Large-Scale Brain Simulations

Excerpt from Neurosciences Institute (La Jolla, California) Scientific Report 2005, in cache

The equations underlying this model provide a foundation for the development of Brain Based Devices governed by a simulated nervous system composed of spiking neurons. (Eugene Izhikevich; see publications 528, 533)

The Institute has a long-standing tradition of constructing anatomically detailed models of the brain, focused particularly on the thalamocortical system. Using the latest anatomical data on the synaptic microcircuitry of cat visual cortex, a large- scale thalamocortical model of unprecedented detail has been developed. This model contains six-layered cortical anatomy with 24 different types of multi-compartmental spiking neurons. It contains

The neurons The STDP and the delays play important roles in shaping the fine temporal structure of neuronal firing and lead to the emergence of neuronal groups.

A simpler version of the model (the same anatomy but no plasticity) was recently scaled up to the size of the human brain - 1011 neurons and 1015 synapses corresponding to 30 cm x 30 cm of cortical surface. The simulation of one second of neural activity in the model took 50 days on a Beowulf cluster of 27 processors. This is the largest simulation of an anatomically detailed neuronal model that has ever been carried out. It provides a benchmark for the resources needed to simulate models with the size of the human brain. (Eugene Izhikevich, Joseph Gally, Gerald Edelman; see publication 584)

References
528. Izhikevich, E.M. (2004) Which model to use for cortical spiking neurons? IEEE Trans. Neural Networks15:1063-1070. (2004).
533. Izhikevich, E.M. (2004) Which model to use for cortical spiking neurons? Soc. Neurosci. Abstr.29:517.2
584. Izhikevich, E.M., J.A. Gally, and G.M. Edelman (2005) Spike timing dynamics in thalamocortical models. Soc. Neurosci. Abstr. 2005 Program No. 276.14.


Version: 12.4.2013
URL of this page
Home
Joachim Gruber