Dear All

This afternoon (YNiC open plan, 4.15 pm start) Prof. Gary Green will be giving a talk on "Connectivity maps based on linear and non-linear dynamics".

Abstract: Many methods have been used to provide evidence of the existence of a connection from one brain area to another. The evidence can be from structural investigations such as DTI and histology or from functional investigations using techniques such as Granger Causality, coherence or other non-causal methods. These tell us that connections exist but they do not necessarily provide evidence of the form of the interaction both in terms of direction, the nature of the processing or the dynamics of the connection. Two 'new' approaches to investigating the dynamics of connectivity will be presented. These are the Adomian Decomposition method and the Fliess Generating Series method. Both are highly related and are based on functional expansions of non-linear differential models of the brain. The can be used with standard ODEs, but can also be used with PDEs, delay-differential and stochastics models Both can be used to compare and eshew models, to fit models to data and to test if existing models are supported by experiment. In passing we show that SPM's DCM approach may suffer from highly correlated posteriors. The approaches are causal and can inform us about directionality, the timing of connections as well as the level and form of linear/non-linear processing. We have implemented them in a simple-to-use set of modules in Python. MEG data from a reading task will be used to demonstrate the techniques and to show evidence that brain areas are causally,directionally connected, on different time scales during this task.

Everyone is welcome to attend. Best wishes Rebecca