A Note on the Evolution of System Theory
The evolution of system theory exhibits three main phases of development. The first phase in the evolution of the theory of systems depends heavily upon ideas developed within physiology. Homeostasis. in particular is the guiding idea: a system is a dynamical whole able to maintain its working conditions.
In order to define a system, one needs (1) components; (2) mutual interactions; (3) the environment in which the system is situated; and (4) a boundary distinguishing the system from its environment.
The main intuition behind this first understanding of dynamic systems is well expressed by the following passage: “The most general and fundamental property of a system is the interdependence of parts or variables. Interdependence consists in the existence of determinate relationships among the parts or variables as contrasted with randomness of variability. In other words, interdependence is order in the relationship among the components which enter into a system. This order must have a tendency to self-maintenance, which is very generally expressed in the concept of equilibrium. It need not, however, be a static self-maintenance or a stable equilibrium. It may be an ordered process of change, a process following a determinate pattern rather than random variability relative to the starting point. This is called a moving equilibrium and is well exemplified by growth” (T. Parsons, The Social System, New York, Free Press, 1951, p. 107).
The main result achieved by the first phase of development of system theory has been the proof that the system as a whole is defined by properties not pertaining to any of its parts. Global equilibrium, say, is a property of the whole system, not of its parts. However, the definition of a system as the whole resulting from the interactions among its components contains a number of hidden assumptions. Subsequent developments of system theory have sought to address and understand these hidden assumptions. There follow the main assumptions hidden within the initial definition of a system:
- The first assumption is that all the system’s components are given in advance, before its constitution. The underlining idea is that the system collects and organizes elements that are already there. We shall discuss this problem under the heading of the system’s constitution.
- The second assumption places all changes on the side of the environment. What about systems able to learn and to develop new strategies with which to cope better with survival or other problems they may encounter? Systems endowed with this property will be called adaptive.
- The third assumption becomes explicit as soon as the problem of the historical continuity of the system through time is addressed. What happens when a new component enters the system or is generated internally? What happens when a component is no longer part of the system, or dies out? These groups of questions can be summarized as the problem of the reproduction of the system.
The overall outcome of constitution, adaptation and reproduction is complexity, albeit this of a type rather different from any of the mainstream conceptualizations of complexity.
The first two assumptions have been able to produce an extensive body of literature, whose main results can be summarized by distinguishing two different types of both constitution and adaptation.
The two forms of constitution are the bottom-up type of constitution from components of the system (that are already available), and the top-down constitution from (a previous stage of) the system into its components. This latter form of constitution again assumes two guises: as constraints on initial conditions and the phase space of the system components, and as the development of a new organizational layer(s) of the system.
In their turn, organizational layers are a structural condition needed by developing adaptive systems. In fact, an adaptive system needs both (1) rules governing the system’s interactions with its environment and with other systems; and (2) a higher-order layer that can change such rules of interaction. These changes may be purely random, or may follow pre-established, or acquired patterns. In this regard, a hypothesis can be advanced which claims that the main difference between non-living natural systems on the one hand, and living natural systems, psychological systems and social systems on the other, is that the former have only one single organizational layer of interactions; the latter, more complex, systems have at least two layers of organization: the one governing interactions and the one capable of modifying the rules of interaction.
Furthermore, the persistence over time of living systems is made possible by multi-stability, a form of dynamic stability in the face of perturbations that prevents the destabilization and rapid disappearance of such systems.
The third of the three above-mentioned assumptions is the most important one. Indeed, the unfolding of the third hidden assumption has dramatically modified the entire landscape of system theory. The theory of autopoietic systems is possibly the best-known result connected with the problem of systems reproduction. In this regard, it is worth considering that the theory of autopoietic systems is itself in need of further generalizations. The simplest generalization of these is well represented by Niklas Luhmann’s theory of social systems. The second possibility is well represented by Robert Rosen, who some twenty years before the birth of the theory of autopoietic systems proposed what he called (M,R)-systems, later developed into the theory of anticipatory systems. As a result, Rosen’s theory is both more general and more precise than the theory of autopoietic systems. Forthcoming posts shall slowly present the main aspects of Rosen’s proposal.