Applying information-theoretic measures to computation and communication in neural ensembles

The novel methods used to develop brain machine interfaces (BMIs) present systems neuroscientists with the opportunity to now ask fundamentally new types of questions. This paradigm shift has been driven by three key developments. First, it is now possible to perform simultaneous, massively parallel recording of brain signals from microelectrode arrays chronically implanted in multiple brain areas over several months. Second, advances in computing power enables sophisticated, real-time signal processing to be performed on these large data sets. Third, recent theoretical advances clarify how the mutual information between individual neurons or between distinct neuronal populations could play a key role in neural coding. We contend these advances present a high-leverage opportunity to accelerate understanding of neuronal computation and communication.