3.7 The Architecture of Functional Systems

25

does it really create two equal offspring, simultaneously annihilating itself? Does it

essentially bud off excreta in a less vital, perhaps almost moribund version of the

parent, which thereby gains a new lease of life? Does it gather its vital forces and

concentrate them in a fresh new organism, accepting senescence and death for itself?

The Rôle of Memory

The picturesque idea of human (and, as far as we know, other animals) memory as

a vast warehouse of facts to be retrieved at will, closely analogous to the digital

memories of modern computers, would appear to be very far from the truth. Man, in

particular, appears to possess immense power of bringing past experience (including

that of fellow members of the species, via written or other records) to bear on the

present situation. In terms of the schemata of Figs. 3.1 and 3.2, this input should be

included in the regulatory response upper RR.

3.7

The Architecture of Functional Systems

Almost any system is confronted with the problem that as its complexity increases,

more and more channels of communication are required (cf. Sect. 3.5), with greater

and greater information capacity, if every component of the system is to remain fully

integrated. A useful way of coping with this problem is to organize systems hierar-

chically, such that the amount of information is distributed more or less uniformly

across levels; by this means the information flow within and between levels remains

manageable. One way of quantifying the degree of hierarchicality is to determine

the distribution of path lengths between pairs of components; the closer it is to a

power law distribution, the more hierarchical the system (cf. Chap. 12).

As the size of a system (as measured by the number of constituent components)

increases, if every component had to be individually designed and fabricated, the bur-

den of doing so would soon become overwhelming. In artificial systems, such as very

large-scale integrated circuits, this problem is evaded by a combination of functional

modularity and structural regularity. The latter is anything that reduces complexity,

in the sense discussed in Sect. 11.5 (e.g., the repetition of components). Thus, even

the most sophisticated integrated circuits have essentially only two types of basic

components, pMOS (p-type metal–oxide–semiconductor field-effect transistors) and

nMOS (their n-type equivalents).

Functional modularity is the structural localization of function. ¹ In other words,

some function is separated into structural units (“modules”); these are able to carry out

some information processing internally, which diminishes the amount of information

that needs to flow between modules (cf. the rôle of nervous centres, Sect. 3.5 and

Chap. 24). It may even arise that design principles developed for modules at one

level in a hierarchy can be reused for modules at other levels. Functional modularity

1 See Lipson (2007).