Deterministic and stochastic identification of neurophysiological systems

Examples of the use of deterministic and stochastic identification of concrete neurophysiological objects are examined. Deterministic identification was used for the efferent fiber-muscle system. Transition characteristics obtained reflect the dynamic nonlinearity of this system. For stochastic identification, Wiener analysis was generalized for the case of spike trains at the input and output of the system and a statistical criterion was suggested for comparison of the spike trains. By means of the generalization, Wiener identification of a model neuron and a real system of afferent fibers — synapses — neuron was undertaken in the molluscan nervous system. Quantitative estimates were obtained of the relative contribution of nonlinear components in a description of this system. Accuracy of prediction of response of the Wiener model to testing input influences is estimated and compared with the results of testing the real system. The data confirm the efficiency of this method of identification when used to construct functional models of neurophysiological systems.