and response to treatment. Many in our field want to move to a hard medical model based on etiological mechanisms. This essay explores the feasibility of this move and asks whether psychiatric disorders have the needed single clear level of explanation for an etiologically based nosology. I propose seven criteria for a good explanation: (i) strength, (ii) causal confidence, (iii) generalizability, (iv) specificity, (v) manipulability, (vi) proximity and (vii) generativity. Applying them to cystic fibrosis, a gene-level approach to etiology performs well across the board. By contrast, a detailed review of alcohol dependence and a briefer review of major depression suggests that psychiatric disorders have multiple explanatory perspectives no one of which can be privileged over others using scientific data alone. Therefore, a move toward an etiologically based diagnostic system cannot assume that one level of explanation will stand out as the obvious candidate on which to base the nosology. This leaves two options. Either a hard medical model will be implemented that will require a consensus about a preferred level of explanation which must reflect value judgments as well as science. To take this approach, we need to agree on what we most want from our explanations. Alternatively, we will need to move away from the traditional hard medical model that requires that we ground our diagnoses in single biological essences, and focus instead on fuzzy, cross-level mechanisms, which may more realistically capture the true nature of psychiatric disorders. Molecular Psychiatry (2012) 17, 11-21; doi:10.1038/mp.2011.70; published online 14 June 2011