Stochastic Resonance in interacting flows of cars and neural spikes.

The problem of interaction between cars at road junctions has been studied for a long time. It is well known that the interaction generally decreases the traffic efficiency and that randomness in the car flow can help cross-junction traffic efficiency. We show that proper long-range correlations in the flow noise, related to 1/f like noises, provide superior traffic properties as compared to Poissonian or periodic car traffic. Moreover, a stochastic resonance phenomenon sensitive for the shape of the spectrum occurs. By a small modification, the model has been made relevant for neuronal spike traffic. Neural spike trains generated by Gaussian 1/f noise show superior spike traffic properties and efficiency of information transfer. © 2000 American Institute of Physics. [ABSTRACT FROM AUTHOR]