ENCODING CHAOS IN NEURAL SPIKE TRAINS

Recently, it has been shown that interspike interval (ISI) series from driven model neurons can be used to discriminate between chaotic and stochastic inputs. Here we extend this work to in vitro experimental s tudies with rat cutaneous mechanoreceptors. For each of the neurons te sted, we show that a chaotically driven ISI series can be distinguishe d from a stochastically driven ISI series on the basis of a nonlinear prediction measure. This work demonstrates that dynamical information can be preserved when an analog chaotic signal is converted into a spi ke train by a sensory neuron.