Applied Math Colloquium

Friday, Oct 19, 2012, 11:30 AM
Cullimore Lecture Hall, Lecture Hall II
New Jersey Institute of Technology


Temporal Coding in Spiking and Bursting Neurons

German Mato


Balseiro Institute, Argentina



It is generally assumed that neural systems perform some kind of computation. In order to elucidate what computation is being done we have to understand the relation between environmental stimuli, behavior and neural responses. This is the coding problem. Typical questions are which features of the response convey more information about the stimuli, how correlations between responses affect the transmission of information or what is the capacity of the code to convey information. These questions, that can be addressed using techniques of Information Theory, are not different from what one would ask about any other coding and communication system (or an input-output black box). Neural systems, on the other hand, have very specific dynamical properties, that arise from the properties of their ionic channels. The specific nature of information processing in the nervous system is determined by the details of this neural hardware, but given the complexity of these highly non-linear dynamical systems this relationship has not been unraveled in a systematic manner. We address this question using a different approach: we first analyze very simple dynamical models with the hope that they will reveal aspects of the neural code that cannot be uncovered using conventional black box approaches. These simplified systems can be linked to more realistic ones using tools of theory of dynamical systems (such as bifurcation theory). We will show that the temporal neural code of spiking and bursting neurons depends critically on the mathematical properties of the underlying dynamical system. Moreover, the analysis reveals which one of the coding strategies bursting neurons can use (among all the possible ones).