A one-dimensional spiking neural network model of the midbrain superior colliculus

We propose a biologically realistic spiking neural network that accounts for the dynamic spatiotemporal transformations within the midbrain superior colliculus (SC) motor map for saccadic eye movements. The model is constrained by observed firing patterns of saccade-related SC cells, where burst durations and peak firing rates vary systematically with their location in the motor map, while keeping a constant number of spikes in their bursts. Our functional network model reproduces the spike trains of single cells in an SC population encoding visually-evoked saccades. In our one-dimensional network the SC neurons are described by adaptive integrate-and-fire models, and lateral excitatory-inhibitory connections. The network scheme is suitable for a full 2D extension. Furthermore, the model offers a basis for neuronal algorithms for spatiotemporal transformations and bioinspired optimal control signal generators.