Statistical Performance Analysis And Array Of Neural Cell Model Of Stochastic Resonance

In the present thesis, we investigate the efficiency of information transmission in neuron model and the positive roles of internal and external noise during the information transmission process. Firstly, we summary the development of stochastic resonance theory in the neuroscience, and determine to mainly study the noise in the behavior of neuron colonies. In the bistable nonlinear array analogous to neuron models, for a common periodic noisy signal, the array response shows a resonance-typed behavior as the internal noise intensity increases. Moreover, as the array size increases, the signal-to-noise ratio (SNR) gain can be enlarged and exceed unity in certain regions of internal array noise intensity. It is demonstrated that the performance of the infinite array can be closely approached by an array of two subsystems by the cyclestationary stochastic process theory and numeral simulations. This result is of significance in the array sesign based on noise-enhanced effects.Furthermore, we apply the above mentioned array stochastic resonance theory in the information transmission of FitzHugh-Nagumo collective neurons. In the real information transmission system, the input is often an aperiodic Gaussian signal and the internal noise plays an important role in the information transmission. When external noise is fixed, the increase of internal noise leads to array stochastic resonance. It indicates that neuron has self-adptive ability. In the same way, when internal noise is fixed, the correlation coefficient ratio also produces the same phenomena. It shows how the external environment affects the neuron. In other words, the neuron might select an optimal noisy environment. The efficiency of information transmission can be improved by these kinds of noise and correlation coefficient ratio does exceed unity in the situation under study. This phenomenon does not change when the DC-adaptive ability occurs in neurons regardless of subthreshold signal or suprathreshold signal. All of those suggest that noise, as an active factor of affecting neuron function, is worth studying. This mechanic of information processing has directive sense for both neuron colonies and complex nervous system such as the small-world network.