Imaging neuroscience and theoretical neurobiology

Professor Karl Friston FRS

The Group develops advanced mathematical techniques that allows researchers to characterise brain organization. This involves creating models of how the brain is wired and how it responds in different situations. These models are used to interpret measured brain responses using brain imaging and electromagnetic brain signals.

Work within this Group has been concerned with two issues.

First, in imaging neuroscience, there have been considerable advances in Bayesian estimation and inference in causal models. The Group has developed procedures to make Bayesian inferences about evoked brain responses using posterior probability maps. This, and other Bayesian estimation advances in the context of hemodynamic responses and dynamic causal modeling, rests upon some basic work using expectation maximization in the context of hierarchical models. Work on dynamic causal modeling of interactions among brain areas has been implemented in the SPM software package. Much of this work is a prelude to integrating EEG and fMRI.

Second, the programme has developed generative models of brain function and mean field approximations to characterise ensemble dynamics in neuronal populations. The former theme has focused on backward connections in the brain and their role in perceptual synthesis. The framework lends itself to a Bayesian interpretation in the context of hierarchical generative or forward models that could be plausibly implemented in cortical hierarchies. This theoretical work accommodates many previous imaging results and will be used to motivate questions that will be addressed using multi-modality imaging (EEG and fMRI).

The UCL Institute of Neurology promotes teaching and research of the highest quality in neurology and the neurosciences. The Institute of Neurology and the National Hospital for Neurology and Neurosurgery are members of  UCL Partners, Europe's largest academic health science partnership.