Generating arbitrary rhythmic patterns with purely inhibitory neural networks

A novel approach for the prediction and generation of coupled neural oscillation among arbitrarily connected inhibitory neurons is proposed. Based on Scheduling by Multiple Edge Reversal (SMER), a very simple distributed algorithm, neural network building blocks can be configured for the generation of complex rhythmic patterns with a very high independence from individual neuronal models. A method for the organization and simulation of the new approach is illustrated by mimicking the main rhythmic gait patterns of an hexapodal animal.