In the past decades, reductionism has dominated both research directions and funding policies in clinical psychology and psychiatry. The intense search for the biological basis of mental disorders, however, has not resulted in conclusive reductionist explanations of psychopathology. Recently, network models have been proposed as an alternative framework for the analysis of mental disorders, in which mental disorders arise from the causal interplay between symptoms. In this target article, we show that this conceptualization can help explain why reductionist approaches in psychiatry and clinical psychology are on the wrong track. First, symptom networks preclude the identification of a common cause of symptomatology with a neurobiological condition; in symptom networks, there is no such common cause. Second, symptom network relations depend on the content of mental states and, as such, feature intentionality. Third, the strength of network relations is highly likely to depend partially on cultural and historical contexts as well as external mechanisms in the environment. Taken together, these properties suggest that, if mental disorders are indeed networks of causally related symptoms, reductionist accounts cannot achieve the level of success associated with reductionist disease models in modern medicine. As an alternative strategy, we propose to interpret network structures in terms of D. C. Dennett's (1987) notion of real patterns, and suggest that, instead of being reducible to a biological basis, mental disorders feature biological and psychological factors that are deeply intertwined in feedback loops. This suggests that neither psychological nor biological levels can claim causal or explanatory priority, and that a holistic research strategy is necessary for progress in the study of mental disorders.

Reductionist thinking in neuroscience is manifest in the widespread use of animal models of neuropsychiatric disorders. Broader investigations of diverse behaviors in non-model organisms and longer-term study of the mechanisms of plasticity will yield fundamental insights into the neurobiological, developmental, genetic, and environmental factors contributing to the “massively multifactorial system networks” which go awry in mental disorders.

Borsboom et al. discuss the implications of network structures for neurobiology-based reductionism, but inherent in the network approach is that dimensional models of psychopathology are untenable as well. Insofar as mental disorders are complex dynamic constellations of symptoms, the “trait reductionism” of dimensional psychopathology frameworks suffers from the same limitations as neurobiological reductionism.

Culture contextualizes the contents and intentionality of many mental statuses. Cognitive mediation of cultural information shapes these contents and intentionalities, as well as many of the false beliefs of pathology. Flexibility of cognitive mediation processes and resulting beliefs and pathologies may vary by individual, be a key mechanism of the feedback loop, and help characterize network connections.

In this commentary on Borsboom et al.’s target article, we argue that researchers should be aware of the historical development of models in neuroscience. Considering the importance of causality in anatomo-clinical approach and stressing the complexity of mental phenomenon, we provide new insight on reductionism and representation limitation.

Progress in psychiatric research has been hindered by the use of artificial disease categories to map distinct biological substrates. Efforts to overcome this obstacle have led to the misconception that relevant psychiatric dimensions are not biologically reducible. Consequently, the return to phenomenology is once again advocated. We propose a process-centered paradigm of biological reduction compatible with non-reductive materialism.

Borsboom et al. propose that the network approach blocks reductionism in psychopathology. We argue that the two main arguments, intentionality and multiple realizability of mental disorders, are not sufficient to establish that mental disorders are not brain disorders, and that the specific role of networks in these arguments is unclear.

Borsboom et al.’s formulation provides an opportunity for a fundamental rethink about the “brain disease model” of addiction that dominates research, treatment, policy, and lay understanding of addiction. We also demonstrate how the American opioid crisis provides a contemporary example of how “brain disease” is not moderated by the environmental context but is instead crucially dependent upon it.

The target article by Borsboom et al. proposes network models as an alternative to reductionist approaches in the analysis of mental disorders, using mood disorders such as depression and anxiety as examples. We ask how this framework can be applied to neurodevelopmental disorders such as autism spectrum disorder (ASD) and attention deficit hyperactivity disorder (ADHD). Specifically, we raise a number of promises and challenges when conceptualizing neurodevelopmental disorders as networks.

Borsboom et al. confuse biological approaches with extreme biological reductionism and common-cause models of psychopathology. In muddling these concepts, they mistakenly throw the baby out with the bathwater. Here, we highlight recent work underscoring the unique value of clinical and translational neuroscience approaches for understanding the nature and origins of psychopathology and for developing improved intervention strategies.

A network structure explains why reductionism is not possible for mental illness, but the same argument applies for the somatic symptoms of functional disorders. Because the covariation of symptoms of functional disorders cannot be explained in terms of symptom-to-symptom causality, explanation requires a network of biological mechanisms having emergent properties that cannot be reduced to biology.

Neurobiology-based interventions for mental diseases and searches for useful biomarkers of treatment response have largely failed. Clinical trials should assess interventions related to environmental and social stressors, with long-term follow-up; social rather than biological endpoints; personalized outcomes; and suitable cluster, adaptive, and n-of-1 designs. Labor, education, financial, and other social/political decisions should be evaluated for their impacts on mental disease.

Borsboom et al. correctly note that the use of latent variable models in cross-cultural research has resulted in a futile search for universal, biological causes of psychopathology; however, this is not an inevitable outcome of such models. While network analytic approaches require further development, network models have the potential to better elucidate the role of cultural and contextual variables related to psychopathology.

Borsboom et al. have written a trenchant critique of biological reductionism in psychopathology. After commenting on recent controversies concerning the network perspective, I discuss ways of integrating biology into the network enterprise.

Borsboom and colleagues argue that reductionism in psychopathology research has not provided the expected insights. Instead, they suggest a systems approach of interacting syndromes, which, however, falls short of a perspective for empirical testing. Here, a combination of both approaches is suggested: a reductionistic empirical approach allowing testability, synergistic with a constructivistic systems appraisal of syndrome networks – a constructive reductionism.

We agree with Borsboom et al. in challenging neurobiological reductionism, and underscore some specific strengths of a network approach. However, they do not acknowledge that a similar problem is present in current psychosocial frameworks. We discuss this challenge as well as describe valuable parallels between symptom and neurobiological network theories that will substantially augment psychopathological research when integrated.

The network approach as a novel way of understanding psychopathology has helped address some of the existing issues associated with traditional biological interpretations. Nonetheless, it has similarly failed in explaining the fundamental etiology of mental conditions – a persistent conundrum that arguably could be adequately addressed only by evolutionary formulations, specifically evolutionary mismatch and life history theories.

Modeling the extremes of mental/emotional conditions requires explicit accounts of evolutionary-developmental sources of human neurodiversity, not merely psychopathology. The target article's approach could be improved by incorporation of a hierarchical scheme wherein mental/emotional infrastructure interacts across differentiated layers of function. The notion of “symptom networks” thus calls for differentiation into hierarchically interacting components of mental/emotional evolution and development.

In experimental psychology and neuroscience, technological advances and multisensory research have contributed to gradually dismiss a version of reductionism. Empirical results no longer support a brain model in which distinct “modules” perform discrete functions, but rather, a brain of partially overlapping networks. A similarly changed brain model is extending to psychopathology and clinical psychology, and partly accounts for the problems of reductionism.