Aurel Lazar, Reconstructing Natural Visual Scenes From Spike Times


A single neuron is not sufficient to represent an entire visual scene. An ensemble of neurons with different receptive fields tiling the visual space is needed. Fig. 2 schematically illustrates a class of neural circuit architectures for encoding visual scenes (signals). A space-time varying signal is encoded by a parallel neural circuit consisting of neurons, each composed of a spatio–temporal DSP cascaded with a BSG. The output of the neural circuit is a multidimensional spike train whose spike times are observable.