CCE Theses and Dissertations
The Effects of Learning on Evolvability and its Evolution
Date of Award
2005
Document Type
Dissertation
Degree Name
Doctor of Philosophy (PhD)
Department
Graduate School of Computer and Information Sciences
Advisor
James D. Cannady
Committee Member
Sumitra Mukherjee
Committee Member
Gregory Simco
Abstract
Altenberg has defined evolvability as the "ability of the genetic operator/representation scheme to produce offspring that are fitter than their parents." Consequently, the evolution of evolvability describes evolutionary changes in the evolvability of a population over time. Based on these definitions, this dissertation makes the following argument that individual learning ability will affect both the evolvability of a population and also the evolution of evolvability within that population. Individual learning ability alters the genetic makeup of a population over time via the process known as the Baldwin Effect. The genetic makeup of a population over time constitutes what is known as the population's genetic diversity dynamics. Finally, both evolvability and the evolution of evolvability have been linked to a population's diversity dynamics. Thus, it is quite reasonable to expect that individual learning will impact both a population's evolvability and the evolution of that evolvability.
In support of the above argument, experiments have been performed to demonstrate that individual learning affects both evolvability and the evolution of evolvability. Results from these experiments have revealed three novel effects of individual learning on evolvability and its evolution. First, individual learning stabilizes a population's genetic diversity dynamics, and consequently also its evolvability, in the face of changes in environmental complexity. Second, when both the genetic operator and the genetic representation are able to evolve, the adaptive power of the representation affects their evolution and thus the evolution of evolvability. Third, the adaptive power of the representation also affects the rate at which evolvability evolves, accelerating it when the adaptive power is low and retarding it when the adaptive power is high. Collectively, the results of these three experiments provide evidence firmly establishing that learning does affect evolvability and also the evolution of evolvability.
NSUWorks Citation
Grant William Braught. 2005. The Effects of Learning on Evolvability and its Evolution. Doctoral dissertation. Nova Southeastern University. Retrieved from NSUWorks, Graduate School of Computer and Information Sciences. (422)
https://nsuworks.nova.edu/gscis_etd/422.