C. Rasche, R.J. Douglas, M. Mahowald, Institute for Nueroinformatics, ETH/UZ, Gloriastr 32, CH-8006 Zurich, Switzerland.
The Silicon Neuron (SN) is an analog VLSI circuit that has the functional characteristics of real neurons. It emulates efficiently many of the ion currents that generate action potentials and control the dynamics of their discharge. The ion currents are modelled according to the Hodgkin-Huxley principles: Time dependence is simulated by a follower integrator. Voltage dependence is mimicked by the differential pair and its related devices which have a sigmoidal current voltage relation similar to that of voltage dependent membrane channel conductances in their steady state. By simulating intra-cellular calcium concentration we obtain spike frequency adaptation in various forms like real pyramidal cells. Further typical characteristics are non-linear subthreshold current voltage relation and discharge behavior like point processes. The parameters of these circuits can be set so that the Silicon Neuron emulates a particular class of biological neurons. We present here how our SN simulates a cortical pyramidal cell.