370 Michael Wheeler
latter makes sense is a significant step towards establishing that the former does
too.
Among other things, coding talk about genes is supposed to help us make good
on the claim that genes are special developmental factors, that they count as being
privilegedcausal elements in the developmental process. The way that coding talk
is supposed to achieve this feat is nicely captured by Lorenz’s [1965] image of the
non-genetic causal factors in development as nothing more than the building blocks
out of which organisms are systematically constructed according to a blueprint
stored in the genes. On such a view, the real challenge for developmental biology
is to understand how genetically specified instructions organise those available
developmental materials into an organism. This way of looking at things really
does make genes special.
There are many gene concepts in the literature, ranging from the essentially
abstract, generically Mendelian notion of a gene as a trait difference marker to
various attempts to give molecular substance to the idea.^1 There are even some
who argue that most of the assumptions that historically have underpinned the
term ‘gene’ have been shown to be problematic, meaning that the very concept of a
gene is now, in many ways, a misleading one that perhaps biology could do without
(e.g., [Dupre, 2005]). For the purposes of the present investigation I intend to put
both definitional diversity and strategic critical eliminativism aside, and stipulate
that we should be thinking of a gene as an entity with some sort of molecular unity,
that is, as a stretch of DNA that possesses some sort of ontological integrity. To
make this idea firm enough for the job at hand, we need to resist the tempting
thought that the way to establish the molecular unity in question is by holding
that genes are those parts of the genome that code.^2 Why this is should be clear
enough: I have been assuming that there is conceptual space for the following
result: there are genes but they don’t code for anything. If genes simply are the
coding parts of the genome, then this result is not available. A negative answer
to the question ‘do genes code?’ would imply that there are no genes. So we need
to achieve the desired molecular unity without appealing to coding. But how?
One answer would be to appeal to causally underpinned structural isomorphisms
that exist between (a) sequences of DNA and (b) certain developmental elements
that are causally downstream of those sequences. The most likely candidates
for the latter are proteins, since the claim that systematic causal mappings exist
between sequences of DNA and amino acids in proteins is not generally thought
to be controversial (see the description of protein synthesis in section 4 below). Of
course, if (i) all there is to coding is some sort of systematic causal dependence,
and (ii) genes may rightly be said to code for proteins, whether or not they also
code for traits, then the recognition of systematic causal mappings between (a)
and (b) would herald significant progress in our investigation. Genes would code
(^1) For a recent review, see Griffiths and Stotz forthcoming. See also theRepresenting Genes
project athttp://www.pitt.edu/∼kstotz/genes/genes.html.
(^2) Contrary to some accounts, an organism’s genome is not simply its complete set of genes,
but much more besides (see [Dupre, 2005]).