The Cost of Modularity 273
unlikely to occur because of fi tness-decreasing “side effects.” Without such a supplementa-
tion, the parcellation path to a modular structure cannot be regarded as being satisfactorily
explained either.
Another issue is identifi ed that demands an explanation in terms of evolution: the some-
times occurring congruence of functional and structural modules in metabolic and in many
gene regulatory networks. Such congruence is a precondition for the multiple use of a
functional module in an organism. It is shown that this congruence, though almost always
present in technical artifacts due to their design methodology, is not trivially present in
biological systems but only brought about by evolutionary processes. Here as well, a gap
is diagnosed in the biological arguments about the evolution of modular systems. This
latter gap is not to be closed by cost considerations. It poses, in contrast, epistemic costs:
Where congruence is found, it needs to be explained. And where it is absent, an account
of the system needs to be developed that is more differentiated than those currently
available.
Notes
- Simon consistently takes the analytical perspective and leaves open “whether we are able to understand the
world because it is hierarchic [i.e., modular] or whether it appears hierarchic because those aspects of it which
are not elude our understanding and observation.” He gives reasons for supposing “that the former is at least
half the truth—that evolving complexity would tend to be hierarchic—but it may not be the whole truth” (Simon
1969: 208). - I focus on biological modularity and its relation to modularity in engineering, and do not discuss the somewhat
different and still unsettled issue of the modularity of mind. For a discussion of the latter issue see, e.g., Callebaut
and Rasskin-Gutman (2005), García (2007), and Sarnecki (2007). - A recently proposed explication of two different concepts of functional modularity (García 2007) combines
functional and structural criteria in each of the considered cases, functional integration and functional indepen-
dence of modules. In each of the two cases this results in counting only those subsystems as modules that satisfy
simultaneously the modularity conditions of functional and structural approaches. I discuss modules of this kind
in section 15.7 but stay, when arguing about functional models, with the classical concept of functional decom-
position, which largely disregards criteria of network structure. - I further explain this in section 15.6. In contrast to the biological case, the coincidence of functional and
structural modules usually holds with respect to technical artifacts, as also discussed in section 15.6. - It was clear from the very beginning of the modularity debate (e.g., Simon 1969) that any analysis of a system
in terms of modular components distorts the picture of the network (Krohs and Callebaut 2007). - Evolutionary modules, however, are sometimes even defi ned as subsystems that are both functional and
developmental units (Brandon 2005; see also Schlosser 2005). - “[C]omplex systems will evolve from simple systems more rapidly if there are stable intermediate forms than
if there are not” (Simon 1969: 196). - Similarly, it is doubtful whether a “gene for high mutability” could be selected in diploid organisms under
usual selective regimes (Wagner, Mezey, and Calabretta 2005). - Fell (2007) supports the validity of this fi nding by comparison with similar results obtained from evolutionary
approaches to engineering electric circuits by means of genetic algorithms (Koza et al. 1999; Bennett et al.
2000). - I do not claim that the authors of these arguments are not aware of such adaptive advantage of integration.
The arguments, however, do not cover this issue.