The main objective of this investigation is a study of mechanisms of simulated evolution of primitive agents in simple environment. There is an artificial world inhabited by some population of agents. Agents can receive information from their external and internal environment and perform some actions. Behavior of every agent is governed by modular artificial neural network. Weights of this network are coded in agent's genome. The population evolves through modifications of agents' genomes. As a result of simulated evolution it has been formed a hierarchy of goals in the population of simple software agents. There also has occurred the differentiation in the behavior repertoire of the population.

Mathematical modelling in actual problems of science and technics