International Encyclopedia of Systems and Cybernetics

Since long time, and still on-going, there is a controversy between the proponents of hierarchic or parallel models or organization, in general.

A specific case, i.e. "Cognitive systems architectures" has been considered by Marta OLIVETTI BELARDINELLI (1998, p. 11-26). She advocates for an "integration of perspectives". Historically the hierarchic view has been dominant, at least in western thinking. This was probably a result of the search for short -cuts in the management of complex situations of all kinds, as the hierarchic ordering leads easily to unconscious or conscious algorithms of treatment ("Act according to the rules and so, avoid problems")

It should however be observed that:

1- Any hierarchical order has been constructed in a historical dimension

2- The hierarchical view, when strongly established, tends frequently to block needed adaptation

In turn the parallel way implies the possibility of unlimited adaptability. In OLIVETTI BELARDINELLI's words: "There maybe some justification for claiming the superiority of hierarchical structuring, which relies on input-specific systems with differentiated processing, interconnected according to a structure compatible with the classical architecture of the nervous system. However, the limitations imposed by this type of structure on the explanation of adaptation processes seem far more serious than those characteristic of connectionism, that is, regarding the optimization of processes as complexity increases"(p. 17)

In short, hierarchic structures and models could represent the more or less periodic summing-up of the stabilized results of adaptation processes through parallel activity.

→ Bottom up; Parallel distributed processing; Top down