Georg Ivanovas From Autism to Humanism - systems theory in medicine

6. Systemic Medicine

previous -- home -- content -- next

6.5 Central and distributed control

The reason why physiological processes are nonlinear and far from an equilibrium becomes immediately intelligible when the notion of control is introduced. In artificial systems two kinds of control are distinguished, central and distributed control. Systems with a central control are linear, homogeneous and have a modular hierarchy. Systems with distributed control are heterogeneous, nonlinear and self-organizing. The elements are able to find dynamically a new solution on their own (Gershenson 2007: 2).

This shall be demonstrated with the model of traffic lights (Gershenson 2005). Usually traffic lights follow a fixed green wave mode where the lights change according to a given rhythm. This rhythm is either fixed or controlled by a central computer. Its program considers the usual traffic volume. But such a program is not optimal as it does not take the real conditions into account. It is non-adaptive. All such programs fail after a football game or during Christmas rush. Then very often the police regulates the traffic. This is approximately the principle of self-organising traffic lights (SOTL). In this system, every traffic light is equipped with a certain set of rules and knows the traffic density of its own and the surrounding crossroads.

The traffic lights are in so far self-regulating, as they have to find the optimal solution on their own in co-operation with the surrounding traffic lights. No central decision maker exists. This network system performs much better than any fixed green wave method, at least in the computer simulation as it has not been applied yet. The results of the two approaches can be seen in the illustration (Gershenson 2005). (1)

It is evident that the oscillations of the self-organizing traffic lights are about the same as the oscillations in healthy physiological processes. The oscillations of the fixed green method shows a somehow similar rigidity of the pattern as seen in disease. Furthermore, it is evident, either, that physiological processes are self-regulating by nature, although there are influences from the centre, defining a certain goal.

The difference between systems with a central and a distributed control is the difference between trivial and nontrivial machines (chap. 4.5). “A homogeneous system will be easier to understand and control. A heterogeneous system will be able to cope with more complexity with less elements, and will be able to adapt more quickly to sudden changes. If there is a system of ten agents each able to solve ten tasks, a homogeneous system will be able to solve ten tasks robustly (if we do not consider combinations as new tasks). A fully heterogeneous system would be able to solve a hundred tasks, but it would be fragile if one agent failed. Heterogeneity also brings diversity, that can accelerate the speed of exploration, adaptation, and evolution, since different solutions can be sought in parallel. The diversity is also related to the amount of variety of perturbations that the system can cope with” (Gershenson 2007: 52-53).

There is a main difference in how these systems are controlled. Systems with a central control are easy to understand and easy to control. It is only necessary to change the chain of command. The higher in the hierarchy the change is performed the more effective is an intervention.

Self-organizing systems are different. This shall be demonstrated with a simple example. Breast-feeding is a recursive process (chap. 4.2) with a typical self-organisation: The mother provides milk in the quantity and to the time the baby needs it. This involves also anticipation, another trait of self-organisation. Thus, such systems have also to be ‘cognitive’ (Gershenson 2007: 51). Typical strategies of central control in breast-feeding are to be breast-feed every 4 hours. Often the baby is weighed before and after drinking. Then the drunken quantity is compared with a list of reference values established in bottle-fed babies and the ‘missing’ quantity is given by the bottle.

IfIf one regards breast-feeding as self-regulating a different form of control has to be performed. Control would be, for example, to provide the necessary relaxation of mother and child, or to support the mother in crises as in mastitis. Or in days when the demand of the baby suddenly rises it is helpful to breast-feed more often, but shorter, with the aim to instruct the organism of the mother to adapt to the new situation. Some of the techniques to improve the adherence to breast-feeding are laid down in the Ten Steps to Successful Breastfeeding of UNICEF. Mothers applying such principles were 6 times more likely to exclusively breastfeed their babies (Declercq et al. 2009).

In the context of self-organisation the notion of control has a totally different meaning than in central control. In ‘complex adaptive systems’ which are essentially unpredictable and uncontrollable (chap. 2.8). there is no classical control anymore. Control in this context means to influence the system to do what one wants. This is performed by so-called mediators. “A mediator arbitrates among the elements of a system, to minimize conflict, interferences and frictions; and to maximize cooperation and synergy” (Gershenson 2007: 41). A mediator might be traffic lights, a drug, an advice, a glass of Champaign or a bunch of flowers.

In our example of breast-feeding the notions of control differ in many ways. In the self-organizing approach the aim is to promote breast-feeding and to support this process as best as possible with different mediators. In the central control the aim is to bring the child to a certain weight and the mother is a mediator in doing so.


(1) Publication of the illustration by cpurtesy of Carlos Gershenson

previous -- home -- content -- next