Novel class of neural stochastic resonance and error-free information transfer

We investigate a novel class of neural stochastic resonance (SR) exhibiting error-free information transfer. Unlike conventional neural SR, where the decrease of a system's response with too much noise is associated with an increase in the baseline firing rate, here the bell-shaped SR behavior of the input-output cross correlation emerges versus increasing input noise in spite of no significant increase of the baseline firing rate. The neuron thus acts as an error-free detector for weak signals. An integrate-and-fire model with short-term synaptic depression convincingly validates our experimental findings for SR in the human tactile blink reflex.