International Encyclopedia of Systems and Cybernetics

The capacity of a system to select and decide, within limits, its own behavior.

This concept of autonomy was introduced by the French biologist P. VENDRYES (1942). It is quite different from MATURANA and VARELA's concept, introduced about 20 years later, but in no way incompatible. VENDRYES described a hierarchy of various levels of autonomy: tropic (or metabolic), kinetic (motor) and psychic and mental. This corresponds more or less with LABORIT's hierarchic concept of the brain's organization.(1954 and 1976). This concept of autonomy is essentially based on the relative possibility of the system to control its relations with its environment. It also implies a specific concept of time: present time is the connection between an already strictly determined past and a still more or less probabilistic future offering yet possibilities to select between various choices. Such selection converts one – and only one – of these options into reality, suppressing all the others. This concept thus implies a limited probabilistic view of the future, quite consonant with the more recent concept of chaos.

In the same vein, K. BERRIEN observed "1) Choice is the human equivalent of probabilistic outputs in the more general systems theory" and "2) Although some of the determining parameters (can be specified), their precise measurement is subject in principle to an uncertainty of the same general kind as originally indicated by HEISENBERG" (1968, p.97)

VENDRYES theory describes as follows the goals, means and mechanisms that characterize autonomous systems:

- Counter-randomness: The system uses regulating devices to compatibilize random inputs with its internal determinism;

- Internal determinism: The system is able to maintain the fluctuations of its internal processes within stable limits. This connects VENDRYES' views with CANNON's homeostasis and with MATURANA's autopoiesis;

- Interrelation functions: The system possesses mediating functions that coordinates its internal states with the environmental variations;

- Internal medium ("invironment"): This is the internal space of the system, where the interrelations between the subsystems take place. This concept comes from Cl. BERNARD and connects VENDRYES autonomy with J. MILLER's living systems, and again with CANNON's homeostasis, also inspired from Cl. BERNARD.

- Perturbation: Any input not yet controlled by the system. However, many inputs are needed by the system, and their assimilation is an essential aspect of the system’s symbiosis with its environment.

- Random relation: Any not pre-defined interrelation of the system’s inputs with its internal states. VENDRYES explains that the relations between two systems are deterministic when already functionally linked and random when not linked. The system may gain – up to a point and with time – predefined controls on critical inputs. This is another link with autopoiesis.

- Articular relation: A mechanism which permits the selection at any moment of a position, state or motion among various that are possible. The model of articular relation has been widely developed by VENDRYES. (see specific heading)

- Reserve: A potentially usable store of energy or matter within the system, that can be used at any moment by the corresponding control to offset in real time or within a time lag, a specific perturbation. This is a need in order to maintain the system distinct from its environment. Controls are efficient only if associated with specific reserves. (see specific heading)

Another viewpoint is given according to a subtle definition by R.H. HOWE: "Autonomy is the unity of computation and construction" (1975, p.5)

In this case, computation should be understood as that type of computation which occurs in concrete neural networks (i.e. any system with internal communication between numerous elements in an organizational closure way), and construction as self-reproduction.

In the last resort, it must however be emphasized that a system's autonomy depends always of its possibility to obtain sufficient energy inputs from its environment. In A.S. IBERALL words "… an autonomous system is a thermodynamic engine" (1973, p.4)

This basic condition, the existence of an organizationally closed set of rules being also provided for, allows for the autonomous existence of the system.

Still more generally, more autonomy also means more interdependence among more components and subsystems within the system and more extended and complex relations with a more precisely specified environment. Man for example, needs many more specific types of inputs than a tree.