Artificial life models of neural development

Artificial neural networks are computational models of nervous systems. Natural organisms, however, do possess not only nervous systems but also genetic information stored in the nucleus of their cells (genotype). The nervous system is part of the phenotype which is derived from the genotype through a process called development. The information specified in the genotype determines aspects of the nervous system which are expressed as innate behavioural tendencies and predispositions to learn. When neural networks are viewed in the broader biological context of artificial life, they tend to be accompanied by genotypes and to become members of evolving populations of networks in which genotypes are inherited from parents to offspring. Artificial neural networks can be evolved by using evolutionary algorithms. An initial population of different artificial genotypes, each encoding the free parameters of an individual neural network (e.g. the connection strengths and/or the architecture of the network and/or the learning rules), are created randomly. Each individual network is evaluated in order to determine its performance in some task (fitness). The fittest networks are allowed to reproduce (sexually or non-sexually) by generating copies of their genotypes with the addition of changes introduced by some genetic operator (e.g. mutations, crossover, duplication). This process is repeated for a number of generations until a network that satisfies the performance criterion (fitness function) set by the experimenter is obtained. © 2003 Elsevier Ltd. All rights reserved.