EPSRC Network on Evolvability in Biology & Software Systems

Evolvability, Genetics & Development in Natural and Constructed Systems: Abstracts of the EPSRC Evolvability Network Symposium

Tewin Bury Farm Hotel, Hertfordshire, England, UK
26-28 August 2003


University of Hertfordshire Computer Science Technical Report 389
C. L. Nehaniv, P. J. Bentley & S. Kumar (Editors)

Evolvability: Connections with Development, Duplication-Divergence and other

Evolvability: Connections with Development, Duplication-Divergence and other Mechanisms in Natural and Constructed Systems

CHRYSTOPHER L. NEHANIV

Adaptive Systems Reseach Group
Algorithms Research Group
University of Hertfordshire
College Lane
Hatfield Herts AL10 9AB
United Kingdom

C.L.Nehaniv@herts.ac.uk


We survey some properties of evolutionary systems exhibiting open-ended complexity increase, and ask whether these properites are prerequisites for or consequences of evolvability. These include modularity, sensitivity to changing requirements, duplication-divergence phenomena, extradimensional bypass, and robustness to and control of various types of variability. These have for the most part been studied independently by workers in evolutionary computation and biology.

Darwinian evolution characterized by heritable variation and selection is not by itself sufficient to account for the capacity to vary nor for heritable phenotypic expressions of fitness. For instance, a genetic system and also a genotype-phenotype relationship are necessary. Existence of an evolutionary process by itself can thus not fully explain the advant of genetic systems, complex and flexible genotype-phenotype mappings, nor even the heritability of fitness. This presents a challenge both to biologists seeking to understand the capacity of life to evolve and to computer scientists who seek to harness biological-like robustness and openness in the evolution of artificial systems.

Evolvability has been variously defined as the ``ability to produce adaptive variants when acted on by the genetic system" (Wagner & Altenberg, 1996), as the ``capacity to generate heritable phenotypic variation" by developmental biologists (Kirschner & Gerhart, 1998); and is characterized by `evolutionary watersheds' opening the ``floodgates to future evolution", such as the advant of segmentation or body plans (Dawkins, 1987). On the other hand, inappropriately constrained variability can lead to lack of stability, `cancer', nonheritability of fitness, and severe limitations on evolutionary power.

The evolution of biological complexity has often harnessed duplication-and-divergence mechanisms (e.g. Ohno, 1970). Such mechanisms underlie multicellular differentiation, intimately related to developmental processes in organic multicellular entities, and division of labour (in multicellularity but also in populations of social insects).

Genetic regulatory networks (GRNs) have nice dynamical properties and are now being modelled and explored in constructed evolutionary systems. Constructed systems, e.g. based on inheritance of duplicated copies of genetic regulatory networks that acquire different dynamical behaviours in interaction with their environments and with many instances of each other, can be used to harness and study possible evolutionary developmental systems similar to those which have been derived by organic evolution. Their evolvability properties should be systematically characterized. We consider how, in such evolutionary developmental systems, such properties as the following might arise: appropriate modularity, robustness to genetic and phenotypic noise (in inheritance and in development, respectively), redundancy, phenotypic plasticity, balances between freedom at different levels (e.g. Michod & Roze 1999), and suppression of enough undesirable variability while allowing enough useful variability.