Revise Description of Phenomena Category

I recommend that the description of this “Phenomena” category be revised to make it clear that the basic phenomenon explained by PCT is CONTROL. The current description reads as follows:

Behavior is the basic phenomenon explained by PCT. PCT recognizes that intentional behavior is the control of perception. Topics in this category describe examples of behavior. Discussion may include speculation about the kinds of perceptions controlled and proposals how to test those speculations.

I recomend revising it to read like this:

PCT has its origins in William T. Powers’ realization that intentional behavior is a process of CONTROL. Control is an objective phenomenon that involves acting to keep variable aspects of the world in predetermined or reference states, protected from the effects of disturbances. A more detailed, graphical explanation of the phenomenon of control – in this case, as it is done by humans – is given by Rick Marken in a talk entilted PCT in a Teacup; a PowerPoint version is available here and an MP3 video version is available here. This Phenomena category should include descriptions of different examples of control as well as suggestions regarding the type of models that can explain this controlling.

I recommend that the original description be left as is. Reference to the work of any individual other than W.T.Powers is, in my opinion, improper in a category descriptive header. Also, the suggested revision omits the important sentence: “Discussion may include speculation about the kinds of perceptions controlled and proposals how to test those speculations.”

How about this:

Behavior is the basic phenomenon explained by PCT. PCT is based on recognition that behavior is a process of CONTROL . Control is an objective phenomenon that involves acting to keep variable aspects of the world in predetermined or reference states, protected from the effects of disturbances. PCT explains this process as the control of perception. Topics in this category describe examples of the controlling done by living systems. Discussion may include speculation about the kinds of perceptions being controlled and proposals for how to test those speculations.

You have to ask others than me. Personally I prefer the original version. Others may disagree with either or both of us.

As much as I love democracy, I think that in this case it’s not a matter of what the majority likes and dislikes; it’s a matter of what’s right and wrong. And the current description of this Phenomena category is misleading at best and wrong at worst. The current description starts like this:

Behavior is the basic phenomenon explained by PCT. PCT recognizes that intentional behavior is the control of perception.

Behavior is, indeed, the basic phenomenon to be explained by PCT. But what is behavior? The current description says that it is the “control of perception”. But “control of perception” is not a phenomenon; it is a theory that explains a phenomenon; the phenomenon of control. My change in the description makes this clear:

Behavior is the basic phenomenon explained by PCT. PCT is based on recognition that behavior is a process of CONTROL . Control is an objective phenomenon that involves acting to keep variable aspects of the world in predetermined or reference states, protected from the effects of disturbances. PCT explains this process as the control of perception.

I think it is very important to make this point in the description of the Phenomena category so that discussions in that category center around observation rather than just theory. Talking about theory without connecting it to the phenomena it explains – the existence of variables (CVs) being kept in reference states by the disturbance-resisting actions of the system – can be reserved for the Philosophy section of the Fundamentals category.

Here’s a proposed rewrite of the category intro:


This category is for discussion of the observable phenomena of behavior. PCT explains purposeful behavior as control: acting to keep variable aspects of the environment in states which the subject organism prefers. Observable phenomena of behavior include:

  1. The controlled variable (CV).
  2. Its reference value.
  3. The subject’s means of perceiving it.
  4. The subject’s means of affecting it.
  5. A representative selection of other influences that could affect it (disturbances).

Discussion may include speculation about the kinds of perceptions being controlled and proposals for how to test those speculations. The place of a given CV in the hierarchy should be indicated not only by a label from the levels that Powers proposed on phenomenological grounds, but also by the subject’s means of control in terms of lower levels of control, and by higher-level CVs which might be controlled (in part) by setting or influencing the reference value for the given CV. This may be done by reference to the rows where these CVs are delineated in the spreadsheet that has been provided for collecting examples of different kinds of controlled variables and the behavior of controlling them.

I agree with you that control of perception is not a phenomenon. But nor is control, which I do not see as distinct from control of perception or control of input.

I have trouble calling control a phenomenon, because by definition a phenomenon is an appearance or immediate object of awareness in experience. From phenomena of behavior in experimentally varied conditions we can infer control, the CV which is controlled, and its reference value. By observing several kinds of phenomena of behavior (including actions, their environmental effects, means of causing those effects, and means of perceiving them) and experimentally intervening in these (by the experimenter’s control of perceptions called disturbances) we can validate such inferences.

I have nonetheless included the CV and its reference value in the list of the phenomena of behavior that I put in my proposed rewrite of the category intro. The CV and its reference value are theory-dependent phenomena which, like gravity as a force, are of a higher order than the “appearances or immediate objects of awareness in experience” from which they are inferred. Control, then, is a phenomenon of a next-higher order, empirically dependent upon the establishment of CVs and their reference values.

But maybe you’re making a distinction between control of perception and control.

Is a sun or moon halo, or a mirage, a phenomenon?

This is not an improvement over the existing desription of the Phenomena category. I point out the mistakes below:

That’s what conventional behavioral scientists do; they have described lots of observable phenomena of behavior (like power laws). But they have done it without understanding that what they call “behavior” is a process of control. So they have been describing “observable phenomena of behavior” that are often irrelevant to understanding how behavior works. I think you have to start any discussion of the phenomena of interest to PCT by noting that, from the PCT point of view, behavior IS control. Leaving that out is a significant mistake.

PCT views purposeful behavior as control and explains it as control of perception.

Number 3 is not the case.

Actually, once you have identified a CV you have identified the perception being controlled. Speculation about the kind of perception being controlled happens after one has already identified a possible CV but greater accuracy is desired. So a better way to say this would be: Discussion may include speculation about the precise nature of the CVs that have been identified and proposals for how to test those speculations.

I think this is being way to specific about what we would want to see in this category. I think what Powers wanted (and what could be accomplished by discussion in this category) was a database of observations of controlling – a database like the one in my spreadsheet that you mention below. Figuring out the hierarchical relationships between the CVs entered into the database (presumably based on some formal or informal testing) is a job for research of a kind that would be best described in the Research category.

I think it would be a good idea to point to that spreadsheet in the description of the Phenomena category. I’ll try to add something on it to my version of the description.

Control is definitely an observable phenomenon; it is the factual basis of PCT. Control is distinct from “control of perception”, which is theoretical proposal of PCT. Not everyone who has a control theory model of behavior applies it as a control of perception model. They don’t because they don’t see behavior as control, in fact, not in theory.

No, we observe the CV and infer the perception that corresponds to it. The inference is that the perception is the same as the CV.

When you see a person doing a simple compensatory tracking experiment, like the one here, you can see the CV (cursor - target position) and its reference state (zero difference) because you can see (from a plot of the data) that the cursor is being kept in that reference state, protected from disturbance by the person’s actions. This controlling is then accounted for nearly perfectly by a model that is controlling a perception that is exactly the same as the CV (cursor - target) relative to a reference of 0.

Yes, as did Powers. It’s good that you did since the CV and its (possibly variable) reference state are the fundamental observations (phenomena, data – whatever you want to call it) of PCT science.

Both the CV and gravity are facts that are explained by theories. The CV is a fact (stability in the face of disturbance) explained as control of perception; gravity is a fact (objects accelerate towards earth) that was explained as a force by Newton and more accurately as a warp in space-time by Einstein.

Yes, it’s an essential distinction that I describe in more detail here.

Best, Rick

You betcha!

Yes. Some explanations of phenomena may be deemed correct at a given time and place. Phenomena are perceptions and their explanations are perceptions. Phenomena are often deemed Real. Theoretical explanations, or conceptual entities in them, are also often deemed Real. The basis for such judgements seems to be their utility in feedback paths for controlling other perceptions. Sun and moon halos have little such utility (‘looks like it might rain’).

How do you perceive control without perceiving a controlled variable?

You don’t. You have to be able to perceive that a variable is being controlled – that it is a controlled variable – in order to see that control is happening. In order to perceive a controlled variable you have to be able to see that a variable is not being affected by disturbances as it would be expected to be. The problem is that in many cases the disturbances to a controlled variable are not visible, such as the gravitational forces that would be affecting the position of the teacup as it is lifted to the lips. But sometimes the disturbances are clearly visible, as in the rubber band demo, in which case you can easily perceive the controlled variable; in the case of the rubber band demo, you are perceiving the controlled variable when you are perceiving the position of the knot relative to the dot.

Here is my final recommendation for the description of the Phenomena Category:

PCT is based on William T. Powers’ realization that what we call behavior is a process of control. Control is an objective phenomenon that is seen when variable aspects of the world – controlled variables or CVs – are maintained in fixed or variable reference states, protected from the effects of disturbances. PCT explains this process as the control of perceptions that are analogs of the CVs. Topics in this category describe different examples of the controlling done by living systems. Discussion may include speculation about the precise nature of the CVs that have been identified and proposals for how to test those speculations.

– RSM

Well, I guess, Bruce, you’re not going to revise the descriptoin of the Phenomena category. Too bad. I think the current description is wrong. Also, I see you added a Motor Phenomena category. I think this is also a mistake becuase it could give the impression that, in PCT, control of variables that are considered “motor”, like limb position, is somehow different than control of variables that seem more “cognitive”, like the type of program being carried out.

Ah well.

Best, Rick

Have you tested that guess about my motivation, or is this intended as a test?

Consider that the means for perceiving (eyes, ears, etc.) and the means for affecting what is perceived are limited resources. When they are directed to variables that are controlled with very high gain they are not available for variables that are controlled with lower gain. Have you ever thought about modeling that bottleneck?

You’ve said quite a bit. Some of it seems to me self-contradictory. When I can get back to it, it will take time for me to integrate a hopefully useful reply.

Since you added the “Motor Phenomena” and “Collaborative biophysics research” topics shortly after I posted my final suggested revision of the description of the Phenomena category, I guessed that that was your answer to my proposed revision. That guess was given a bit more credibility when my proposed revision still didn’t appear for several days after you posted those new topics. And it was given still more credibility by my knowledge that you admire Martin Taylor’s view of PCT, which carefully avoids admitting that PCT is based on Powers’ insight that BEHAVIOR IS CONTROL, in FACT not just in THEORY. But if my guess is wrong and you just didn’t have the time to replace the current description of the Phenomena category with my suggested revision then great.

This is a good example of why the Phenomena category description should be changed. You have described no phenomena here, just your imaginings. What is an example of the phenomenon of “bottleneck”, for example? In PCT we develop models to account for phenomena – the phenomena of control - not other people’s imaginings.

I wait with great anticipation for your reply.

That seems a rather odd thing to say, especially when you use the word “admitting”. Personally, I would replace “carefully avoids admitting that” with “which depends on”. However, I do avoid talking about FACT when I talk about any finding of any branch of science. I regard it as hubris to claim that I am able to determine what is and is not fact (capitalized or not). Based on the long history of science, it’s hard to find anything that was once a generally accepted fact and has remained so for a long time (once, a long time was decades or centuries, but now a long time seems to be better measured in years).

In “FACT”, a good part of the early chapters of PPC is devoted to showing that if you accept Powers description of hierarchical control and further consider what Powers said he was deliberately avoiding, such as lateral interaction, and the distribution of neural impulses that Powers chose to combine into a “neural current” as a simplification you have something so insightful that the reader should not be ashamed of being dazzled.

I suppose such “facts”, which can be checked by reading the early part of PPC, must conflict with a prespecified reference value for a controlled perception of my understanding of Powers’s ideas. That reference value comes from some higher-level perception you are controlling. For me, it simply isn’t worth testing any conjecture about what that (those) higher-level controlled perception(s) might be.

I use the word “FACT” the same way Powers used it in the subtitle to Living Control Systems III: The FACT of Control. The FACT here is the objective phenomenon of control. This phenomenon can be described as: maintenance of a variable in a fixed or varying reference state, protected from the effects of disturbance. This is what living systems can be seen to do if one knows how to look properly: they control.

The FACT of control is the basis of PCT in the sense that there would be no PCT if Powers had not recognized and and developed an explanation of the FACT that living systems control.

I think one should accept Powers’ THEORY of hierarchical control provisionally, to the extent that it continues to account for the FACTS of hierarchical control, not because it incorporates or avoids incorporating certain favored or unfavored features.