Abstract: Information in neural systems is transferred and processed in the form of spike trains, but how a spike train encodes the information it conveys is unknown. The main controversy of the neural coding mechanism is whether information is encoded in the temporal pattern or the rate of the spike train, and how does the random neural spike train express deterministic information. Through a nonlinear neural dynamic model we discover that the spike rate and the randomness of the spiking influence temporal information transfer in a way like stochastic resonance. This predicts an enhancing effect of noise on hearing, which is further confirmed by psychophysical experiments.