Summary: The model considered here makes a clear distinction be­
tween the primary locus and the evolvability modifier loci.
In real organisms however, as well as in the most interesting
artificial life systems, genes have emergent functions that
are not pre­labeled as `primary' or `modifier' loci. Simi­
larly, the phenotypes are not pre­labeled as `good' or `bad',
but their effects on the short or long term survival of their
carriers are emergent properties. These survival properties
may not be inferable from a reductionist description of them.
Hence, real organisms and artificial life systems may exhibit
evolutionary pathologies as emergent phenomena that can
only be observed retrospectively.
The suppression of evolvability for evolutionary patholo­
gies becomes a possibility in evolutionary systems that con­
tain structured populations. These abound in biological
systems, and have been investigated in artificial life as well
(D'haeseleer and Bluming (1994), Ray (1995)).
Because of the complexity of organismal phenotypes, the
effect of one trait on another trait's evolvability may not be
obvious, or even in principle predictable. Therefore, evolu­

Source: Altenberg, Lee - Department of Information and Computer Science, University of Hawai'i at Manoa
Nehaniv, Chrystopher - School of Computer Science, University of Hertfordshire

Collections: Computer Technologies and Information Sciences