[Deterministic and stochastic identification of neurophysiologic systems]. / Deterministicheskaia i stokhasticheskaia identifikatsiia neirofiziologicheskikh sistem.

The paper deals with deterministic and stochastic identification methods applied to the concrete neurophysiological systems. The deterministic identification was carried out for the system: efferent fibres-muscle. The obtained transition characteristics demonstrated dynamic nonlinearity of the system. Identification of the neuronal model and the "afferent fibres-synapses-neuron" system in mollusc Planorbis corneus was carried out using the stochastic methods. For these purpose the Wiener method of stochastic identification was expanded for the case of pulse trains as input and output signals. The weight of the nonlinear component in the Wiener model and accuracy of the model prediction were quantitatively estimated. The results obtained proves the possibility of using these identification methods for various neurophysiological systems.

[Wiener analysis of transmission of signals of identified mollusk neurons]. / Vinerovskii analiz peredachi signalov identifitsirovannymi neironami molliuskov.

Mathematical model of spike train transmission by identified neurons of the mollusc Planorbis corneus was obtained by means of Wiener analysis. Poisson distributed sequences of near-threshold stimuli were used as input trains for construction and testing. It was shown that addition of terms sensitive to the nonlinear properties of the synapse-neuron system to the model decreased the model error from 25% to 16%. When comparing the mean square error value of the predicted model responses to testing stimuli and the real neuron responses to the same stimuli, the prediction error with allowance for nonlinear properties was established not to exceed 21%. Some questions connected with choosing adequate criteria for comparing model and experimental results are discussed.