Integration is not necessarily at odds with reductionism

De Rosa et al. have argued eloquently for integrating social and biological psychiatry, helpfully bridging the existing dichotomised perspective of social versus biological. We agree wholeheartedly that this dichotomy promotes a negative image of psychiatry to the patients for whom we work. The integration of social and biological approaches is crucial in both experimental and therapeutic endeavours. However, an issue of contention in De Rosa et al.’s carefully considered argument is their use of the adjective ‘reductionist’ without qualification by definition; this is not unique to them and is present in the broader medical literature (e.g. a similar integrative proposal given in Greene & Loscalzo, 2017). There are at least three definitions in philosophy of mind and science which are pertinent here:

These definitions imply commitments to different theoretical positions, but they also share intellectual methods, namely,

The problem with the unqualified use of ‘reductionism’ as a term is that we assume everyone agrees on a single definition or concept that is inherently ‘bad’, but ultimately, this fails to take aim at either the commitments or intellectual methods – as such, it becomes an effigy, straw-man or pejorative. De Rosa give examples of how social psychiatry reveals new disorders citing Internet gaming disorder (Petry, Rehbein, Ko, & O’Brien, 2015) as a mental disorder ‘created’ by misuse of technology. Submitting such novel disorders to the ‘tools’ of reductionist analysis, we could conclude that, for example, IGD is a nominal HL abstraction on the dysregulated cortico-mesolimbic system (mechanism, compositionality) where the HL theory provides a vocabulary to describe observable behaviours (e.g. preoccupation with Internet games to the exclusion of other activities, with withdrawal and tolerance). The crucial point is that we do not replace the social or psychological as a tentative ‘higher level’ but rather protect against diagnostic proliferation while building on existing understanding. This aligns with De Rosa et al.’s description of social neuroscience and represents what Craver called mechanistic integration and Kendler (2008) called decomposition and reassembly, whose origin arguably lies in the seminal computational neuroscience of Marr and Poggio (1976) and Marr (1982). Analysing multiple-levels of explanation using a systems approach (molecular biology, through systems neuroscience to environment and psycho-social factors) does not commit to disposing – by reduction – any level of explanation. Perhaps a better approach should be to ‘call-out’ misuse of reductionism by specific definition and instead demonstrate how the intellectual methods fail rather than assume it is simply wrong. Integration requires a melding of theory and intellectual methods; in psychiatry, these are certainly more complex than single-level models, and it is encouraging to see adoption of initiatives such as the Research Domain Operational Criteria project (Cuthbert & Insel, 2013).