Dynamic encoding of amplitude-modulated sounds at the level of auditory nerve fibers.

The relationship between the mean spike frequency of auditory nerve fibers and the levels of tonal stimuli with frequencies equal to the characteristic frequency can be presented as the "input-output" characteristic. In real auditory nerve fibers, the slope of this characteristic increases and its width decreases with increases in the level of spontaneous activity or the ability of fibers to generate spikes in the absence of a stimulus. However, real fibers with low spontaneous activity reproduce the amplitude modulation of sinusoidally amplitude-modulated signal significantly better than fibers with high levels of spontaneous activity. The simulation experiment reported here shows that the reason for the good reproduction of amplitude modulation in auditory nerve fibers is not the static profiles of the "input-output" characteristic but the dynamic properties of fibers which support the tuning (adaptation) of the threshold of the fiber to the level of the stimulus being applied. Because of their steep "input-output" characteristic, auditory nerve fibers with high levels of spontaneous activity can reproduce the modulation of sounds at subthreshold levels when a weak noise is added to the signal, thus demonstrating the property of stochastic resonance.