268 Ulrich Krohs
made clear from a modular stereo set. 2) More housings are needed, one for each module.
This causes additional tara costs. 3) Each module also requires its own power supply pack
and so forth. This causes what can be called autonomy costs. Equivalents to costs of both
kinds do occur in the biological case. 2) Organs, being macroscopic morphological
modules, are wrapped into fascia. And gene clusters, forming the core of many develop-
mental modules, have their regulatory sequences and additional structural markers as a
kind of packaging. 3) Autonomy of organs requires individual vascular and nervous con-
nections. On the molecular level, regulatory cassettes of transcriptional regulators may
serve as an example. Such cassettes constitute a network containing several coactivator
genes, and only these seem to mediate the autonomy of the cassette as a regulatory module
(Kardon, Heanue, and Tabin 2004). It follows that in the biological case there are tara and
autonomy costs as well.
4) In case of a failure of a modular technical artifact, the defect is localized only by
checking whole modules; instead of repairing modules on the level of their components,
they are simply exchanged when defective (White 1999: 475). The failure must not be
localized within a module. This “diagnostic opaqueness” of modules in technology leads,
on the one hand, to decreased diagnostic and exchange costs in terms of hours of work,
and, on the other hand, to an increased requirement of material for repair and maintenance,
since whole modules are discarded instead of only single defective parts. In biological
systems, an equivalent can be found wherever whole, morphologically distinct modules
are discarded in development and self-reproduction, especially where whole cells are
sacrifi ced, for example, during the renewal of epithelia, and in many cases of apoptosis.
So there are material and energetic costs for module-wise replacement.
To supplement adaptive arguments about modularization, costs of the identifi ed kinds
have to be taken into account. It of course requires empirical data to further specify these
costs. It should be noted that with respect to biological organisms only phenomena that
decrease fi tness on a level relevant for selection may count as costs in arguments about
evolutionary adaptation. This includes increases of the energy requirement of processes
going on in the organism, but excludes “costs” of evolutionary processes themselves. This
is in contrast to the case of technical artifacts where costs of the designing process also
need to be taken into account (see section 15.7).
15.6 Mapping Functions on Modular Structures
Up to this point, only the evolution of structural modules has been discussed, since only
modules of this kind are usually regarded as relevant for the organization of a network or
other hierarchical system. However, as presented in section 15.2, another way of decom-
posing biological systems relies on functional criteria. The following two fi ndings require
a closer look at the functional view of biological networks, since they may either be rele-