Proposal for NIPS*95 Post-Conference Workshop

``Network Models of Cortical Function''

Terence D. Sanger

MIT Room E25-526 Cambridge, MA 02139

tds@ai.mit.edu, (617)253-5769, fax - (617)258-5342

The similarity of cellular architecture between different regions of vertebrate cerebral cortex suggests the possibility of a small number of computational mechanisms underlying cortical function. One role of artificial Neural Network models is to investigate these mechanisms through simulation of large collections of units whose properties are derived from neurophysiological measurements. This workshop will draw on models from several different cortical areas in order to address the following questions:

1. Are there uniform computational mechanisms that exist across different regions of cortex devoted to varied sensory and motor modalities?

2. Have theoretical models been successful in guiding new neurophysiological investigations?

3. How faithfully do models need to follow ultrastructural or membrane properties of cortical neurons in order to generate realistic predictions of function?

4. Have artificial Neural Network algorithms gained from Neurophysiological investigations, or is all transfer of technology ``one-way''?

5. To what extent do theoretical models address more than one of David Marr's levels of algorithmic abstraction (general approach, specific algorithm, and hardware implementation)?

6. Is it possible to combine existing models of separate cortical regions into larger ``brain-like'' units?

The workshop is planned as a single day panel discussion including both morning and afternoon sessions. Four speakers per session will be asked to limit presentations of relevant research to 10 minutes and 4 slides. Each speaker will describe computational models of different cortical regions, and speakers will be encouraged to present only the simplest overview of algorithms and results in a manner that will allow direct comparison between methods. The intended audience includes researchers actively involved in cortical modelling, as well as a general audience that can contribute viewpoints from different backgrounds within the Neural Networks field.

• About this document ...