Computational Aspects of the Respiratory Pattern Generator

Abstract

To help evaluate the hypothesis that the central respiratory rhythm is generated by a network of interacting neurons, a network model of respiratory rhythmogenesis is formulated and examined computationally. The neural elements of the network are driven by tonic inputs and generate a continuous variable representing firing rate. Each neural element in the model can be described by an activation time constant, an adaptation time constant, and a step nonlinearity. Initial network connectivity was based on an earlier proposed model of the central respiratory pattern generator. These connections were adjusted interactively until the model trajectories resembled those observed electrophysiologically. The properties of the resulting network were examined computationally by simulation, determination of the phase resetting behavior of the network oscillator, and examination of the localized eigenstructure of the network. These results demonstrate that the network model can account for a number of diverse physiological observations, and, thus, support the network hypothesis of respiratory rhymogenesis.