One of the primary ideas intrinsic to Wiener's theory of cybernetics is the concept of homeostasis. Cybernetics, borrowing the idea from biology, conceives of any existing system as posessing a fundamental drive to homeostasis - an over-riding instinct toward the maintenance of its survival and the assurance of its continued health and security. It is upon this insight and its application to machines, animals, and man, that much of cybernetics rests.
Cybernetics, as a science of control, is not so much concerned with how we can control a system so much as it is an attempt to discern the fundamental principles which allow us to create a self-correcting and self-controlling system. We are enabled to architect this autonomous system through various means which tend to condition behavior within that system, through incentive and negative reinforcement, because of that fundamental drive for homeostasis.
Cybernetics and Homeostasis
Through this insight about the fundamental principle of an organic system being the drive to homeostasis, it's possible for us to see that an organized system (be it a human being, a society, or a machine) will exhibit deterministic (and thus, potentially predictable and hence controllable) behavior because it wishes to continue to exist as it is. It wishes to be healthy, it wishes to be content... it wishes to live.
It's necessary to note (in an attempt to avoid misunderstanding) that there is nothing "wrong" with this homeostatic drive; it is, in fact, necessary to sustain the life of the organism. Thus with homeostasis we have a sort of necessary evil - without it, we lose the stability necessary for life and sanity, and yet because of it, change is fiercely resisted as a threat to survival.
Wiener actually takes a largely positive view of homeostasis, especially as it relates to the social mechanism. He views communication -- the exchange of information -- as the fundamental agent of negentropy, a transmission of meaning which allows the possibility of control, analysis, and creation - in short, as a bridge between the inner, creative dynamics of will and thought, and the outer world. It is the upsetter of established order and fundamentally opposed to the agency which makes homeostasis the organism's goal.
The Significance of Information
"To live adequately is to live with adequate information"
- Norbert Wiener, The Human Use of Human Beings, p.124
The purely homeostatic organism views any external data it receives, (i.e. that which impinges upon its consciousness) not as an opportunity for expansion, but as an immediate threat to its survival, which must be dealt with as efficiently as possible, so that it can return to its secure and comfortable slumber of contentment.
Homeostasis as it relates to cybernetics is thus a process of self-preservation. The organism makes every attempt to remove or remedy the disturbance in the habit patterns of survival which are the basis of its life caused by the experience of new information. The horse brushes away the annoying fly, the human removes sensory inputs to maintain his contentment or concentration.
Cybernetics and the Self
A homeostatic system, cybernetic theory tells us, is inherently selfish. There is no way to put this into a moral context, since it is simply a fact within the nature of the cybernetic machine, which posesses a paticular teleology proper to it.
An organism, be it a culture, a society, or an individual human being, as noted above, is fundamentally interested in the maintenance of its internal "status quo". Needing to "shut off" the sources of disturbance of its slumber, it evolves a faculty of inhibition to do precisely that. Through this faculty of inhibition, its ability to select appropriate inputs and reject inappropriate ones -- a rudimentary type of intelligence and decision-making -- is developed.
The exhibition of this behavior appears to us to be indistinguishable from intelligent, purposeful behavior. This ability is the basis of feedback, which ultimately also reflects an ability to remember past states and to "learn" from them how to behave, and of selectivity and inhibitive faculties of all kinds.
The introduction of a feedback loop presents the posibility of a self-motivated, or (perhaps more in the spirit of the term Cybernetics which Wiener coined), a self-steering machine, which progressively adapts itself to the inputs with which it is faced, responding in accordance with the "strategies" which have worked for it in the past.
As it continues to exist and to learn, it gets better and faster at responding in predictable ways because it is continually selecting and rejecting inputs, forming in itself patterns of behavior, some of which eventually are so incorporated into the neural network that they become almost automatic responses.
"Fighting" against an entrenched habit-pattern is experienced in the same way whether we are trying to train a neural network or a person. The system resists change, not because it is "evil", unnecessary, or stupid, but simply because it has become accustomed to a particular way of solving a given problem. Resistance is pain and effort which necessitates new learning and implies a risk to survival; contentment is the pleasure and effortlessness of achieved homeostasis.
To the outside observer who is attempting to obtain a specific response from the thought-process of the other being, it is possible to evaluate the fidelity of the output to his ideal, desired result, but it is not possible for him to directly alter the thought-processes and internal logic behind that behavior, only to use force to impose his will upon the bewildered pupil he is training.
Instinct and Intelligence
Homeostasis-seeking describes quite accurately the behavior of the so-called lower animals. Until a fresh jolt of hunger, pain, desire for sex, or other such information disturbs it, it exists contentedly in what we may well call a waking slumber. Intelligence alters the balance somewhat in higher animals, who display the capacity to inhibit the immediacy of their desires in hopes of a later "payoff" at a more opportune time -- the capacity for delayed gratification, and as a later development, of inhibition. The originating drive is the same -- the immediate gain is foresaken in hope of higher future return.
It was originally believed that "reflex" was responsible for animal behavior, some blind governing impulse which responded unconditionally to programmed stimuli. Were this the case, we would not observe animals which appear capable of learning through experience. If anything, it seems that the feedback process itself is the only "hard-wired" reflex activity. Even the nervous system itself, which makes this possible, has been demonstrated to have evolved systematically. The fact that Pavlov was able to condition his dogs to respond to a stimuli not related, except experientially, to the receipt of food should be an indication that it is possible for us to initiate "reflex" responses by means other than their programmed natural triggers, through association with that existing behavioral response. Behavioral conditioning is a theory far more inclusive of various observable animal behavior than older propositions like "instinct".
Considering the applicability of many of these concepts to human behavior, and even their incorporation into managerial science and systems theory through the discipline of cybernetics, it becomes increasingly apparent that there is far more to so-called instinctual behavior than simple response to stimuli. The theories of cybernetics can account for much of human as well as animal behavior, and they effectively give a model which can be, and has been, utilized as a theoretical starting-point for artificial life which could display similar behavioral characteristics.
What I find particularly notable is the idea that a sort of intelligence can arise through the evolution of inhibitory faculties, allowing the entity what amounts to a degree of "free will" in selecting its inputs and outputs, as well as the possibility of overriding its programmed responses and homeostatic condition, in order to expand itself. Certainly, it seems an important step in the development of true artificial intelligence.
Wiener, Norbert, On The Human Use of Human Beings, Riverside Press, Cambridge MA 1950.
Wiener, Norbert, Cybernetics (Control and Communication in the Animal and the Machine), MIT Press, Cambridge MA 1948, 1961.
Young, John Frederick, Cybernetics, Illefe Books Ltd, London 1969. pp.12-13,pp.131-133