E.N. Sidnjaeva, N.I. Sidnjaev

It is constructed regressions model in the form of analytical dependence. The main principle of researches consists that any phenomena concerning the higher nervous activity, can be analysed and understood, how some activity in a network consisting of logic elements, accepting only two condition ("all or anything"). Thus for any logic expression satisfying specified above conditions, the network of logic elements having behaviour described by this expression can be found. In article the mathematical model of neural bioexcitability for construction of a function chart artificial neuron is experimentally proved. Analytical dependences of transitive functions show that sigmoidalniay function possesses selective sensitivity to signals of different intensity that corresponds to the biological data. The greatest sensitivity is observed near to a threshold where small changes of a signal net lead to notable changes of an exit. On the contrary, in areas considerably above or below threshold level function is not sensitive to signal variations.