Cognitive network modelling employs mathematical models to infer hidden neural mechanisms from recordings of behaviour and brain activity–functional magnetic resonance imaging (fMRI) and electroencephalography (EEG) data–acquired during cognitive tasks. We utilize a suite of mathematical models, including hierarchical Bayesian, reinforcement learning, and machine learning algorithms, to infer underlying cognitive and neuronal mechanisms.

On one axis, our research examines how basic learning signals, such as predictions and “correction” signals or prediction errors, are represented in the brain using probabilistic learning tasks and multimodal imaging (EEG and fMRI). This foundational knowledge serves as a substrate for our parallel line of inquiry that investigates how distortions in the representation of these learning signals contribute to maladaptive behaviours and pathological states. Specifically, we are interested in how these mechanisms could underpin the neurobiology of severe mental disorders, including schizophrenia and major depression.