On the parametric stability of a central pattern generator

In this study, one subunit of the nerve network that gives rise to swimming oscillations in Hirudo was implemented in electronic hardware and characte rized for its range and stability. Based on documentation from in-vivo stud ies, the network was re-created in analog VLSI hardware on a nerve-by-nerve , synapse-by-synapse basis. The eleven cells that comprise the network are tonically active, and network topology reveals the presence of numerous rec iprocally inhibitory pairs and cyclically inhibitory rings of cells. IC-bas ed artificial nerve cells were individually configured to match the respons iveness of their living counterparts, and the network assembled. Parametric tests on tonically active single cells, reciprocally inhibitory pairs of c ells, cyclically inhibitory rings of cells, and the entire network revealed that the entire network was far more immune to variation in inward and env ironmental parameters than any suboscillator type alone. (C) 2000 Elsevier Science B.V. All rights reserved.