Hardware computation of conductance-based neuron models

We review di4erent applications of silicon conductance-based neuron models implemented on analog circuits. At the single-cell level, we describe a circuit in which conductances are programmed to simulate various Hodgkin–Huxley type models; integrated in a hardware/software system, they provide a simulation tool; an illustrative example is the simulation of bursting neurons of the thalamus. At the network level, we present a mixed analog–digital architecture, where the connectivity and the plasticity rules are implemented digitally and are therefore very 8exible. These circuits provide valuable tools for real-time simulations, including hybrid applications where single-neuron or network models are interfaced with biological cells. c © 2004 Elsevier B.V. All rights reserved.