Coordination of Subsystems

Functional coordination of subsystems is important for all organic structures, certainly the cell, the brain and technical implements, but likewise society, companies, states and ecosystems. All subsystems are limited in scope, perspective and information. They need to cooperate in a given situation to form an integrated functional whole, taking into account, and making mutually consistent, all relevant information. In the brain this leads to the conscious state ( von der Malsburg, 1997; von der Malsburg, 2002; Baars, 1994). The point of functional coordination is to let all subsystems serve the same goal at the same time, directly or indirectly, and be in tune with the environment.

Subsystems have internal parameters that regulate transition between dormancy and activity, the relationship to other subsystems, parameters that keep track of case distinctions and of levels of uncertainty or ambiguity, and that decide on the level of confidence with which they can influence cooperative decisions (Triesch, von der Malsburg, 1999). Subsystems have to decide autonomously whether their application is warranted in a given situation, by recognizing relevant patterns in the current state of other systems and the environment. Pairs of subsystems must autonomously find out whether to pay attention to each other and how to interpret the signals that are exchanged, starting with general statistical observations and possibly ending in the erection of structured interfaces.

Dynamics must inherently thrive to optimize processes in the sense of reduction of malfunction, especially reducing internal and external contradiction and frustration. At each node where signals meet, individual pathways must be optimized to predict the others as best they can. It is an important scientific goal to find a quantitative measure that is to be optimized by system operation on all time scales. The basic means to approach desirable states is redundancy. The decisive measure is the level of agreement reached by subsystems at common target nodes. Globally consistent states are reached by self-organizaton. To speak in terms of a methaphor, let molecules of a liquid be viewed as subsystems, their internal parameters being position and orientation, their interactions having the form of mutual forces. For each pair of molecules, mutual forces leave a lot of uncertainty as to relative position, but in the process of crystallization all these vague interactions intermesh in cooperative ways to create a globally ordered, solid crystal. In an analogous fashion, organic systems reach global certainty as a consistent meshwork of many subsystems, although individually these may have started out with large uncertainty.

While systems evolve and develop they generate new subsystems, often by differentiation from less specific predecessors. Criteria for the generation and differentiation of subsystems may be avoidance of duplication of effort and achievement of division of labor, and at the same time, utility for other subsystems. Examples for the process are the differentiation of visual feature detectors or the evolution or co-option of genes.