An attempt is made to design a model of intracellular chemical processes of excitability regulation based on the prediction of probable after-effects of a signal and the estimation of its biological value. While constructing the model, the author tried to take into account as much data relating to actual chemical processes in the neurons as possible. Modeling results show that recognition, event prediction, decision-making modulation of excitable membrane properties, and registration of the action results may be implemented without difficulty by a relatively plain system of chemical reactions. Reactions of this kind probably take place inside the neurons. The rate of decision-making may be significantly high depending on whether the reactions proceed concurrently. The model resembles a neural network, but it functions within a neuron. The role of physical links between neurons is played by the substances specific to a pair of synapses. The use of the model instead of a formal neuron in the development of neural network would result in a sharp change in their properties: for example, the retrieval ability of the neural network is maximized by choosing an optimal threshold of neurons.<>