Beliefs, factual and symbolic

Yes, I saw you question about the disturbance and forgot to answer,

No! In PCT the term “disturbance” refers to a variable that has effects on the controlled variable that are independent of those produced by the system controlling that variable.

I just wrote that off the top of my head. Here’s how Bill defines it in B:CP:

Disturbance: Any variable in the environment of a control system that (a) contributes to changes in the controlled quantity, and (b) is not controlled by the same control system.

These definitions are exactly equivalent when it is understood that “controlled quantity” was the term Bill used at the time for what we now call the “controlled variable”.

Right. If my definition of “disturbance” was the same as yours – “environmental forces or effects which cause the controlled variable away from its preference state” – and my definition of the controlled variable was the same as yours – “controlling for helping other people” – then my scenario would make no sense.

But my definition of “disturbance” is the PCT definition – a variable that has effects on the CV that are independent of those produced by the person controlling the CV. In my scenario, the disturbance variable was whether or not the helpee asks for help, a variable that has two possible values: “help me” and “don’t help me”.

I assumed that the helper’s CV was the perception of a contingency of the following kind:

If (the person says “help me”) then (give help) otherwise (don’t give help). (1)

[NB. “Giving help” and “Not giving help” are the helper’s behaviors that are part of the perception (the CV defined by equation (1)) that the helper is controlling.]

I assumed that the reference specification for the value of this CV is “true”. The CV will habe the value “true” when the helpee says “help me” and the helper helps and when the helpee says “don’t help me” and the helper doesn’t help. If the helpee says “help me” and the helper doesn’t help or if the helpee says “don’t help me” and the helper helps, the value of the CV will be “false”.

So, the helper controls the CV – keeps it at the reference value “true” – by helping when the helpee disturbs the CV by saying “help” and by not helping when the helpee disturbs the CV by saying “don’t help me”.

And these models have to be tested against data in order to know whether or not they explain the observed interactions between control systems. I have done that with models of conflict and Tom B. has done that with models of cooperation. Have you or any of the others studying collective control done that?

I am not familiar with any tests that demonstrate this. Do you know of any?

And I have tested models of learning by example against data. Do you know of any others who have done so?

Can you point me to any work on “collective control” (other than the stuff done by Tom B. and me) where the phenomenon was observed first followed by a model that was designed to account for the observation?

I was trying to describe the difference between a belief in the existence of god a statement that a person can believe to be true or false, and they can share this statement, as a proposition, with others, who can state whether they believe it or not. On the other hand, ‘belief’ in the presence of God is a current perceptual experience that an individual has that may or may not be similar to the experience another person has, and does not have a propositional meaning, just an experiential one. It can’t be easily distilled into words and shared exactly with other people to reproduce it exactly. But a belief in God is binary - yes or no (maybe also not sure!) and can be shared easily (although it is up to the listener or reader of the message whether they agree with it or not!).

This is another one of your peculiar posts inasmuch as I just don’t see the relevance of it to the discussion here. I’m sure that’s my failing but I would sure appreciate it if you (or anyone else who understands this) could help me understand how this is relevant to whatever the heck it is we’re discussing here;-)

Best, Rick

Sorry Rick, I think I posted it in a divergent thread from the initial question which was about the difference between ‘desire’ beliefs and ‘factual’ beliefs that Bruce mentioned a while back…

Even so, I still don’t get it. You say that the statement “I believe in the existence of god” differs from the statement “I believe in the presence of God” in that the first statement can be answered true or false while the second can’t. I’m not sure who is doing the answering – the person who says they believe or a person listening to that person. But assuming that it’s the person listening to someone who says those things and asked if you believe those things also, I would have no problem answering both “no”.

What was the point of making the distinction anyway?

Best, Rick

Hi Warren and others!
I think this is an important question but the terms desire and factual are not very successful. I am used to think with terms like propositional, verbal, explicit in one hand and tacit, nonverbal or procedural in another hand. The latter type means roughly the way how someone’s output functions work as Rick said earlier. The previous type means those special kind of control systems in us by which we control certain perceptions verbally eiher in imagination or in language. Often (outside PCY) only these language bound (and thus necessarily collective) cases are called big difference is that these language bound beliefs are typically always conscious while the other kind of beliefs are typically unconscious. Both can be more or less consistent with environment and other beliefs and thus either desires or factual.

T. Eetu Pikkarainen
(Kännykästä / From mobile phone)

I agree that only the tiniest beginning has been made at control-theoretic research in a populated environment.

Research into the phenomena of collective control cannot be done without first identifying phenomena to be investigated. Phenomena first.

The fruits of such research, results providing data to model a given phenomenon, cannot be set as a precondition for recognizing the phenomenon. Phenomena first.

The familiar phenomena of control have been researched in a depopulated laboratory environment, in which the subject is controlling and the investigator is ‘only observing’. Research and modeling of two systems conflictively controlling the same variable is very similar. The environmental feedback function is quite simple. A simple model of cooperative control of the same variable is very little different.

Phenomena of control in a populated environment differ from the familiar phenomena of control in a depopulated laboratory environment. They differ because the environmental feedback function is not so simple. The control systems involved in the more typical situations are not all controlling the same variable. Instead, for one individual’s control loop to be completed, some variable which functions as a link in the environmental feedback portion of that loop must be controlled by a different individual. The environmental feedback function is interrupted, and the first individual’s control is thwarted, unless and until another control system controls that variable in a way that completes that link in the environmental feedback path for the first individual.

How to identify, isolate, and measure real examples of such behavior is not an easy problem. Many people have worked at the identifying and isolating of the phenomena, but I am not aware of anyone having solved the problems of measurement suitably for computer simulation. Now that perhaps we are able to acknowledge that the phenomena exist we can begin to address the measurement problem.

In the very simple example of a handshake, if one person puts out their hand and the other doesn’t reciprocate, there is no handshake. If the second person does reach and grasp the first one’s hand, but then continues the movement to complete a martial arts maneuver that results in the first person lying on the floor, there is no handshake. If the second person reaches out and slaps the first person’s palm it’s an alternative protocol with similar function, but not a handshake. When Louis Armstrong came from behind the curtain onto Ed Sullivan’s stage back in 1961, Ed put out his hand, Satchmo gave him five, whirled around, slapped his hand again, whirled and slapped again, did it again, Ed smiling and still holding his hand out. My mother, watching, exclaimed disapprovingly ‘he won’t shake his hand!’ As an 11th or 12th grader in KKK central Florida I had not a clue either.

A handshake does not occur in isolation. What are the transactions which may include a handshake? Assuming that we have identified a variable or variables that are controlled by carrying out a transaction, how is that control of intended outcomes or sequelae different when a handshake is included and when it is not? For example, some employment interviews conclude with a handshake, others do not. Is there a relationship to which interviewees get hired? Intuitively, I think we would assume that a handshake that concludes an interview is simultaneously the initial move in a ‘welcome aboard’ communication, and that without it there is no welcome aboard.

Of course real life is not unidimensional. We might hear words like ‘We appreciate your applying. Good luck’ from the person initiating the handshake. (The interviewee probably would not initiate the handshake; why?). Are there differences of length of handshake, facial expression, and posture that are consistently observed when it’s a handshake of farewell, “good luck in your future endeavors”?

Observable phenomena, correlations, OK. Are these data upon which to model the behavior? How to get modelable data without the data ceasing to have a meaningful relation to actual social phenomena.

These are real phenomena, but they are not reducible to the physical variables by which they are made perceptually apparent. A handshake is a symbolic abstraction, a significant interaction between two people, not reducible to movements of hands or to muscle tensions and pressures on receptors in the skin. It is a communicative act. Each person demonstrates by this limited, stylized collective control of a handshake “I am willing and able to cooperate with you in collective control of other variables”. (One may of course be lying or may change his mind.) This general meaning of a handshake is made more particular by context, e.g. job interview, positive or negative decision.

Thank you Rick, the first part of your message was an important correction to one of my die hard beliefs. To the rest of the message I must try to return later.


In what way is it not so simple? Is it not simple because of the way it is implemented or is it not simple because of the nature of the function itself. Here’s a little diagram to show you the difference:

The top figure is a complexly implemented feedback function connecting system output to a CV. The different forms represent complex components of the feedback path, some or all of which could be people. The lower graph is a complex (and highly non-monotonic) function showing the relationship between variations in system output (on the x axis) and the CV. This could be the function relating system output to CV for the complexly connected feedback path at the top of the figure.

So what kind of complexity of the feedback function are we talking about?

Best, Rick

I think that you believe this represents what we are talking about.

I did go on to say ways in which the environmental feedback function is more complex when it intersects the environmental portion of other control systems’ control loops. It is more complex for the same reasons that the behavior of a living nematode or whale or marmoset or mouse or human when placed on an inclined plane is more complex than that of one of Galileo’s metal balls when placed on the same inclined plane. When the environment includes other control systems, it is a more complex environment. The environmental feedback function for an individual’s control of a CV may include some or none of that complexity. Social phenomena always include some.

Suppose the red pentagon represents a segment of the environmental feedback path, the state of which is subject to control by another control system (not shown). It could be in the input side of the environmental feedback path (not shown) or it could be an effective segment on both the output and input sides relative to the perceived/affected aspects of the environment labeled CV. You’ve shown it on the output side. For the effects of system output to affect the CV, the segment represented by the red pentagon must be in a certain range of conditions. The first system’s ability to control that CV through that environmental feedback path (including the input side) is conditional upon control within that range by the second system (not shown).

If the East Chop road is washed out by the nor’easter, Fred cannot get to Cronig’s Market by driving along the East Chop road. To control that perception by an environmental feedback path which includes driving along that road, Fred must wait for the town’s Highway Department personnel and those of the state Department of Public Transportation, among others, to rebuild the road. The break in the road is a physical contingency in the environment (if you try to drive past the break in the pavement your car falls of the pavement and becomes undriveable, not to mention injury to the driver). The process of restoring the road, with its delays, is a social contingency. This was very well laid out in 1993 by Bill Powers, Kent McClelland (here and in the cited paper “Perceptual Control and Social Power”), and Chuck Tucker.

Of course Kent’s work has advanced since 1993. We are not limited to what Kent, Bill, and Chuck said then, but nor has anything of what was said then been abandoned. The challenge remains of quantifying data suitable for modeling in the familiar ways. Gosh, we might have to do something unfamiliar.+

No, it’s my attempt to figure out what you are talking about. You keep saying that the environmental feedback function is more complex when other control systems are in the environmental portion of a person’s control loop compared to when they are not. This implies that you have in mind some way of measuring the complexity of the feedback function. What is that measure and how does the complexity of a feedback function affect control?

Yes, but you never said what that complexity is. In my previous post I suggested two possibilities; 1) something about the types of components and connections between them that make up the feedback function or 2) something about the mathematical relationship between system output and CV.

The input side of the feedback function is the system’s output; the output side is the CV, which is the input to the system. That’s all there is to the feedback function.

No, the red pentagon component of the feedback function is shown where it belongs, between the input (system output) and output (CV).

Of course. Same for any feedback function. For example, flipping up the light switch won’t get the CV to be in the desired state (“light on”) unless power is being delivered to the switch.

Thanks for the references. But what I really need to know in order to understand your point is what you mean by the complexity of the feedback function. Specifically, how to you measure it?

Best, Rick

The Powers reference was a brilliant discussion of the role of different types of feedback functions (he called them “contingencies”) in individual and social human behavior. The two types of feedback functions most relevant to social behavior he called “man-made physical contingencies” (like roads) and “man-made social contingencies” (like traffic laws). These feedback functions are easily described as contingencies relating output to CV. For example, roads create a contingency such as “you can drive to place X contingent on staying on the roads Y that take you there, otherwise not”. And traffic laws create a contingency such as “anyone who drives in place X above a certain speed Y will be arrested, otherwise not”.

The man-made physical contingencies like the roads are in force as soon as they are completed. The man-made social contingencies are in force as soon as people are trained to enforce them. In one case, the contingency (feedback function) is a physical property of the world (like the road) and in the other it is a behavior (controlling) carried out by another person (like a policeman). But in both cases the feedback function can be described very simply as a contingency. There is no difference in the complexity of feedback functions – they all can be expressed as contingencies between action and result; output and controlled variable.

I couldn’t find anything about feedback functions in the other two papers. But the paper by Powers made my day, and I didn’t even feel lucky. So thanks for the reference. Every time I go back and read Bill’s stuff I realize what a brilliant and articulate person he was. And on top of that he was a really nice guy.

The question is what is collective control and how do we model it. Eventually we’ll work our way to the topic of different kinds of beliefs.

Yes, the feedback function is a mathematical expression in a model. I am talking about phenomena and you are talking about a model of phenomena. You ask how the environmental feedback function for collective control can possibly differ from the environmental feedback function for control in an unpopulated environment.

The environmental feedback function is a mathematical expression representing a causal relationship between control outputs into the environment and environmental effects upon the sensory inputs of the same control system. Typically, this involves only immutable physical laws, that is, mathematical representations of causal relationships which are completely dependable. F=MA and so forth. Any factor in the transmission from control output to controlled sensory input which is not derived from completely predictable physical law is a disturbance.

When you talk about a model of control phenomena, the output side of the environmental feedback function is Qi. You say that Qi is the CV.

Getting measures of Qo and Qi for participants in social phenomena is usually impractical and can be unethical, and it certainly cannot be required of me in this discussion. Fortunately, control is an observable phenomenon which we can verify without measuring Qi and Qo and without expressing the physics of the environment between control output and sensory input mathematically as an environmental feedback function. And, fortunately, we have a long and honored history of investigating control on those terms in this research community.

When we observe control, the CV is observed in the environment. There are links of cause and effect between the subject’s output and the CV. This is one side of the environmental feedback path. There are also links of cause and effect between the CV and the subject’s sensors, the other side of the environmental feedback path. A fundamental task in the TCV is to employ the corresponding sensors and to arrange matters to ensure that your environmental feedback path from the CV has the same effect on your sensors as the corresponding path has for the subject’s sensors. So already there’s a lot to consider out there in the environment. In the modeling point of view as represented by your diagram of an environmental feedback function there are no environmental factors between Qi and the sensors, and the CV is not represented separately from Qi.

Disturbances may intrude directly on the CV, or at any point in either side of the environmental feedback path. In the TCV, one test is to interrupt the environmental feedback path between the putative CV and the subject’s relevant sensors and observe whether that interrupts control. In practice, unless the subject is constrained by accepting artificial constraints in an experimental situation, the subject very probably will act to remove the obstruction. In general, elements of the environmental feedback path may themselves become controlled variables if, due to some disturbance to them, their efficiency in transmitting energy from outputs to CV or from CV to sensors is degraded. This is an important factor in collective control to which we shall return.

There may be many links in an environmental feedback path. They may be represented by distinct mathematical expressions within the environmental feedback function (a mathematical expression representing the cause-effect relation of control system output to input through the environment). The little polygons in your diagram presumably represent distinct mathematical expressions within the environmental feedback function.

I have taken the liberty of labeling the axes with the appropriate variables for the control system equations (B:CP p. 274 of the 1973 edition). The relevant equation of the four is
Ke (with subscript e) is the constant K representing the instantaneous ‘steady-state condition’ of the environmental feedback function.

However, the little polygons in your diagram cannot include any causal links between the CV and the sensors for reasons that are explained above and illustrated in a canonical diagram of an elementary control loop:

In the modeling point of view exemplified here and in your diagram the CV that we observe in the environment does not exist. It melts away into a cloud of “physical properties that convert action or behavior into effect on input quantity”.

This omission of the environmental CV from the equations, diagrams, and terminology of the modeling point of view is the source of countless muddled disputes on CSGnet as to what is controlled, and whether the CV is the perceptual signal p, the input quantity Qi, or that in the environment which the subject, the experimenter, and others present verify to be controlled. The very existence of an ECV or CCEV has been denied. Just because our senses are our only means of knowing it is there does not mean that it is not there. To me, it is patently obvious that all three are controlled, because control brings all three into correspondence. If we look for which causes the controlled condition of the others, we can only say that the functioning of the control loop as a whole is causative, reducing its error output to zero.

There may be many links in an environmental feedback path. An obvious example that presently exists is a Waldo on the output side and magnifying imaging equipment on the input side as equipment for a neurosurgeon.

The stability of the links in the environmental feedback path is important. An unstable link in the environmental feedback path degrades control. I once drove a car with worn steering linkage across country coast to coast. The steering linkage was worn, resulting in a control-free gap of 3-4 inches between pressure to steer to the left and pressure to steer to the right. In a small town, a fellow driving an MG convertible wanted his passenger to hand something to one of my riders in the back seat reaching out through the window. His steering linkage was nice and tight, precise control. I could ease my car close to his, but a bump in the road could make the wheels turn left toward him ahead of the pressure I was applying to the steering wheel, and to correct I would have to move through several inches before corrective pressure to the right made contact, and overshoot was likely. He kind of scoffed at what a nervous driver I was. I told him if he could see the play in my steering wheel he’d be nervous too. That control-free gap between pressure to steer to the left and pressure to steer to the right was not a source of disturbance, it was a property of that link in the environmental feedback path between the output of my arms and hands on the steering wheel and the direction of motion of the car, a property which degraded the precision of my control. Because it degraded the gain in that control loop, I was unable to resist environmental disturbances as well as I wanted to. The deficiency of that car, localized in the steering linkage, itself became a controlled perception. I was glad to find a replacement vehicle for use in the many environmental feedback paths that required me to drive someplace.

This illustrates the importance of distinguishing disturbances from properties of links in the environmental feedback path that affect their efficiency and efficacy as links transmitting energy of outputs to a CV or transmitting energy which (at the end of the path) is perceived as the variable state of the CV. It also illustrates that we control perceptions of the links in an environmental feedback path, and perceptions of how well they serve that function in our control.

A great deal of the field of collective control concerns aspects of the environment that people make use of as links in the environmental feedback paths of various control loops. Much of the passage about work that I have quoted from Kent’s chapter concerns the creation and maintenance of such artifacts. Perhaps you could now go back to earlier posts in this topic and reconsider them from this point of view.

Earlier in the “Beliefs, factual and symbolic” thread you said:

There are several points in this current post that suggest that this is not the case. I will point them out and explain why each represents a significant change that is not an improvement to PCT.

This represents a change in PCT because it implies that you can deal with phenomena and models separately. In PCT, models and phenomena are intrinsically linked. Feedback functions are not just mathematics (models); they are also properties of control phenomena that can be represented mathematically in models of those phenomena. Two examples were given by Powers in the article you pointed me to. One was a road and the other a traffic cop. Both are feedback functions that can be modeled as a mathematical contingencies – if X then Y else not Y – between system output (X) and the variable controlled by the controller (Y) .

This is an unnecessary and misleading addition to the PCT model. The CV is a perception; what is in the environment is the basis of the CV. The CV is shown as being in the environment because it is the CV from the perspective of the observer. What you call the “links of cause and effect between the CV and the subject’s sensors” could only be links from the environmental basis of the CV to the sensors, things such as glasses or hearing aids. These links are implicitly included in the feedback function, which is an an observed relationship between variations in system output and variations in the CV.

And they don’t need to since the CV is a perception that is a function of what’s on the sensors.

This is just plain wrong. Poetic, but wrong. You really lose much of the beauty of PCT by looking at things this way. But I will try to restore the beauty for you by demonstrating that the CV that’s shown in the environment of the control diagram most definitely does not "melt away into a cloud of ‘physical properties that convert action or behavior into effect on input quantity’. It exists and it’s the most important reality in that diagram.

I’ll demonstrate this to you using my “What is Size?” demo. In that demo you can control either the perimeter (P) or the area (A) of a rectangle. That is, you can control either one of two different perceptions, P = k(H+W) or A = k(H*W), of the same physical reality.

These perceptions, P and A, actually exist as possible CVs. The fact that these two different CVs exist when they are being controlled can be seen via modelling because the feedback function is different depending on which perceptual variable is the CV. When P is the CV then, the feedback function is CV = O + W; when the CV is A the feedback function is CV = O * W. So, the environmental feedback function connecting output (O) to the CV is determined by the CV. This wouldn’t happen if the CV were, as you say, non-existent. I think the fact the the nature of the CV changes the nature of the environmental feedback function proves that the CV exists as a real aspect of the environment.

Agreed, and it’s very important in modeling. I was going to present on this at the conference; maybe I still will.

If it is, it isn’t obvious to me since those feedback paths connect outputs to controlled variables and I don’t recall seeing much about what variables people are controlling. I learned more about the social implications of man-made feedback functions from Bill’s brief little paper then from all of the stuff I’ve read on “collective control”. Bill’s paper was written in 1993. I think he tried to point sociologists in the right direction but, apparently, to no avail.

Yes, p is controlled. Yes, the controlled quantity qi is controlled, the quantitative effect which the controlled aspect of the environment has upon the senses. And also yes, we observe in the environment a (variable) reference value at which outputs of a control system maintain the state or condition of an aspect of the environment despite disturbances which would otherwise change it.

When we test our hypothesis about the CV, we do not disturb qi directly. We do not shine a variable source of light into the subject’s eyes, we move something in the environment from which light reflects into ours and the subject’s eyes.

Yes, Bill defines ‘function’ in two senses, as a description and as that which is described:

“FUNCTION: A rule making the state of one variable dependent on the states of one or more other variables. A physical entity embodying processes described by the function rule.”

The environmental feedback function is a property of the environment, and it must be observed and measured in the environment.

“The disturbance function and the environmental feedback function ke are properties of the environment and thus are measurable.” B:CP p. 290 (2005).

That sentence is not found in the original 1973 edition of B:CP. The appendix “Control System Operation and Stability” was extensively rewritten for the 2005 revised edition. The 1973 edition calls ke the ‘environment function’ and says little about it other than normalization of qi and ke relative to “the maximum value of the perceptual signal p” (p. 275).

It is informative to compare the two versions of this appendix and see what changes Bill found to be necessary. During 30 years of building generative computer simulations of behavior, Bill came to emphasize the importance of modeling the environment. I think it was in connection with later revisions of the Little Man demo that he remarked that the model of the environment is where the complexity lies, the control-theoretic parts are much more straightforward. An advantage of robotics is that it does not have to model the environment, the robot is interactive with the actual environment.

Your simulations all have extremely simple simulated environments. For ball catching, the field is perfectly flat, the runner moves without legs and catches without arms. There is no wind. The center fielder does not need to control perceptions of where the right and left fielders are, or into whose territory the ball is flying. It is proper to exclude aspects of the environment which are irrelevant to the interception question being asked, namely: How does the fielder get to the place where the ball will fall? The toy environment is a horizontal plane, and the runner’s line of sight to various frictionless parabolic ball paths vertical to that plane. For the abstract purpose of the model you can ignore the environmental feedback path, e.g. you can ignore the possibility of disturbances at any point along the ball’s parabolic path as determined by the mathematics of physical law, or at any point along the runner’s path as determined by the mathematics of control. If the camera-carrying runner stumbled, those data were discarded.

I am calling your attention to situations in which control by others is part of the environmental feedback function. For example, there are many situations in which qi does not arrive at the subject’s sensors unless intervening or otherwise related parts of the environment are in their reference conditions, as perceived and controlled by other control systems.

For what purpose was the social ritual of a handshake initiated and reciprocated by the two individuals whose performance of it is memorialized in this painting?
The perception of the ending of the American Civil War was was controlled through an environmental feedback path that included the handshake depicted here, as well as some documents bearing the signatures of the two men, the control activities of the other men in the room and the control activities of literally armies of others not depicted, their economic and political supporters, etc. A rather capacious et cetera.

Even though so many players and interests are involved, that’s a relatively straightforward example. Here’s an example that is more complicated. For what purpose did one man refuse to reciprocate the handshake initiated by the other, as shown in the photograph of the event?

American Col. Sidney Mashbir was acquainted with Japanese Lt. General Torashiro Kawabe before the war. This photo was taken as Kawabe arrived in Manila to prepare the details for Japan’s surrender ceremony. Was Kawabe seeking to reaffirm their prior friendship? Was Mashbir controlling a perception of Kawabe being publicly shamed? What is ‘shame’? Doesn’t it have to do with the shamed person’s perception of how others perceive him? How do you quantify shame? Shame was a prominent variable in Japanese culture (see Benedict The Chrysanthemum and the Sword). Does that change the quantification? Does it make the quantity of the ‘shame’ variable greater for Kawabe than the quantity that Mashbir was controlling for Kawabe to experience? How do you shame someone (control a perception of another person’s emotion of shame)? Or, instead, was Mashbir unwilling for his fellow American soldiers to know about his prior familiarity and perhaps friendship with Kawabe? Or, assuming they knew, was he indicating to them as well as to Kawabe that any such relationship was ended? He’s looking down, visor blocking eye contact, turning (left foot lifted and knee bent to pivot to the right) and his thumb is rising to point toward a place behind his back, which may indicate a communication “Let’s just get over there and get on with the business at hand.” We can only speculate about these things, but one thing is certain: Kawabe initiated a handshake and Manbir refused it. Where a handshake requires collaboration and thereby suggests possible further collaboration, a refusal to reciprocate an offered handshake indicates that this person is unlikely to collaborate in anything else.

A handshake is means of controlling something other than the handshake. Whether accepted or refused, everybody in these scenes knows what a handshake is and how to participate in it. As far as we’re concerned, everybody in the world, for all practical purposes. It is a human invention. Like language, handshakes pre-exist in the world that a child is born into, as present and real in the environment as breezes and trees.

Until we account for what is of human interest, PCT is of limited value and limited interest. “The disturbance function and the environmental feedback function ke are properties of the environment and thus are measurable.” Modeling the environment is not an easy task.

The CV is exactly equivalent to p. When we observe the outputs of a control system keeping a variable aspect of the environment in a constant or variable reference state then we know that that variable is a CV which corresponds EXACTLY to p in the control system. The feedback function is the effect of system output on the CV which is the same as saying that the feedback function is the effect of output, via the environment, on a perceived aspect of the environment.

This means that you can’t know what the feedback function is until you know what perception – what CV – the system is controlling!! This is demonstrated in my “What is Size” demo. When you are controlling the perimeter of the rectangle (a perception) the feedback function from mouse output, O, to perimeter value, CV, is CV = H + O, where H is the varying height of the rectangle. When you are controlling area as the CV (a different perception of the SAME physical reality) the feedback function is CV = H * O. This change in the feedback function involved no change in the environmental connection from O to the lines on the screen that are the basis of the perception of perimeter and area nor has there been any change in the lines themselves!!. All that is changed is the perceptual aspect of the environment you are controlling. That change is enough to change the feedback function.

Once you understand the above, then we can start talking about feedback functions.

Yes, we do!

If this were true – if, when we do the TCV we are not disturbing qi, which is exactly equivalent to disturbing the CV (or p) – then it would be impossible to test to determine the variables people are controlling when they behave. This would mean that PCT is untestable. So this idea of yours is a rather significant change from PCT as described by Powers, who went out of his was to show how to test PCT.

Correct. And in very simple examples of control in an unpopulated environment it is easy to ignore the fact that it is an aspect of the environment.

You haven’t responded to any of the questions that I have asked. I’ll wait.

Whether it is easy or hard to ignore, the important fact is that the “What is Size?” demo shows clearly and quantitatively that the nature of the feedback function depends not only on the nature of the environment in which behavior occurs; it also depends on the nature of the perceptual variable that is being controlled – the CV.

In the “What is Size?” demo, the feedback function changes when the controlled perception changes and the environment remains the same. The lines on the screen and the connection to them from your output (mouse movement) make up the environment in which you are doing your controlling. This environment is the same whether you are controlling a perception of area or perimeter! The feedback function, however, is quite different depending on what perceptual aspect of that environment is being controlled.

This means that you learn nothing about a person’s behavior by just looking at the nature of the environment in which the person is behaving. In order to determine the nature of the feedback function you have to know what perceptual variable (CV) – what aspect of the environment – the person is controlling.

The main shortcoming I see in your approach to understanding social behavior is that you don’t seem to try to identify the perceptual variables – CVs – around which any particular behavior is organized. You seem to think that by determining the nature of the environment in which a person behaves you can understand their behavior. The problem is that, as in the “What is Size” demo, there are many possible perceptual aspects of the SAME environment that a person might be controlling. So, as Powers has explained repeatedly, if you want to understand a person’s social or individual behavior you have to first know what perceptual variable(s) he or she is controlling. In simple terms, what that means is that you have to know what people want before you can determine what they have to do to get it.

You’re right. They seemed like questions that are more appropriate for a historian. But I’ll answer them:

[quote=“bnhpct, post:58, topic:16110, full:true”]
Q: For what purpose was the social ritual of a handshake initiated and reciprocated by the two individuals whose performance of it is memorialized in this painting?

A: My guess is that Lee was controlling for being perceived as having agreed to surrender and Grant was controlling for being perceived as having accepted Lee’s surrender.

A: The man who refused was probably controlling for being perceived as not having agreed to whatever was proposed by the man offering his hand.

I think that if you learned how to correctly analyze behavior from a PCT perspective you would find that it accounts for things that are of considerable human interest. For example, read Bill’s brilliant essay Degrees of freedom in social interactions on the social implications of PCT.

As I noted above, the environmental feedback function, ke, is a property of the environment and of the perceptual aspect of that environment that is being controlled, both of which are, indeed, measurable.

Yes. And when the actions and intentions of others are constituent inputs of the controlled perception they need to be included in the model of the environment.

The phrase “actions and intentions of others” refers particularly to control by others of environmental variables which the subject perceives and which are salient for the subject’s control. And it goes without saying (but to forestall cavil I probably have to say) that for any of the modeled participants these are the subject’s perceptions of the actions and intentions of others.

I enumerated a variety of purposes a person might have for initiating a handshake, and purposes a person might have for reciprocating or refusing to participate in enacting the social ritual of a handshake.

Do you consider that an adequate specification of the CV? What exactly is a ‘surrender’ in general, and what exactly was that particular ‘surrender’. Are they controlling the same CV? How is this possible without conflict?

You don’t seem to try very hard to identify the perceptual variables – CVs – around which this behavior is organized.

Obviously, we can’t directly inquire into what the individuals involved were controlling in these events, but history is not to be scorned. A record of experimental investigations is a form of history.

In both of these cases, the presence of witnesses is important, but even when there are only two participants each is witness with respect to the other. There are many situations in which an important CV is a perception of how others perceive what one is doing. The photograph is an especially obvious example.

The main problem is methodological. Without valid and agreed method there can be no valid and agreed data. (Note that validity and agreement are terms of collective control which are essential to science.) We do not have an accepted methodology for investigating such CVs. Methodologies exist and have been refined for a long time, but to obtain by them data suitable for a PCT simulation of behavior remains almost entirely unexplored.