• Type:#article
  • Year read:#read2022
  • Subject: (in brackets, can also bracket keywords in text)
  • Bibtex: @roefs2022
  • Bibliography: Roefs, A., Fried, E. I., Kindt, M., Martijn, C., Elzinga, B., Evers, A. W. M., Wiers, R. W., Borsboom, D., & Jansen, A. (2022). A new science of mental disorders: Using personalised, transdiagnostic, dynamical systems to understand, model, diagnose and treat psychopathology. Behaviour Research and Therapy, 153, 104096. https://doi.org/10.1016/j.brat.2022.104096

Example citation

Key takeaways

  • They outline their big network approach research programme:
    • Mapping of networks in a big cohort
    • Experimental pre-clinical studies
    • Network-informed interventions

Critique of DSM and ICD

  • Risk factors and symptoms are continuously distributed in the population
  • Comorbidity rampant but single-diagnosis treatment

Transdiagnostic approaches

  • RDoC
  • Unified Protocol
  • Transdiagnostic mechanisms (e.g., rumination, sleep problems)
  • HiTOP

The focus (of RDoC) is on conceptualising mental disorders as brain disorders, and it is mostly biologically oriented. The framework reflects a dynamic research strategy that aims to clarify basic biological, cognitive and behavioural processes, but it is not meantto serve as a diagnostic guide or to replace current diagnostic systems.

Network approach

  • The symptoms constitute the disorder, rather than being caused by an underlying entity
  • Mental disorders are stable network states, outside events can trigger transition into another stable state (getting treatment, adverse life events)
    • Resilience to outside influence is determined by number of feedback relations, how easy it is to activate symptoms in the first place, and strength of connections
  • Comorbidity is understood as overlapping symptoms, and represent real complex relations between disorders as currently defined by DSM/ICD

Overall, the network approach assumes spreading activation to be the engine that drives the mental disorder: connected elements synchronise, maintain each other actively in vicious cycles, and become a self-sustaining entity.

Common cause (DSM) and network approach do not need to be mutually exclusive

Proximal vs Distal causes.

  • Proximal causes directly related to symptom mechanisms, usually a node or edge in the network
  • Distal causes are indirectly related to symptoms and can be represented as forces outisde the network (genes, personality, SES, early adversity)
  • Proximal causes: experimental research
  • Distal causes: Epidemiology & longitudinal research

Borrowing from other network fields

Network approach to psychopathology has much to learn both conceptually and can borrow statistical tools from these related fields.

  • Eco-systems, stock market, the weather etc.
  • Studying elements of systems will not be sufficient to understand the macro-level behaviour of such systems.
  • One potential tool: identifying early warning sign that predict transitions between states

Limitations of network studies to date

  • Single-disorder
  • Cross-sectional (between-individuals)
  • Self-rated

Single-disorder and cross-sectional data to date cannot account for processes bridging diagnostic boundaries. Between-individual approach can’t look into personalised dynamic processes and mechanisms within-individuals.

Disentangling nomothetic relations that hold for most people, versus idiographic relations that are unique to certain individuals, is crucial if we want to keep the promise of the network approach as a new paradigm for the treatment of mental disorders.

Their project: Mapping, Zooming, Targeting


Estimating networks to inform diagnosis and treatment.

  • Time-series data collection and analysis
  • Network-informed diagnoses
  • Digital phenotyping
  • EMA


Experimental research to identify causal processes.

  • Manipulating network elements (such as central symptoms) to study their causal role
  • Forecasting relapses and gains, tipping points and early warning signals. Similarities to sudden gains

For example, when targeting anxiety, binge drinking may improve as well. This is a desired outcome in therapy, but it is difficult to tell whether recovery was hastened by first reducing the anxiety, the binge drinking, or both. Note that simultaneous improvements across one or more nodes in addition to the target node may still be informative, though it does not permit drawing conclusions about a causal role of the targeted element.


Interventions informed by networks. Comparing them to interventions that are not network-informed. The goal of a network-based intervention is to bring back the system to a healthy state, by targeting nodes and edges that are influential.

How to choose target Network nodes? Some indicators are Node strength, Node closeness. One difficulty is that the psychological network nodes can’t be observed but only estimated. There’s also multicollinearity between symptoms, and some symptoms are more severe than others. New centrality measures may be needed.

Some open questions

  • Should treatment targets be based on symptom intensity and/or on centrality indices, or other aspects?
  • What variables besides symptoms should be included in analyses?
  • What broader context beyond the ESM (experience sampling methodology) data is needed (clinical theory, expertise)?

Careful evaluation

  • Does network-based interventions actually change network structures?
  • Do they act differently than control interventions?
  • Does change in network structure lead to relief of symptoms?
  • Selecting techniques: from already existing toolbox, e.g., exposure or insomnia techniques.


  • Ecological Momentary Assessment (EMA) during waitlist for therapy
  • Therapists need easy-to-use tools that help in treatment planning
  • Therapists need training in network-informed interventions