Noise-enhanced temporal association in neural networks

We consider a network of globally coupled neuronal oscillators subject to random force, and investigate numerically dynamic responses to external periodic driving. The order parameter, which measures the overlap between the configuration of the system and embedded patterns, is found to exhibit stochastic resonance behavior, as manifested by the signal-to-noise ratio (SNR). The optimal noise level at which the SNR reaches its maximum is found to depend on the driving frequency. On the other hand, as the randomness in the driving amplitude is increased, the system undergoes a transition from the memory-retrieval state to the mixed-memory one. The noise effects on the temporal-association state in the absence of external periodic driving are also investigated, revealing similar noise-enhanced resonance.