Cognitive Network Modelling

The goal of the Cognitive Network Modelling team is to characterise disease states using mathematical modelling and symptom-relevant design of tasks. We validate and apply mathematical models that infer subject-specific pathophysiology from non-invasive measures of behaviour and neuronal activity.

These models aim to quantify both physiological and computational principles that underlie (mal)adaptive cognition, including aberrant learning and decision-making, in individual participants. Hierarchical models of brain dysconnectivity applied to EEG and fMRI measurements are of particular interest: How does Bayesian learning go wrong when integrating prior beliefs with novel information?

The long-term goal is to use these models for a mechanistic re-definition of psychiatric diseases leading to pathophysiologically interpretable diagnostic classifications and individual treatment predictions.

Our main lines of research are:

  1. Design of cognitive tasks focused on specific symptoms probing different aspects of maladaptive cognition in psychiatric disorders.
  2. Development of modelling techniques for inferring effective connectivity, synaptic plasticity and neuromodulation from fMRI and EEG data, e.g. hierarchical Bayesian modelling, dynamic causal modelling (DCM), and Bayesian model selection (BMS).
  3. Experimental neuroimaging studies on the physiological determinants of individual mechanisms underlying aberrant learning and decision-making.
  4. Systematic model validation in physiological, pharmacological and patient studies.
  5. Clinical applications: Model-based diagnostic classifications that are pathophysiologically interpretable and allow for individual treatment predictions.