An Optimization framework for Nonspiking Neuronal Networks and Aided Discovery of a Model for the Elementary Motion Detector

摘要: We present a general optimization procedure that given parameterized network of nonspiking conductance based compartmentally modeled neurons, tunes the parameters to elicit desired behavior. Armed with this tool, we address elementary motion detector problem. Central established theoretical models, Hassenstein-Reichardt and Barlow-Levick detectors, are delay lines whose outputs from spatially separated locations prescribed be nonlinearly integrated direct engender direction selectivity. The neural implementation delays---which substantial as stipulated by interomatidial angles---has remained elusive although there is consensus regarding neurons constitute network. Assisted procedure, identify parameter settings consistent connectivity architecture physiology Drosophila optic lobe, demonstrates requisite concomitant selectivity can emerge nonlinear dynamics small recurrent networks simple tonically active synapses.