Borsboom et al. aim to show that reductive research strategies are misguided in the context of psychopathology research. More specifically, they claim that adopting network models as an alternative framework for the analysis of mental disorders will show how the reductive aspirations of the traditional research are ill-founded. It seems, however, that they have misconstrued their target: The ultimate aim of neuroscientifically based research (or more generally, physiologically based research) on psychopathology is not to reduce mental disorders to neural phenomena, but to eliminate the current notions of mental disorder altogether by changing the nosological practices in fundamental ways.

The focus of the analysis that Borsboom et al. are offering is on a DSM-based nosology. The DSM defines mental disorders in a symptom-centred way. There is nothing inherently wrong with this: After all, it is from symptoms that all clinical work starts off, and the DSM can therefore be a useful diagnostic tool for clinicians. But Borsboom et al. are not interested in clinical practices in psychiatry, at least not primarily; rather, they are interested in psychopathology research – the question of what mental disorders really are, and how we should conduct research on defining their true nature. Symptoms, however, are mere signs, marks of the underlying disorder or illness; and even in clinical practice, the ultimate aim is not just to remove the symptoms, but to cure the physiological condition they stem from.

Therefore, those who are explicitly proposing reductionist research on psychopathology tend to stress that basing our scientific understanding of mental disorders on the DSM is ill-founded. In particular, the Research Domain Criteria (RDoC) project ultimately aims to replace the DSM-based classifications of mental disorders. The main thrust of this project is in the conviction that the DSM nosology is invalid: that it clusters together disorders that are symptomatically similar, but aetiologically and physiologically different (cf. Cuthbert Reference Cuthbert2014; Cuthbert & Kozak Reference Cuthbert and Kozak2013; First Reference First, Kendler and Parnas2012; Insel et al. Reference Insel, Cuthbert, Garvey, Heinssen, Pine, Quinn, Sanislow and Wang2010). Therefore, the idea is that we should aim to abandon the superficial DSM classifications and replace them with more valid classifications based on physiological aetiologies. Thus, this agenda is eliminativist, not simply reductionist, and Borsboom et al. seem to be aiming at a wrong target.

Philosophically, the tension between the DSM and the RDoC is easy to appreciate. The core of the issue is whether mental states – mental disorder types, in this case – are identical with their physiological realisers. In current philosophy, this translates into the question as to whether mental states are multiply realised (Fodor Reference Fodor1974; Putnam Reference Putnam, Capitan and Merrill1967). This issue is also at the heart of the analysis Borsboom et al. are offering. However, Borsboom et al. simply take for granted that the answer to this question is positive, thus ignoring the fact that there are a number of reasons to be skeptical of multiple realisation (e.g., Bechtel & Mundale Reference Bechtel and Mundale1999; Bickle Reference Bickle1998; Reference Bickle2003; Polger & Shapiro Reference Polger and Shapiro2016; Shapiro Reference Shapiro2000). As illustrated in Figure 1, the apparent cases of multiple realisation (Fig. 1A) have the tendency to become dissolved either by kind splitting (Fig. 1B) or by realiser merging (Fig. 1C). In the former case, the purportedly multiply realised mental state or function (M) splits into two (or more) separate entities (N₁ and N₂) as it is understood that the mental state or function does not constitute a single, unified psychological or neural entity (e.g., the way that “memory” or “attention” split to several different psychological and neural functions). In the latter case, the purportedly multiply realised mental state or function is identified with a single, unified neural state or function (N) as it is understood that the different realisers are actually physiologically the same (e.g., the way that intentions to grasp objects can be identified with the average neural activity of specific neural ensembles).

Figure 1. Figure 1A represents the multiple realisation hypothesis. Figure 1B represents kind splitting. Figure 1C represents unification or merging. (Source: Pernu, Reference Pernu2018)

The RDoC framework can now be seen to aim at kind splitting: The apparently (symptomatically) homogeneous clusters of mental disorders (classified in terms of the DSM) can be predicted to split into new, homogeneous sub-clusters, each aligned with their physiological constitution (cf. Pernu Reference Pernu, Tekin and Bluhm2019). Therefore, mental disorders would be multiply realised no more. The result is not reduction (as in Fig. 1C), but elimination: Our current understanding of mental disorders – “folk psychiatry” – will be fundamentally transformed, and the symptomatically defined notions of mental disorders (M) will give way to new notions, aligned with their neural-level realisers (M₁ and M₂).

In principle, there is a more forceful argument on offer for challenging the reductionists (or eliminativists). One could claim that different mental states – different mental disorders in particular – could be realised by the same physiological states. In other words, one could claim that it is not multiple realisation, but rather “multiple realisation in reverse” that the non-reductivists should be focusing on. And, in fact, in some places in the target article, Borsboom et al. point to this sort of an analysis. Note that this line of thought does not have to be in any way particularly controversial: Neural plasticity and reuse would already indicate that the same neural basis could give rise to different mental functions (cf. e.g., Anderson Reference Anderson2010).

If mental disorders would indeed be multiple realised in reverse, that would undermine the reductionist research strategies in a quite straightforward way: We would not be able to read off mental disorders from biomarkers, for any of them could ground different mental disorders. However, not only do Borsboom et al. fail to focus their analysis on this issue, one can immediately point to a fundamental philosophical problem: Multiple realisation in reverse would breach the core idea of nonreductive physicalism – namely, the idea that the physical basis is sufficient to exhaustively fix all of the higher levels of reality (mental disorders among them), that is, the idea of mind-body supervenience. Basing the analysis on the thesis of multiple realisation in reverse (rather than on the traditional thesis of multiple realisation) would therefore be bound to amount to a Pyrrhic victory.