Title: Recent developments in Bayesian modelling of brain dynamics
Authors: Karl Friston - UCL (United Kingdom)
Peter Zeidman - University College London (United Kingdom) [presenting]
Abstract: Some recent developments in modelling brain activity and connectivity for cognitive neuroscientists are introduced. These methods enable experimenters to specify forward models which describe how brain circuitry gives rise to imaging data e.g. fMRI, EEG, MEG. To test hypotheses, these models can be fitted to the data and compared based on their evidence. We will provide an overview of novel methods for group connectivity studies: Parametric Empirical Bayes, PEB, and Bayesian Model Reduction, BMR. These tools can, for example, be used to distinguish patients from controls based on their brain connectivity, or to predict clinical scores. We will also describe a recent application of these methods for Bayesian fusion a multi-modal neural circuitry model which may clarify the relationship between fMRI, EEG and MEG data. Together, these methods may offer new ways to test interesting hypotheses about the brain, drawing on data from multiple imaging modalities.