Next: Contents

Notes on

Information Maximization in Single Neurons

Martin Stemmler Innovationskolleg Theoretische Biologie Humboldt-Universität zu Berlin Invalidenstr. 43 D-10115 Berlin Christof Koch Computation and Neural Systems Program California Institute of Technology 139-74 Pasadena, CA 91125 The senses provide a rich array of inputs to the nervous system, such that the amplitude and time scale of stimuli span many orders of magnitude-- from the noise of a leaf falling to the ground in the forest to the continuous racket of a jackhammer at an urban construction site. By adapting to the synaptic inputs a neuron receives from its own ``environment", a single neuron must find a compact and efficient representation of natural stimuli.

This set of notes is meant to accompany the paper "Adaptation of Ionic Conductances to Maximize the Information in a Neuron's Firing Rate. " A short paper entitled "Information Maximization in Single Neurons" is accessible via the underlined link.

Time evolution of the firing rate probability distribution: if the peak firing rate is limited, the best neuronal strategy is to use all firing rates equally often in response to a range of stimuli: this corresponds to the uniform (flat) probability distribution.

Contents

Martin Stemmler

stemmler@www.klab.caltech.edu

3/8/1998