Living Control Systems II

Contents

Foreword, W. Thomas Bourbon

ix

Foreword

In 1979, Bill Powers wrote a prophecy:  “A scientific revolution is just around the corner, and anyone with a personal computer can participate in it....  [T]he particular subject matter is human nature and in a broader scope, the nature of all living systems.  Some ancient and thoroughly accepted principles are going to be overturned, and the whole direction of scientific investigation of life processes will change.” (William T. Powers, “The Nature of Robots: Part 1:  Defining Behavior,” BYTE  4(6), June, p. 132) Powers foresaw the overthrow of the idea that either stimuli from the environment, or commands from the mind or brain, are sole causes of behavior.  In its place, he offered the concept that people (and in their own ways all other organisms) intend that they will experience certain perceptions and behave to cause the perceptions they intend.  The social, behavioral, and life sciences had simply missed the fact that living things control many features of their environments.  Powers acknowledged that fact, and he realized that to an organism the environment exists only as perceptions, hence his insight that organisms act to control their own perceptions.  His formal statement of the new concept was control theory, and he said amateurs, working with personal computers on their tables at home, would be major players in the revolution.  Thirteen years later, the revolution is not accomplished, but it is underway.

Powers’ perceptual control theory is new, but he is not the first to describe many of the key ideas in the theory.  Over 2200 years ago, Aristotle wrote about intention—”that for the sake of which,” the desire or wish that causes actions that result

<snip>

W. Thomas Bourbon
Nacodoches, Texas
February 1992

Read more (pdf)

Preface

This is a volume of explorations, all unpublished and some, quite possibly, unpublishable in the ordinary course of things. A good many of them, such as the paper on evolution, are the result of following my nose, seeing where control theory will lead if given its head. Others have resulted from people asking me questions I couldn't answer, enticing me to think about subjects—like epistemology—that I wouldn't have thought interesting without some nudging.

Many of these papers were written as a form of thinking out loud, trying to put some order into disordered thoughts by arranging them as if for publication (with no intention of seeing them in print). In some cases, as for the paper on transfer functions (somewhat shortened because the data were taken four computers ago and are no longer at hand), the ideas were set down as a promise to myself; a promise to continue the work and get it into some finished form before offering it for publication. A few of those promises may yet be kept; most, I hope, will be kept by others.
Anyone with the will to do so is welcome to finish any of these unfinished thoughts if they look interesting. At this stage of my life, that would be a much more satisfactory outcome than doing it myself; the point of all this is to see others grasping and using control theory in the study of living systems.

If it hadn't been for Greg Williams, most of this material would have sat yellowing in my files or awaiting a cosmic ray to delete it from old computer disks. Greg went through mountains of my old papers and numerous disks simply because he thought it ought to be done before, in my irresponsible way, I got rid of them in a fit of neatness. He has judged this collection to be worth preserving; I can't tell if this is true, because my mind is on what to write next, not those old words which have solidified into print. But I trust Greg Williams' judgment on scientific matters better than I trust that of almost anyone else, and certainly more than my own.

I must also acknowledge the material contributions to the conception and production of this volume by Edward Ford, at this writing the president of the Control Systems Group. Ed is a down-to-earth practical person, who deals daily with real human problems. For some reason he gives boundless encouragement to my theorizing, even though he then has to ask me what it means so he can translate it into something usable.

My thanks to Greg and Ed, good friends, good critics, and persistent benefactors.

William T. Powers
Durango, Colorado
February 1992

A Note on the Text

As in the previous volume of Living Control Systems, I have minimized alterations to the original manuscripts, attempting mainly to provide consistency of spelling and grammar within each paper. All deviations from the originals, for better or for worse, are solely my responsibility.

In coordinating this project, I have been aided by several individuals, including my wife Pat, who typed several of the papers and wrote the software used to redraw all of the figures; Dag Forssell, who provided—and typed—the elusive complete manuscript of "CT Psychology and Social Organizations" (which he had originally obtained from Rick Marken) and also provided the frontispiece holograph; Clark McPhail, who provided the manuscript of "Control Theory for Sociology"; and Mackenzie Yearsley, who provided the manuscript of "A Bucket of Beans." I thank all of them.

Ed Ford got this endeavor started, then made sure it was completed. Many, many thanks, Ed, for your ongoing support!

Most of all, I thank Bill and Mary Powers, who graciously permitted me to stay at their home, rummage through their files, photocopy whatever I wanted, and even cart precious manuscripts around the country. For the record: Bill and Mary are simply great folks!

Gregory Williams
Gravel Switch, Kentucky
March 1992

[1959]
Some Implications of Feedback Theory Concerning Behavior

In the course of developing a general feedback theory of human behavior, we have come across an interpretation of goal-directed behavior which seems to offer a departure from some concepts of stimulus and response currently employed by many experimentalists and theorists in psychology.

This interpretation is based on three facts about feedback control-systems: (1) Feedback systems are always goal-seeking; (2) feedback systems determine their own input-signals, and not their output-efforts; and (3) the output-efforts of feedback control-systems have the "purpose" of preventing environmental disturbances from affecting the input-signals.

Let us first show that these are valid statements about how all stable feedback control-systems operate, and then show how they might influence current interpretations of behavior if the systems involved should prove to have the properties of feedback control-systems.

Behavioral properties of control-systems

Diagram One is a very general representation of any feedback control-system, showing all the variables essential to such a system and all the functions performed on these variables within the system and its environment. Any feedback control-system can be represented in this form.

The oval labelled E represents the environment of the system. This environment consists of environmental variables in

Unpublished manuscript

[1971]
The Wiener Feedback Model—A Strategic Error

The The way in which feedback was presented in Cybernetics has kept psychologists from seeing how to apply feedback concepts to behavior. In effect, the concept of feedback was presented in such a way that it could be assimilated into the stimulus-response concept of behavioral organization: the idea of successive-state analysis, basic to Wiener's "stationary time series," may be responsible for erroneous concepts of how feedback operates, and his basic feedback diagram, often reproduced, is inappropriate to most feedback situations. As a result, the fact that there are feedback loops from output to input via the environment has been ignored, or else dealt with in inappropriate terms.

Successive-state analysis

In the 1964 edition of a basic psychology text, D. O. Hebb explains feedback as a succession of stimuli and responses, each response modifying the next stimulus for the next response. (Hebb, 1964)
This analysis uses the concept of successive states of a system-environment relationship. There are difficulties with this approach even in neural-net models, since the nervous system must be assumed to be "clocked," which it is not; in dealing with overall behavior the difficulties are worse because neither stimuli nor responses are discrete point-events, and they do not neatly alternate in time.

Unpublished manuscript

[1972]
Emotion

There are three major categories of sensations. One category is associated with what we learn to identify as events that take place outside the body; sensations of light, of smell, of sound, of temperature can arise or change without our taking any action at all. In control system terms, these are the quantities that seem especially susceptible to disturbance by outside agencies. Then there is a set of sensations that seem to be under more or less direct voluntary control: sensations of effort, of joint angle, of facial expression, of touch, of movement—these sensations hardly ever appear unless one wishes them to appear, and external events have very little power to affect them in comparison with our own control over them. The third category is odd in that it seems to have its location entirely inside the body, and yet at times seems to be under control by the external world, or by factors in ourselves over which we have no voluntary control at all: these are feelings or emotions. This chapter represents an attempt to bring the third category of experience into the scope of this model.
Emotions are well-known to have two components, one thought-like and one sensation-like. Physiologists have long identified the sensation-like component as arising from changes in the biochemical state of the body-alterations in the levels of action of the major organ systems. In fact for all strong emotions, there appears to be a more of less common physiological state, termed the "general adaptation syndrome." The adrenal cortex releases adrenalin into the blood‑

Chapter 26 of the preliminary version of Behavior: The Control of Perception, used for teaching a course at Northwestern University.

[1972]
An Experiment with Levels

A truism accepted by practically every working therapist is that progress in therapy depends on the patient's becoming aware of the true nature of his problem. This observation, indeed, was one of the reasons for which I suspected that awareness is associated intimately with the functions of the organizing system. If awareness is directed to the functioning of some subsystem in the hierarchy, this notion implies that the subsystem is being modified, or can be modified if there is intrinsic error. The requirements for successful therapy are quite clear if one accepts these ideas: awareness must be directed to the systems that are in conflict, while they are in operation, and there must be intrinsic error. The awareness alone is not sufficient: there must also be what is termed "affect"—an experiencing of the emotional consequences of the conflict. Since one does not seek help in order to become free of pleasant emotions, the consequence is that one must experience unpleasantness in order to make any important change in his own organization. It is not enough to achieve "insight" into oneself; the conflicted systems must be brought into operation, both at the same time, and the consequences of the conflict must be experienced as a disturbance of intrinsic state while awareness is directed to the conflicted systems. In terms of having more subjective meaning, one must consciously experience both or all of the desires that are in fact causing the conflict, so that the emotional consequences become plain in relationship to the action of seeking both conflicted goals.

Chapter 27 of the preliminary version of Behavior: The Control of Perception, used for teaching a course at Northwestern University.

[ca. 1974]
The Illusion of Control

Any reader of Harvey Wheeler's compendium, Beyond the Punitive Society, must, if at all analytically inclined, wish he could add a chapter of his own. I will try to avoid the temptation, even though the liveliness and variety of the contributed chapters makes the temptation strong. Throughout this volume the discussions revolve around critiques, modifications, and applications of B. F. Skinner's provocative views but this is not a book review, either. Let us say that Wheeler's volume has inspired and focussed my own analytic response to Skinner's work, by bringing together all the most cogent arguments concerning it. My contribution here will not be a 20th chapter, but a meta-chapter derived from a point of view not completely shared by any of the contributors. An uninvited meta-chapter, I hasten to add.
The title of this article suggests, I hope, that there is one serious question that has not been raised by Skinner, by his critics, or by his proponents either whole- or half-hearted. That is the question of just what we mean by "control of behavior," and whether what we have in mind by this concept is possible, much less practical or desirable. Nearly everyone assumes that it is possible, and bases his arguments on that axiom. My purpose here is to cast doubt on that axiom.

What is behavior?

Unpublished manuscript

[1979]
Control Theory for Sociology

As a theoretician and an interloper in this volume, I will not attempt here to translate from control theory into Meadian sociology. That task is better left to the real sociologists who, seeing that such translations were possible, invited me to provide some background at a 1978 meeting at Kent State University. Having benefited from discussions at that meeting, I will not follow the format I brought to Kent State, but will try to speak directly to what now seem the most germane points.

The basic issue involving control theory is that of the nature of sociological principles and laws. Control theory is above all a theory of individual organization; it purports to describe how any one person is put together. If it has anything to say to sociologists, its contribution will be in the form of suggestions about the kinds of interactions that are to be expected when living systems organized as control systems encounter each other. I understand that this was one of Mead's concepts as well: that social laws are emergent, and arise from individual characteristics.

In the following I will first talk about control theory and its relationship to other theories, and only afterward explain what it is. Understanding what it is not is of considerable help in interpreting its meaning, and avoiding the trap of reinterpreting it in terms of something more familiar, but quite different.

Written for an unpublished compendium of papers given at a conference on George Herbert Mead at Kent State University, Kent, Ohio, 1978

[1980]
CT Psychology and Social Organizations

This paper was written as theoretical background for D. T. Campbell's investigations of the uses of social indicators in judging social programs. The orientation is that of "CT psychology," CT standing for control theory; familiarity with the basics of this approach is assumed, and the theoretical underpinnings of CT psychology will be used here without justification.

The following consists of a systematic development of a number of themes without regard to their obviousness (or for that matter, their factual correctness). The object is to see what kinds of theoretical statements applying to social interactions appear to follow from a belief that all organisms, human beings especially, are organized as hierarchies of negative feedback control systems. In lieu of a review of the principles of CT psychology, we will begin with some statements that follow deductively from these principles and will be important to later arguments.

Some generalizations from the principles of CT psychology

Human beings are assumed to have certain fundamental properties:

1. The world with which human beings knowingly interact consists of a hierarchy of representations in the human mind (for which read "brain" or any other acceptable substitute). These representations consist of information-carrying signals

Working paper, March 1980, distributed by W. T. Powers and D. T. Campbell. Supported in part by NSF Grant BNS 7925577.

[1983]
A Bucket of Beans

Applications of Control Theory
to the Understanding of Human Nature
and the Design of Social Systems

At this meeting you will quite probably hear proposals for solving the difficulties that face our society and our world. These proposals will have to do with applying cybernetic ideas to the design of large systems; industrial systems, economic systems, and political systems. Some will just be high-tech proposals, but others will discuss how people ought to interact with each other and organize their society, whatever machines they employ in the process. If past meetings of our cybernetics society are any guide, you will hear a great deal about how a system ought to work, but little about the people who are supposed to make it work and who have to be persuaded to live under it.

This is a general shortcoming of what is called social systems thinking. One famous example of a social system design is expressed in the principle, "From each according to his abilities, to each according to his needs." The unspoken assumption about human nature on which this design depends is that human beings are willing to give all they are able to give, and will demand only what they do in fact need. When human nature proves to be different from what was assumed, the tendency seems to be to fall back on force as a means of making the theoretical system seem to work in practice. Another example, just to show I am not taking sides, is the principle of competition in a free marketplace. Theoretically, competition is supposed to drive quality up and prices down. In practice, what it produces just as often is either market

Tutorial presented at the American Society for Cybernetics Meeting, October 7,1983, at Palo Alto, California.

[1983]
Deriving Closed-Loop Transfer Functions
for a Behavioral Model, and Vice Versa

Transfer functions have been used in representations of human behavioral organization since World War IL In general, a transfer function is a concise mathematical statement of the way the action or output of a behaving system depends on a stimulus or input to that system. Given a valid transfer function, one can predict the output of a system for any input, static or varying, provided the external setting remains the same. The purpose of this paper is to show both how a transfer function can be derived without the usual reliance on an analytical model, and how the method of derivation itself came from a relatively new approach to modelling behavior: a closed-loop, rather than open-loop, approach. The method and the approach may have applications in fields other than behavioral modelling: weather prediction, for example, or possibly economic modelling. The basic method is not new, but perhaps my simple-minded way of arriving at it is.

Transfer functions and mediation theory

There is a family resemblance between a transfer function and other kinds of principles that have been used to express theories of behavior. In nearly all experimental studies of behavior, there is one underlying model, based on what can be called mediation theory. Under mediation theory, what an organism does is strictly a function of what happens to it. The

Presented at the American Society for Cybernetics Meeting, October 1983, at Palo Alto, California.

[1983]
Learning and Evolution

In the present context, "learning" has a special restricted meaning: the process by which an organism acquires control of what happens to it. Much of what is studied under the label of learning should, I think, be thought of as performance. The capacity to reproduce past experiences is a natural feature of all control systems; studying that capacity is a way of characterizing the abilities of those systems, but does not explain their development. The principal requirement we must impose on any learning theory that is to accompany the present model is that it must be able to explain how new control systems come into being—especially the first new one.

Control by bias of random output

The basic problem that any theory of learning must solve is what Petrie (1981) refers to (in the context of applying control theory to education) as the "Meno Paradox." Plato had one of "Socrates' yes-men" (Petrie's phrase) wonder how we can enquire about any new subject: if it is new, we do not know what to enquire about, and if we do know what to enquire about, it is not new. To paraphrase: if an organism can neither sense nor control some new variable, it has no means of altering its behavior in the ways required to acquire control. If it can alter its behavior appropriately for bringing the new variable under control, it already has the required control system, or most of it. How could a really new control system come into

Excerpt from unpublished manuscript "CT Psychology—A New View of Organisms."

[1986]
An Agenda for the Control Theory Group

As I was looking forward to this second annual meeting of the Control Theory Group, I found myself getting strangely depressed. It took a while to figure out why. Our numbers have grown—at least the mailing list has grown—to somewhere between sixty and seventy, and there are indications that it will grow much more in the year to come. The thought occurred to me that we're approaching a time when we will no longer have to concentrate on overcoming opposition. I wondered if I might have lost the knack for doing anything but showing what's wrong with conventional ideas and writing introductions to control theory. That unpleasant idea led to wondering just what I would do if someone handed me a blank check, and said, "Go to it." And that wondering has led to this talk.

It's strange how difficult it is to find applications for the first new conception of behavior in over three centuries. We've all had this trouble, but those of us who have had the most trouble are those who know the most about the life sciences. I think that's a clue. Think of clinical psychologists (we have a few of them here). What could clinical psychologists do with control theory? Well, they could play how-and-why, they could try to find out what the client is controlling for. But how would they use these techniques? Probably to try to help people get over anxieties, phobias, compulsions, depression, or stress. And that's where I think the difficulty starts.

No matter what discipline in the life sciences you are at‑

Distributed at the Control Systems Group Meeting, August 20-24, 1986, at Kenosha, Wisconsin.

[1987]
Control Theory, Constructivism, and Autopoiesis

The objective of this essay is not to achieve a final synthesis of the three concepts in the title, but to explore similarities, differences, possibilities, and alternatives within the consensual domain these words suggest. My biases as a control theorist with a favorite axe to grind will probably be visible just as one can see biases in other approaches, reflecting their authors' primary orientations. My conclusion—for the benefit of those who seek certainty in science and thus can be excused early from this discussion—is that human consciousness and the world in which it lives are likely to remain a mystery even after we have applied our best intellectual tools to them. The most we can hope for while exploring waters this deep and murky is to find a few hints as to directions we can take.

One effect that is hoped for from this essay is a clarification of the relationship between the approach taken by Humberto Maturana and my own considerably simpler approach to an understanding of human nature. The term "simpler" is not intended to be approving: it can also be taken to mean "less advanced" if one is so inclined. The article "What Is It to See?" (Cybernetic, Fall-Winter 1985) is my working source for reference here to Maturana's ideas, in addition to understandings gleaned elsewhere, I hope correctly. Readers who are coming into the middle of this conversation can find my view, a little out of date, in Behavior: The Control of Perception (Powers, 1973).

Distributed at the American Society for Cybernetics European Conference, March 15-19,1987, at St. Gallen, Switzerland.

[1987]
Control Theory: The Road to Utopia

When I first heard that this was to be yet another theme meeting I raised some churlish objections, muttering that people go to scientific meetings to report on their own work, not somebody else's. Naturally, my mind then insisted on returning to the topic of utopias and I finally realized that I no longer could think of anything else to talk about. The truth is that any theoretician who dwells on human behavior thinks that if everyone just believed in the right theory (guess whose), the result would be a vast improvement in the human condition. At least I don't think you will hear many theoreticians telling you that the road to utopia is the one we're on now.

Since the subject has been so conveniently raised, I would like to tell you why I think that control theory could take us a long way toward a more ideal society. But before I do that, there's a problem that has to be dealt with in this meeting of cyberneticists. The problem is that most of you don't know what I am talking about when I say the words "control theory."

Or perhaps I should say that many of you think you do know what control theory is about, but have something in mind that is very different from the conception I'm involved with. The biggest problem is this word control.

A few meetings ago I was asked by a rather bristling young woman why I thought that controlling people was a good thing to do, or words to that effect. I was too dumbfounded to give a coherent answer, and even if I hadn't been so shocked,

Distributed at the American Society for Cybernetics Meeting, December 1987, at Urbana, Illinois.

[1987]
On Violence and Aggression

A group Statement on Violence reprinted in the June 1987 issue of this Newsletter presents a case against the idea that violence is a scientifically-established property of human beings. Manson and Wrangham, in partial rebuttal, remind us of the evidence that at least some aggression may have genetic bases. Hinde, one of the original authors of the Statement, rejects the idea that behavior can be explained only in terms of self-interest (as suggested by Manson and Wrangham), and recommends, in the interest of world peace, more research. I recommend, in addition to both points of view, a new theory.

I would like to introduce the control-system theorist's view —an outsider's view without a doubt to this discussion, by proposing some concepts that take neither side, but suggest some new possibilities. If Frans Plooij's interpretations of my theoretical notions are any indication, ethologists should have no difficulty in seeing the points I am making and improving on them.

Suppose, for the sake of argument and to bypass rhetoric, that the basic principle of living systems is found in their ability to achieve control of what happens to them. The lowliest organisms, single-celled creatures, do this modestly well, and adult human beings do it superbly (if not always successfully). The most successful organisms control the physical variables that affect them in the most immediate and important ways.
They do this in the manner of control systems: systems that

Unpublished comment on articles in the Newsletter of the Society for Human Ethology, Volume 5, Number 2, June 1987.

[1988]
The Good, the True, and the Real

Cybernetics is suffering from a case of overapplied epistemology. To show what the problem is I will indulge in some exaggerations, but they do not go beyond the uttermost boundaries of actuality.

Two primary epistemologists of cybernetics, von Glasersfeld and Maturana, have informed us that experienced reality is constructed (Maturana adding, "... in an informationally closed nervous system"). I have added my share to the confusion. The result has been an enthusiastic reception by those who find the restrictions of mathematics, logic, and scientific experiment distasteful. These enthusiasts have taken the conjectures of the philosophers and made them into a new—but not really so new—legislated non reality for cybernetics, a sort of inverted reality.

In this new reality, the old reality appears as a cartoon strip in which white-coated ex-nerds, now grown to terrifying stature, conspire to convert us into obedient cyborgs chained in front of television sets while our exposed brains are basted with the poisons of technology. Art, music, literature, and Bambi are compressed by a gigantic technological foot into data banks as the System fills our vacant minds with pi calculated to an infinite number of decimal places.

Against this threat, the New Epistemologists stand arrayed in aggressive permissiveness. The rules of this permissiveness are very strict, and often entail inexpertly-hidden political agendas. No longer may we say that some ideas are right and

Unpublished manuscript

[1988]
A Manifesto for Control Theorists

I. Control theory was first offered as a model of behavior because artificial control systems seemed to act in some ways like organisms.

A. In the form of regulating devices they could maintain variables like temperature, pressure, chemical concentration and composition, flow, position, sound level, and almost anything else measurable in stable predetermined conditions, despite unpredictable disturbances.

B. In aircraft, autopilots could keep airplanes flying level and on course at constant altitude, despite changes in weight distribution, wind direction, engine efficiency, turbulence, and air density, automatically—meaning without outside direction of the details of their actions.

C. In homing devices they showed characteristics of goal-directed or purposive behavior, directing themselves toward distant moving or even evasive targets.

D. In general, they showed the properties that William James called characteristic of life: the persistent and repeated reaching of constant ends by variable means. Control systems behaved in ways that the traditional life sciences had proclaimed impossible for any entity in a world governed by physical laws. Yet they behaved like organisms.

Distributed at the Control Systems Group Meeting, September 28-October 2, 1988, at Kenosha, Wisconsin.

[1989]
After Galileo

My subject for this year concerns programs, principles, and system concepts. I will talk about programs of action that the CSG might want to pursue in the next ten years, the principles we might want to use to organize and justify the programs, and a system concept that might give us a reason for continuing to exist. I won't take these subjects in order, because as you all know that isn't how levels of organization work: they interact with each other and with the real world, and they reorganize in the process. Consider these ideas to be work in progress. You might even want to take notes when you hear something that suggests a better idea to you.

I look on the work of the CSG as being divided into two main parts: theory and practice. Of course our theoreticians are strongly inclined to experimentation, and our practitioners are hardly silent on the subject of theory, so this division is only functional, a matter of which hat you're wearing. I think we would agree that it is a theory that informs our practices, and our practices that give us observations that our theories must take into account, so everything is working as it should.

Theory and practice could keep one satisfied for a lifetime, but I am not satisfied. I want something more. I want what we do to lead to a revolution in the life sciences, a revolution that will change the way all of them go about understanding the nature of living systems. To be blunt, I want a real science of life to develop to replace what now exists, which is not science. I will have to amplify on that a little.

Distributed at the Control Systems Group Meeting, September 20-24, 1989, at Kenosha, Wisconsin.

[Ca 1989]
The Epistemology of the Control System Model of Human Behavior

In the paper in which this model of human behavior was introduced (Powers, Clark, & McFarland, 1960), the following statement appears:

... in all the feedback systems we will discuss, it is of no concern at all to the feedback system what actual effects are produced in the environment. The system reacts only to the signals injected into it by its feedback function, and for any one system nothing else exists. Even when we speak of systems which deal in human interrelationships, these complex systems not only do not "care" about what is actually going on in the "real" environment, they cannot even know what is going on "out there." They perform the sole function of bringing their feedback signals, the only reality they can perceive, to some reference-level, the only goal they know. If we were discussing servo-mechanisms, such anthropomorphisms would be unnecessary, but when we are talking of the very systems in which we live, now and always, which we must employ even to think, anthropomorphism is an essential ingredient of understanding. (pp. 78-79)

In 1973, I published a book called Behavior: The Control of Perception (Powers, 1973). Reading "control system" for "feedback system," "perceptions" for "feedback signals," and "perceptual function" for "feedback function," the later usages,

Unpublished manuscript

[1989]
The Fallacy of Behaviorism

In the "experimental analysis of behavior" an attempt is made to adhere to a strictly empirical point of view. Some behaviorists even refer to this as the "scientific" point of view. No conjectures are made about the internal organization of a behaving system; all conclusions and all models are constructed only from observable phenomena. It is apparently thought that by working within such strict guidelines, one can achieve in the life sciences the same kind of rigor and objectivity that is found in the physical sciences.

That myth was invented by J. B. Watson, the founder of behaviorism (although he probably got its basis from biologists). While Watson is now considered old-fashioned even by behaviorists, the basic principles he expounded are still defended. Watson proposed that by varying the conditions surrounding any organism and observing subsequent changes in its behavior, he would be able to amass enough data to describe how a given organism will respond under any conditions. It was not necessary, he claimed, to invoke such mysterious concepts as consciousness, goals, motivations, emotions, or any other such thing that is known only subjectively. It was not even necessary to know how an organism works in order to observe the laws that govern its behavior.

I call that set of assertions a myth because in fact behaviorists (including, in his time, Watson) do make assumptions about how organisms work. If these assumptions had been made explicitly and competently and had been tested in the

Unpublished manuscript

[1989]
Human Nature from the Standpoint of Control Theory

The most common conception of human nature is based on empiricism: watching how people behave, remembering and classifying everything, and sequencing the classes. To understand human nature then means being able to say "Under circumstances A, people in category B will react to event C by producing behavior D." This entails finding ways of determining

- that circumstances A do in fact exist.

- that a person is in fact in category B.

- that event C has in fact occurred.

- that behavior D does then in fact occur.

Each part of such a statement of human nature carries an uncertainty. Also, one or more of the parts may actually entail multiple uncertainties. For instance, people are characterized not just as belonging to one category, but as belonging to many at once: [female][heads of households][with no more than a 7th-grade education][with more than three children] [receiving ADC payments][between 1/1/85 and 12/31/: ][in the state of New York]. Seven independent conditions must be established as existing in order to say that a person is in category B. If each condition has a probability of 0.97 of actually

Unpublished manuscript

[1990]
Standing at the Crossroads

I'd like to try today to give you the sense that psychology is standing at a crossroads—and not only psychology, but all the sciences of life. We are about to experience the advent of something for which many people have searched, an organizing scheme that pulls together all the disparate schools of thought, specializations, movements, and evanescent fads that make up various fragmented branches of the life sciences.

The organizing scheme is called "control theory." This theory explains a phenomenon, as theories are supposed to do. The phenomenon in question is called control. Everyone has heard this word, and most people have occasion to use it from time to time, but in science it has become part of the meta-language rather than designating a subject of study. A scientist does a control experiment, or demonstrates how manipulation of stimuli and rewards can control an animal's movements, or advocates a proper diet to control cholesterol level, or competes for control of a department. This word is used as part of a background of ordinary language, but it has not been part of the technical language of the life sciences.

The reason is quite simple: nobody in or out of science understood the process of control until about the beginning of World War 2. By understanding the process, I mean being able to define it, characterize it, measure its parameters, predict how it will proceed, and recognize it in a real system. This doesn't mean that control was impossible to accomplish before World War 2: after all, most people accomplish digestion with‑

Distributed at the Control Systems Group Meeting, August 15-19, 1990, at Indiana, Pennsylvania.

[1990]
The X-Phenomenon

Many people in cybernetics and other fields are talking about control and control theory. In most cases, the principle apparently used to explain control is that of compensation, following Ashby and other notable figures. Others seem to think that control is a matter of prediction, analytic or stochastic. Still others appear to believe that control can be achieved by a system that analyzes its environment and computes the output that will have the desired effect. Some have believed that control can be achieved by (somehow) matching a measure of information in a system's output to a measure of information in disturbances. A few have assumed that control is a process that takes place entirely inside an organism, modifying what is otherwise an input-output conversion process.
In the face of all these explanations of control, I propose to abdicate temporarily as an authority on whatever the phenomenon is to which they refer and to apply for a position as the discoverer of a new phenomenon. I will call it the X-phenomenon; we can defer proposals about a better term to substitute for X.

The X-phenomenon is created by the interaction between an environment and a type of system I will call an X-system. The nature of an X-system is recursively defined by the X-phenomenon, which can be observed both in the laboratory and in natural systems. Only living organisms naturally exhibit the X-phenomenon. I will describe a canonical experiment that can distinguish between an X-system and other organizations.

Distributed at the Control Systems Group Meeting, August 15-19, 1990, at Indiana, Pennsylvania.

Published Works by William T. Powers on Living Control Systems, 1989-1991

1989a "Volition: A semi-scientific essay," in Wayne A. Hershberger, ed., Volitional Action: Conation and Control, North-Holland, Amsterdam, et al., 21-37.

1989b "Quantitative measurement of volition: A pilot study," in Wayne A. Hershberger, ed., Volitional Action: Conation and Control, North-Holland, Amsterdam, et al., 315-332.

1989c (with Richard S. Marken) "Levels of intention in behavior," in Wayne A. Hershberger, ed., Volitional Action: Conation and Control, North-Holland, Amsterdam, et al., 409-430.

1989d "Problems with the Smiths," Continuing the Conversation: A Newsletter of Ideas in Cybernetics (16), Spring, 3-4.

1989e "By any other name," Continuing the Conversation: A Newsletter of Ideas in Cybernetics (16), Spring, 11-12.

1989f "Things I'd like to say if they wouldn't think I'm a nut: Or, overgeneralizations that aren't that far over," The Control Systems Group, Inc. Newsletter, May, 5.

1989g "Foreword," in Edward E. Ford, Freedom from Stress, Brandt Publishing, Scottsdale, Arizona, ix-xi.

Prepared by Gregory Williams