[Date Prev][Date Next][Thread Prev][Thread Next]   [Date Index] [Thread Index] [Author Index

The other passage, as promised... (Part One)



I finally have some time to type in the other part of my father's discussion about anticipatory vs reactive control mechanisms in modeling natural systems:
 
From  Page 9 of Anticipatory Systems:
 
"I was astonished to find this profusion of anticipatory behavior at all levels of biological organization. It is important here to understand why I found the situation astonishing, for it bears on the developments to be reported subsequently, and raises some crucial epistemological issues.
 
We have already seen, in the few examples presented above, that an anticipatory behavior is one in which a change of state in the present occurs as a function of some predicted future state, and that the agency through which the prediction is made must be, in the broadest sense, a model. I have also indicated that obvious examples of anticipatory behavior abound in the biosphere at all levels of organization, and that much (if not most) conscious human behavior is also of this character. It is further true that organic behaviors at all of these levels have been the subject of incessant scrutiny and theoretical attention for a long time. It might then be expected that such behavior would be well understood, and that there would indeed by an extensive body of theory and of practical experience which could be immediately applied to the problems of forecasting and policy-making which dominated the Center's (Center for the Study of Democratic Institutions) interests. But in fact, nothing could be further from the truth. The surprise was not primarily that there was no such body of theory and experience, but rather that almost no systematic efforts had been made in these directions; and moreover, almost no one recognized that such an effort was urgently required. In retrospect, the most surprising thing to me was that I myself had not previously recognized such a need, despite my overt concerns with modeling as a fundamental scientific activity, and despite my explicit involvement with biological behavior extending over many years. Indeed, I might never have recognized this need, had it not been for the fortuitous chain of circumstances I have described above, which led me to think seriously about apparently alien problems of policy-making in a democratic society. Such are the powers of compartmentalization in the human mind.
 
In fact, the actual situation is somewhat worse than this. At its deepest level, the failure to recognize and understand the nature of anticipatory behavior has not simply been an oversight, but is the necessary consequence of the entire thrust of theoretical science since earliest times. For the basic cornerstone on which our entire scientific enterprise rests is the belief that events are not arbitrary, but obey definite laws which can be discovered. The search for such laws is an _expression_ of our faith in causality. [Note from Judith: This was something my father believed also, it's just that he believed the natural laws were not adequately figured out by science, as yet.] Above all, the development of theoretical physics, from Newton and Maxwell through the present, represents simultaneously the deepest _expression_ and the most persuasive vindication of this faith. Even in quantum mechanics, where the discovery to the Uncertainty Principle of Heisenberg precipitated a deep re-appraisal of causality, there is no abandonment of the notion that microphysical events obey definite laws; the only real novelty is that the quantum laws describe the statistics of classes of events rather than individual elements of such classes.
 
The temporal laws of physics all take the form of differential equations, in which the rate of change of a physical quantity at any instant is expressed as a definite function of the values of other physical quantities at that instant. Thus, from a knowledge of the values of all the relevant quantities at some initial the values of those quantities at the succeeding instant are determined. By iterating this process through an integration operation, the values of these quantities, and hence the entire behavior of the system under consideration, may be determined for all time. Carrying this picture to its logical conclusion, Laplace could say "An intelligence knowing, at a given instant of time, all forces acting in nature, as well as the momentary position of all things of which the universe consists, would be able to comprehend the motions of the largest bodies of the world as well as the lightest atoms in one single formula... To him nothing would be uncertain; both past and future would be present in his eyes."
 
This picture of causality and law, arising initially in physics, has been repeatedly generalized, modified, and extended over the years, but the basic pattern remains identifiable throughout. And one fundamental feature of this picture has remained entirely intact; indeed itself elevated to the status of a natural law. That feature is the following; in any law governing a natural system, it is forbidden to allow present change of state to depend upon future states. Past states perhaps, in systems with "memory"; present state certainly; but never future states. It is perfectly clear from the above discussion why such a commandment is natural, and why its violation would appear tantamount to a denial of causality in the natural world.
 
A denial of causality thus appears as an attack on the ultimate basis on which science itself rests. This is also the reason why arguments from final cause have been excluded from science. In the Aristotelian explanation of system behavior in terms of final causes is the province of teleology. As we shall see abundantly, the concept of an anticipatory system has nothing much to do with teleology. Nevertheless, the imperative to avoid even the remotest appearance of telic explanation in science is so strong that all modes of system analysis conventionally exclude the possibility of anticipatory behavior from the very outset.
 
And yet, let us consider the behavior of a system which contains a predictive model, and which can utilize the predictions of its model to modify its present behavior. Let us suppose further that the model is a "good" model; that its predictions approximate future events with a sufficiently high degree of accuracy. It is clear that such a system will behave as if it were a true anticipatory system; i.e. a system in which present change of state does depend on future states. In the deepest sense, it is evident that this kind of system will not in fact violate our notions of causality in any way, nor need it involve any kind of teleology. But since we explicitly forbid present change of state to depend on future states, we will be driven to understand the behavior of such a system in a purely reactive mode; i.e. one in which present change of state depends only on present and past states.
 
This is indeed what has happened in attempting to come to grips theoretically and practically with biological behavior. Without exception (in my experience), all models and theories of biological systems are reactive in the above sense. As such we have seen that they necessarily exclude all possibility of dealing directly with the properties of anticipatory behavior of the type we have been discussing.
 
How is it, then, that the ubiquity of anticipatory behaviors in biology could have been overlooked for so long? Should it not have been evident that the "reactive paradigm", as we many call it, was grossly deficient in dealing with systems of this kind? To this question there are two answers. The first is that many scientists and philosophers have indeed repeatedly suggested that something fundamental may be missing if we adopt a purely reactive paradigm for consideration of biological phenomena. Unfortunately, these authors have generally been able only imperfectly to articulate their perception, couching it in terms as "will", "Geist," "elan," "entelechy," and others. This has made it easy to dismiss them as mystical, vitalistic, anthropomorphic, idealistic, or with similar unsavory epithets, and to confound them with teleology.
 
The other answer lies in the fact that the reactive paradigm is universal, in the following important sense: Given any mode of system behavior which can be described sufficiently accurately, regardless of the manner in which it is generated, there is a purely reactive system which exhibits precisely this behavior. In other words, any system behavior can be simulated by a purely reactive system. It thus might appear that this universality makes the reactive paradigm completely adequate for all scientific explanations, but this does not follow, and in fact is not the case. For instance, the Ptolemaic epicycles that are also universal, in the sense that any planetary trajectory can be represented in terms of a sufficiently extensive family of them. The reason that the Copernican scheme was considered superior to the Ptolemaic lies not in the existence of trajectories which cannot be represented by the epicycles, but arises entirely from considerations of parsimony, as embodied for instance in Occam's Razor. The universality of the epicycles is regarded as an extraneous mathematical artifact irrelevant to the underlying physical situation, and it is for this reason that a representation of trajectories in terms of them can only be regarded as a simulation, and not as an explanation."
 
I'm going to take a break and type the next part in later this morning-- it turns out to be a bit longer than I thought.Time for coffee...
 
Judith