The Cost of Modularity 271
(Kornberg 1965; Owen, Kalhan, and Hanson 2002). Moreover, the cofactors that are
involved in the cycle occur in many other metabolic pathways as well. In contrast to such
a plethora of external interactions, only two reactions integrate each intermediate into the
cycle. So the external interactions of the cycle seem to be stronger than the internal ones,
which is the opposite of what is required for structural modularity.^21 In biological systems,
other than in engineered ones, functional and structural modules are not usually
congruent.
15.7 A Lesson from the Function-Structure Map
In the case of technical artifacts, the congruence of F-modules and S-modules was seen
to originate from design methodology. Unsurprisingly such congruence is not generally
found in biological systems, since evolution cannot follow any methodology. Nevertheless,
in many developmental and evolutionary modules, a unique attribution of function(s) to
a structural module is in fact possible (Wagner, Mezey, and Calabretta 2005). In such cases
biological F- and S-modules coincide. In developmental modules something similar is
found to what is in the subjects of more classical physiological disciplines, where func-
tional descriptions often map fairly reliably on structural descriptions (Krohs 2004). An
evolutionary and developmental module—for example, one that gives rise to a signaling
pathway—forms an S-module that is simultaneously an F-module. A characteristic of such
modules is that they contribute, identically or with slight variations, to different capacities
of an organism. It is precisely this multiple involvement that is regarded as one of the
main benefi ts of biological modularity (Schlosser and Wagner 2004; Callebaut and Rasskin-
Gutman 2005). But, in light of the result of the discussion here, the congruence of F- and
S-modules needs an evolutionary explanation.
Functionality within networks depends on the properties of the network components
and on the structure of the network. According to the explication of type-fi xed structures
given in section 15.6, any change of an F-module is the result of a change of the structure
of the system. So the explanation of any congruence of F- and S-modules will be equiva-
lent to an explanation of a change of the structure of a system until congruence of F- and
S-modules is achieved. One may be tempted to conclude that this evolutionary change of
structure creates evolutionary costs. In the case of technical artifacts such costs do arise
during the design process and therefore contribute to the price of the product. They are
outweighed by lower costs of construction, diagnosis, and maintenance. However, evolu-
tionary “costs” are not comparable with costs of the four kinds that I have identifi ed here.
Their relevance differs between the biological and the technological case. In contrast to
the costs of design processes in engineering, evolutionary “costs” do not infl uence the
energy requirement, or any other measure for the costs of the ontogenesis or the life span
processes of an organism. And evolutionary processes as such do not cause costs because