Wulfram Gerstner
Keynote speakers
- Kenji Doya
- Alon Halevy
- Astrid Prinz
- Andrew Schwartz
- Shankar Subramaniam
- Arthur Toga
Workshop speakers
- Bart ter Haar Romeny
- Uri Eden
- Klaus Linkenkaer-Hansen
- Tim Clark
- Alan Ruttenberg
- Jeffrey Grethe
- Arnd Roth
- Wulfram Gerstner
- Peter Hunter
- Markus Diesmann
- Andrey Semin
- Pietro Liò
- Albert Cardona
- Giorgio Ascoli
- Rolf Kötter
Workshop 3, How should a neuron be modeled: Biophysical detail vs. abstraction
Wulfram Gerstner
Title: The power and limits of simple neuron models: predicting neural activity spike by spike
École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland
Abstract: Most neuronal data comes from experiments with one electrode per neuron. I argue that for this type of data, single-compartment models of the integrate-and-fire type are not only sufficient but potentially beat more detailed biophysical models. We use data from slice experiments with single neurons under current or conductance injection. The input signal changes on several time scale - from a few to hundreds of milliseconds. While a standard leaky integrate-and-fire neuron is not capable to reproduce this data, variants of integrate-and-fire models are. Because of the simple model structure, parameters of the models can be directly extracted from the data. I present results with the adaptive exponential integrate-and-fire model and with a Spike-Response Model with adaptive threshold for a task where the timing of spikes has to be predicted with a precision of a few milliseconds. The model parameters were optimized on a first set of data (training set) and the performance of the model is measured on an independent set (validation set).
We find that for the independent input set, up to 90 percent of predictable spikes are correctly predicted by the model. Subthreshold voltage is modeled with a precision of one to two millivolt.
References:
Laurent Badel, Sandrine Lefort, Thomas K. Berger, Carl C. H. Petersen, Wulfram Gerstner and Magnus J. E. Richardson (2008c) Extracting non-linear integrate-and-fire models from experimental data using dynamic I-V curves
http://www.springerlink.com/content/l231321784j17157/fulltext.pdf
Biological Cybernetics 99:361-370
Laurent Badel, Sandrine Lefort, Romain Brette, Carl Petersen, Wulfram Gerstner and Magnus J.E. Richardson (2008) Dynamic I-V Curves Are Reliable Predictors of Naturalistic Pyramidal-Neuron Voltage Traces,
http://lcn.epfl.ch/%7Egerstner/PUBLICATIONS/Badel08.pdf
J Neurophysiol 99: 656 - 666
Renaud Jolivet, Felix Schuermann, Thomas K. Berger, Richard Naud, Wulfram Gerstnera, and Arnd Roth (2008b) The quantitative single-neuron modeling competition
http://www.springerlink.com/content/e1601737g4p876j7/fulltext.pdf
Biological Cybernetics 99:417-426
http://lcn.epfl.ch/%7Egerstner/PUBLICATIONS/Jolivet08b.pdf
R. Jolivet, A. Rauch, H.-R. Luscher and W. Gerstner (2006) Predicting spike timing of neocortical pyramidal neurons by simple threshold models
http://lcn.epfl.ch/%7Egerstner/PUBLICATIONS/Jolivet06_abs.html
Journal of Computational Neuroscience 21:35-49
Bio sketch: Wulfram Gerstner is Director of the Laboratory of Computational Neuroscience LCN at the EPFL. He studied physics at the universities of Tubingen and Munich and received a PhD from the Technical University of Munich. His research in computational neuroscience concentrates on models of spiking neurons and spike-timing dependent plasticity, on the problem of neuronal coding in single neurons and populations, as well as on the role of spatial representation for navigation of rat-like autonomous agents. He currently has a joint appointment at the School of Life Sciences and the School of Computer and Communications Sciences at the EPFL. He teaches courses for Physicists, Computer Scientists, Mathematicians, and Life Scientists.