Exploring Filtration Effects in Dendritic Morphology: a Cable-Theoretic Approach

3pm on Wednesday, November 2nd 2011 (Complexity Seminar Room, D1.07)

Abstract

Neural networks are powerful tools for computing, learning and predicting. It is usually thought that it is the network's connectivity that provides information-processing capability, and thus, modellers of biological neural networks typically use simple point neuron models, with no spatial extent. In reality, neurons are connected through highly-branching processes called dendritic trees, whose geometry is thought to affect the signal being communicated as if by a spatiotemporal filter. In this talk, I aim to motivate the study of dendritic filtering and discuss several approaches to solving cable theory problems, with the objective of understanding what the filtration properties of dendritic trees are, and where they come from - morphology, topology, or connectivity.