Versatile networks of simulated spiking neurons displaying winner-take-all behavior.

Yanqing Chen,Gerald M Edelman,Jeffrey L Mckinstry

doi:10.3389/fncom.2013.00016

Yanqing Chen, Gerald M Edelman + Show 1 more

Open Access

https://doi.org/10.3389/fncom.2013.00016

Copy DOI

Abstract

We describe simulations of large-scale networks of excitatory and inhibitory spiking neurons that can generate dynamically stable winner-take-all (WTA) behavior. The network connectivity is a variant of center-surround architecture that we call center-annular-surround (CAS). In this architecture each neuron is excited by nearby neighbors and inhibited by more distant neighbors in an annular-surround region. The neural units of these networks simulate conductance-based spiking neurons that interact via mechanisms susceptible to both short-term synaptic plasticity and STDP. We show that such CAS networks display robust WTA behavior unlike the center-surround networks and other control architectures that we have studied. We find that a large-scale network of spiking neurons with separate populations of excitatory and inhibitory neurons can give rise to smooth maps of sensory input. In addition, we show that a humanoid brain-based-device (BBD) under the control of a spiking WTA neural network can learn to reach to target positions in its visual field, thus demonstrating the acquisition of sensorimotor coordination.

Highlights

Analyses in computational neurobiology have successfully used mean-firing-rate neuronal models to simulate the spatiotemporal patterns of neural activity that arise in interconnected networks of excitatory and inhibitory neurons, such as those in the vertebrate cortex (Von der Malsburg, 1973; Obermayer et al, 1990; Dayan and Abbott, 2001)
We present the model in a dimensional form so that the membrane potential is in millivolts, the current is in picoamperes and the time is in milliseconds: Cv = k(v − vr)(v − vt ) − u − Isyn
Connection strengths were not modulated by spiketiming-dependent plasticity (STDP) but were subject to the short-term synaptic plasticity inherent in modeled neurons (Izhikevich and Edelman, 2008)

Summary

Introduction

Analyses in computational neurobiology have successfully used mean-firing-rate neuronal models to simulate the spatiotemporal patterns of neural activity that arise in interconnected networks of excitatory and inhibitory neurons, such as those in the vertebrate cortex (Von der Malsburg, 1973; Obermayer et al, 1990; Dayan and Abbott, 2001) Certain aspects of these systems may, require the modeling of the dynamic properties of large populations of individual neurons, each calculated with millisecond precision. It has been proposed that local microcircuits of the cerebral cortex can function as winner-take-all (WTA) networks (Douglas and Martin, 2004) In such systems, an individual pattern of input can evoke network responses that suppress possible alternative responses. Indirect physiological evidence (Derdikman et al, 2003; Haider et al, 2010) has been obtained for local excitation and surround inhibition in the cerebral cortex of mammals

Methods

Results

Conclusion