Modularity, variation and variability in genetic representation

Abstract

Nature uses a complex genotype-phenotype map to advance a relatively simple genotype space variational topology to an extremely complex phenotypic variational topology. This dissertation introduces a modular, fixed non-trivial, multi-layered genotype-phenotype map, incorporating an interpretation of the biological processes of transcription and translation into the representation of a Genetic Algorithm (GA), thereby introducing a more flexible phenotypic structure and increasing connectivity. A series of experiments are conducted to examine the impact of the representation on variation and variability, with results indicating that through the inclusion of the proposed multi-layered mapping, only a small fraction of genotypic mutations are adaptive. Results also suggest that slowing genetic drift has an important part to play in the evolution of a population, allowing the representation to foster the creation of highly connected neutral networks, altering genetic drift and changing the evolutionary trajectory by inducing variability.