Borsboom et al.’s target article challenges the reductionism of psychopathology as a brain disorder. The authors argue against reductionism in neurobiological psychopathology research and propose that the focus should be on symptom-to-symptom interaction. We agree with Borsboom et al. in questioning the value of current trends in neurobiological reductionism, however, a similar problem exists in current psychosocial frameworks. We explore this challenge and underscore the specific strengths of an integrated network approach which includes both biological and psychological variables that will enhance psychopathological research.

Borsboom's network theory of mental disorders shifts our attention to symptoms and their relationship, positing that they can influence each other, creating a cascade of causal relationships (Borsboom Reference Borsboom2017) with no need for a common cause – whether neurobiological or psychosocial. We commend the symptom network approach because it proposes that symptoms might have unique relationships and thus differential functional effects. This framework allows for the exploration of multilevel mechanistic approaches at both the micro (e.g., emotion, molecular) and macro levels (e.g., symptoms, coping) that involve biological, psychological, and sociocultural processes and perspectives. This assessment is in agreement with emerging literature indicating that underlying biology, risk factors, functional impairment, and life events are differentially related to specific symptoms (Fried & Nesse Reference Fried and Nesse2015; Olbert et al. Reference Olbert, Gala and Tupler2014; Santos et al. Reference Santos, Fried, Asafu-Adjei and Ruiz2017).

The debate around reductionism in the etiology of psychopathology is not new, nor is it limited to neurobiological research. Reductionism is also a hallmark of psychological theories of psychopathology. For example, Beck's Cognitive Theory of Depression is centered on underlying dysfunctional beliefs of the individual and does not consider potential biological changes. Neurobiological and psychosocial approaches to understand the etiology of psychopathologies have long had a competing history, and this has been seen as counterproductive (Kendler Reference Kendler2008). Both approaches to psychopathology are limited in many ways by their monocausal attempts at defining these disorders: (i) All psychopathology is best explained in terms of neurobiology; and (ii) all psychopathology is best explained in terms of specified mental and/or social mechanisms which cannot be deduced from biology (Kendler Reference Kendler2008). As described in Wittenborn et al. (Reference Wittenborn, Rahmandad, Rick and Hosseinichimeh2016), this monocausal approach is reflected in a large number of studies, where 93% of articles from a literature search of major depressive disorder from 1980–2014 focused on only one key explanatory variable of depression (Wittenborn et al. Reference Wittenborn, Rahmandad, Rick and Hosseinichimeh2016). What is often seen is an intense focus on implicating a popular research topic as the key causal psychopathology factor. This focus changes from year to year without cumulative integration into a more complex and inclusive causal framework.

The network symptomology approach and neurobiology are not as mutually exclusive as Borsboom et al. suggest. The complexity of symptom network models as discussed in the target article is remarkably similar to neural network imaging studies in humans and animal models, which are just beginning to comprehensively identify and characterize the intricate biological pathways affected in mental disorders. Aberrant neural network connectivity and activity in brain networks is associated with various psychopathologies, such as depression, anxiety, autism, schizophrenia, and addiction (Woodward & Cascio Reference Woodward and Cascio2015). These brain imaging studies specifically target the complexity of mental disorders, in contrast to the portrayal in Figure 1 of Borsboom et al. On the contrary, neural network research is often more similar to the holistic approach of their Figure 2, and includes several quantifiable holistic measures, such as small worldness and correlation coefficient, that reflect the complexity of the etiology and symptomology of a multitude of psychopathologies (Drysdale et al. Reference Drysdale, Grosenick, Downar, Dunlop, Mansouri, Meng, Fetcho, Zebley, Oathes, Etkin, Schatzberg, Sudheimer, Keller, Mayberg, Gunning, Alexopoulos, Fox, Pascual-Leone, Voss, Casey, Dubin and Liston2017). Clustering coefficient, a global measure of coordinated activity within individual brain regions, has been documented in schizophrenic patients and their non-psychotic relatives, suggesting that this measure can be used to identify endophenotypes and as a biomarker and risk factor (Lo et al. Reference Lo, Su, Huang, Hung, Chen, Lan, Lin and Bullmore2015). Changes in clustering coefficient are also disrupted in pediatric posttraumatic stress disorder (PTSD) (Suo et al. Reference Suo, Lei, Li, Chen, Li, Li, Huang, Lui, Li, Kemp and Gong2015) and may be a marker for depression-related brain network pathology (Gong & He Reference Gong and He2015). Small worldness, an overall measure of neural network interactions, is often disrupted in schizophrenia (Kambeitz et al. Reference Kambeitz, Kambeitz-Ilankovic, Cabral, Dwyer, Calhoun, van den Heuvel, Falkai, Koutsouleris and Malchow2016). It is also increased in individuals exposed to childhood maltreatment, a major psychological risk factor (Ohashi et al. Reference Ohashi, Anderson, Bolger, Khan, McGreenery and Teicher2017). While many of these imaging studies do incorporate symptom data, they often depend on a single overall diagnostic value or a limited subset of symptoms, in contrast to the network symptomology approach discussed by Borsboom et al.

Observations of the limitations of current reductionist neurobiology research should be used as opportunities to enhance future investigations and strengthen communication and collaboration between psychologists and neuroscientists, especially with regard to disease etiology and behavioral symptomology. While psychiatry and psychology are moving away from monocausal explanations, the same process has been happening in behavioral neuroscience, including increased focus on multilayered gene networks (Wei Reference Wei2017), multi-hit gene x environment models (Samek et al. Reference Samek, Hicks, Keyes, Iacono and McGue2017), a greater consideration for complex transgenerational etiology (Nephew et al. Reference Nephew, Carini, Sallah, Cotino, Alyamani, Pittet, Bradburn and Murgatroyd2017), and the holistic neural network analyses (Zhang et al. Reference Zhang, Geng and Lee2017) discussed in this commentary. We suggest that what is needed is more clinically informed translational neuroscience which fully embraces the vast biological, environmental, and symptomology complexity of most psychopathologies through the inclusion of integrated psychological and biological focused network approaches. Reasonable sacrifices of experimental reductionism will reap substantial gains in replication, robustness, and translation, and ultimately result in clinical progress.