Friday, October 22, 2004

12 noon

Redwood Neuroscience Institute

 

Title " Information Transmission with Renewal Neurons."

 

Carl van Vreeswijk

Neurophysics and Physiology of the Motor System

René Descartes University, Paris, France

 

Abstract:

It is becoming increasingly common to study the statistics of neural data using information theory. Unfortunately the result of such an analysis is usually only compared to that of Poisson processes. This null hypothesis can almost always be rejected. Because real neurons, unlike Poisson processes, have refractory periods, this is extremely unsurprising. I will present the result of a study of information theoretic measures  applied to generalized renewal processes, and will show that many of the features found in information theoretic results from real spike trains can be reproduced with renewal processes.