Neural Information Dynamics - Information theoretic identification of neural algorithms:

Almost everyone agrees that brains exist to process information. Yet, it is much harder to give a detailed and quantitative description of this 'information processing'. This is particularly true when we try to describe the operating principles of neural circuits in terms that do not directly to specific tasks, such as detecting faces or keeping items in working memory. Such a description without reference to specific tasks is absolutely necessary when we want to understand information processing principles, or algorithms, that are conserved over many species and neural structures.
My group focuses on detecting such operating principles in the neocortex, where similar wiring motifs of cortical circuits are conserved over a wide variety of brain areas that are normally associated with vastly of different functions. Yet, their shared circuit architecture suggests some underlying similarity in terms of information processing. In other words the vastly different functions seem to be served by one or a few conserved neural algorithms
The natural choice for describing such a neural algorithm without reference to very specific tasks is information theory, as it is inherently free of semantics of specific tasks. Especially the emerging field of information dynamics, with its ability to separate and measure the active storage, transfer and modification of information time step by time step is of key importance. This is because the tools of infroamtion dynamics allow us to test hypotheses about cortical information processing directly. This is done by first deriving the relevant information theoretic fingerprint of a hypothesized processing strategy and by then comparing it to those information theoretic fingerprints obtained from neural recordings
My groups develops novel methods in the field of neural information dynamics and the identification of cortical algorithms. We specifically focus on information theoretic tests of predictive coding theories, and collaborate with leading electro-physiologists in analyzing the massive parallel recordings that have become available recently.

For details on our work see the recent publications and abstracts on my Google Scholar profile.