Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network

  Advanced Search  

Beyond Bumps: Spiking Networks that Store Smooth n-Dimensional Functions

Summary: Beyond Bumps: Spiking Networks that Store Smooth
n-Dimensional Functions
Chris Eliasmith and Charles H. Anderson
There are currently a number of models that use spiking neurons in recurrent net-
works to encode a stable Gaussian `bump' of activation. These models successfully
capture some behaviors of various neural systems (e.g., storing a single spatial location
in parietal cortex). We extend this previous work by showing how to construct and ana-
lyze realistic spiking networks that encode smooth n-dimensional functions drawn from
a finite functional space. These new networks can capture additional experimentally
observed behavior (e.g., storing multiple spatial locations at the same time).
1 Introduction
Stable Gaussian shaped neuronal activities across a population of recurrently connected
neurons without external perturbation, or stable `bumps', have been successfully mod-
eled by a number of researchers [8, 7, 6]. Bumps have been thought to be present in
various neural systems including the head direction system [5], frontal working memory
systems [8], parietal reach memory systems [4], and feature selective visual systems [3].
Many of these systems can store functions more complicated than a simple Gaussian
bump. For example, there is evidence that parietal areas can hold multiple saccade
targets in memory at the same time, suggesting that a multi-modal function is stored


Source: Anderson, Charles H. - Departments of Anatomy and Neurobiology & Physics, Washington University in St. Louis


Collections: Computer Technologies and Information Sciences; Biology and Medicine