Unraveling Networks: The Conceptual Incoherence of the Network Approach

The network approach to psychopathology promises (personalized) visualizations of the structure of mental disorders, in turn allowing for prediction of disorder development and identification of intervention targets. In this paper, we argue that these goals are problematized by conceptual incoherence between the approach's two pillars: the network theory of mental disorders, and the suite of methods known as "network psychometrics." First, we argue that, if it is to be clinically informative, network psychometrics requires (1) an embedding argument to justify that a causally sufficient set of variables has been modeled, and (2) a concomitant argument supporting the assumption that the human-environment system is near-decomposable. Then, we show that current versions of the network theory of psychopathology raise severe obstacles for such arguments, if not outright blocking them. In turn, the network approach to psychopathology appears to unravel unless changes are made to its pillars.