Development: Three Grades of Ontogenetic Involvement 191doesn’t make DST’s approach to adaptation non-sub-organismal. This is a phe-
nomenon that replicator biology is fully able to accommodate. The concept of
the extended phenotype was introduced in acknowledgement of the fact that an
organism’s traits extend beyond the skin. The concept of the extended phenotype
contends that replicators contain information for building features of an organism’s
environment (e.g. birds’ nests, beehives and beaver dams) [Sterelny, 2001]. The
recognition that inheritable phenotypes can exist beyond thee skin is not sufficient
to make for a non-sub-organismal biology.
DST is sub-organismal with respect to its approach to adaptive evolution. It
falls within the ambit of the DST approach to explain the adaptedness of organ-
isms. And just as in Grade I involvement, the process by which organisms become
adapted is one of selection, recombination and reassortment among a population
of units of inheritance. The adaptedness of organisms is the consequence of the
generation, recombination, reassortment, and selection, of these entities. These se-
lecta are not organisms, but developmental systems, aggregates of which together
constitute an organism.^12 As in Grade I, Grade II adaptation occurs through the
accretion within organisms of adaptive (selected) entities. Selection alone is what
makes adaptive evolution adaptive. So adaptedness of organisms is explained by
the aggregation of sub-organismal selected entities.
The adequacy of selection alone as a cause of the adaptedness of organisms has
recently been called into question. Not just any lineage of reproducing entities can
increase in adaptiveness under a process of mutation and selection. This has been
one of the central findings of research into genetic algorithms. A genetic algo-
rithm must have certain intrinsic, structural properties if it is to undergo adaptive
evolution [Wagner and Altenberg, 1996]. One of the projects of genetic algorithm
research is to address the question, what sorts of properties must programmes
have in order that they may undergoadaptiveevolution. Biologists have recently
adapted that question to the context of evolutionary biology under the rubric of
‘evolvability’. Evolvability is defined as “...an organism’s capacity to generate
heritable phenotypic variation” [Kirschner and Gerhart, 1998]. The question is
‘What properties must organisms have such that lineages of them may undergo
adaptive evolution?’^13 Schwenk and Wagner outline the basic requirements.
On the one hand, phenotypes must be mutable and therefore responsive
to the constantly changing demands of the environment.... On the
other hand, phenotypes must be stable so that the complex dynamics
of their developmental and functional systems are not disrupted.... It
is this fundamental tension — between mutability and stability — that
current evolutionary biology seeks to explain. [2003, 390]But how are stability and mutability achieved? Recent developmental biology
suggests that stability and mutability are both consequences of a fundamental,
(^12) At least aggregates of theirphenotypic effectsconstitute an organism.
(^13) Kirschner and Gerhart [1998], Von Dassow and Munro [1999], Wagner and Altenberg
[1996].indexWagner, G.