Beliefs, factual and symbolic

Rick: “What is germane (at least to me) is the answer to my original question “How are beliefs collectively controlled?” I don’t believe I’ve ever gotten a clear answer to that."

Just one example. When your teachers and parents and others taught you that the earth is round and not flat then they took part into collective control of beliefs.

Rick: “But this discussion has not been a waste of time for me.”

For me neither :slight_smile:

Rick: “Indeed, it led me to realize that there are two very different ways individual organisms (controllers) can be involved in “collective control” : 1) as disturbances to the variables controlled by other controllers and 2) as components of the feedback functions of other controllers. I’d be interested in what you (and others involved in the study of collective control) think about this distinction.”

This is really a basic distinction but a quite abstract one. In real social life these alternatives are often mixed in complex ways. For example a simple cooperative case of helping. My asking for help or just looking helpless is a disturbance for a helper, who tries to cancel that disturbance by offering herself as a feedback link for my control. Martin has a rich theory of these kinds of interactions as protocols.

Thank you both for discussing!


How do you know? Do you have some data on the relative incidence of the “feedback function” versus the “disturbance” type of collective control?

My main “belief” about Kent’s demos of collective control (which certainly may be mistaken) is that they have not been shown to apply to any social phenomenon since they have not been tested against data.

I know. And I report on several models of such conflict-free examples of collective control in my book The Study of Living Control Systems. These include Tom Bourbon’s models of two person interaction, Bill’s CROWD model and Reynold’s flocking model and my model of phonemic drift (and stability).

I described a high level verbal “model” of a handshake from one person’s perspective. I put “model” in scare quotes because I don’t consider these verbal descriptions to be models. And I have developed tasks where a virtual person (a control system) is part of the feedback connection between another person (me) and the variable I’m controlling. I’ve reimplemented that demo in a scenario where you, in the form of a sheepdog, control the position of a renegade sheep as the means of keeping it in the herd. The renegade sheep’s controlling (for avoiding the dog) is the feedback function connecting the dog to the variable it is controlling (sheep together in herd).

I’ve read Kent’s chapter and just the first part of Martin’s book. I don’t remember seeing any examples of the latter kind of collective control in either of these sources… Could you describe a couple of their examples of collective control where the individuals in the collective are part of the feedback functions of others in the collective?

Yes, I’ve 40+ years of experience with this. I have found the denial and reorganization that I’ve experienced in response to PCT evidence both frustrating and exhilarating. Now I just find it fascinating.

That’s an assertion, not an explanation; it doesn’t tell me HOW my belief that the earth is round was collectively controlled. What were the control systems involved; what were their inputs, outputs, controlled variables, disturbances?

Excellent example. And it involves collective control of both type 1 and type 2. I would just suggest adding that your plea for help is a disturbance to the helper only if the helper is controlling for helping; that is, the plea is a disturbance only if helping is a perceptual variable controlled by the helper. And the helper becomes part of the feedback function between you and what you are controlling only if she does more than just offer, but actually helps.

Do you need a page number to believe that this example occurs in both Kent’s and Martin’s writings? I have no reason to doubt your word that you have read them. That you do not remember any examples suggests to me that you read them dismissively, looking for data in forms with which you are familiar and and setting aside anything without such data as being unimportant and unmemorable, not worth the trouble to remember.

Note Eetu’s observation that even the most basic of such examples is abstract. Abstract CVs ride on top of control at lower levels but cannot be reduced to being no more than control at lower levels. What you have been willing to accept as data has been limited to the latter. The problems collecting data about abstract ‘symbolic’ CVs are different from the familiar problems collecting data about motor control. Different measures and different methods are required. Since it is the abstract ‘symbolic’ CVs which are of greatest interest to most people, developing or adapting such measures and methods in ways appropriate for PCT is of great importance for the advancement of our science and for its understanding and acceptance outside our small enclave.

In the TCV, you imagine how the environment appears from the subject’s point of view, you imagine that the subject is controlling some variable that you can perceive, you imagine that the observed condition of that variable is the subject’s reference value for it, and you confirm or disconfirm what you are imagining by controlling it to a slightly diverging reference value.

In every example of collective control the participants are confirming or disconfirming what they imagine that others are controlling. For example, you remember experiences in which your control relied upon cooperative collective control with a person that you have not seen for a while. You imagine that you can continue to rely upon them in this way, should you need to request it. You confirm or disconfirm what you are imagining by sticking out your hand to initiate a handshake, “Hello, Rick! Goodness how many years has it been?” His reciprocating with his hand and cooperating in the execution of a handshake — a conventionalized encapsulation of relying upon one another to accomplish cooperative control — provides some confirmation of what you are imagining. A refusal would disconfirm it. As you pointed out, the gain in the other person’s reciprocated control gives you more nuanced information about your social relationship with him. The manner of the handshake is often a means of controlling relative status and relative control capacity (social power). A 'power handshake’ asserting dominance by rotating your hand to a horizontal position can be countered by swiveling to the other’s left side to join him (generally a him for this dominance stuff, right?) in facing an imagined or actual audience, with the appearance of sweetly holding hands. Gender differences are typically ignored in discussions. There’s a lot more dimensionality that we could get into, but we’re keeping it simple.

Likewise, a request for help begins with imagining the other person helping, which may be confirmed or disconfirmed by what they do next. The imagined perception is a looped-back reference value within a loop participating in collective control.

If the CV is an interpersonal relationship of reliability of each to the other for participation in collective control, at need, what are the data for that CV and how do you collect and measure such data? The data are not in approximation of hand to hand, the angles of elbows, movements of hands and digits, pressures of fingers. The data of collective control are of an entirely different kind. The CV of mutual reliance can be controlled by means other than a handshake.

Here’s a report of some fMRI data.

No, I believe it. But there are problems with this example that I pointed out in the part of my reply to Eetu’s comment that you didn’t quote. What I said was:

I suppose one could say that calling the helper a “helper” implies that that person is controlling for helping. But I think it’s important to explicitly state that the plea for help is a disturbance only if the person from whom one is soliciting help is controlling for a perception of themselves helping. Leaving out the assumption that the helper’s controlled variable is to “help” in this scenario makes it sound like the plea for help is a stimulus that causes the helper to help. But the plea is not a stimulus – that’s a behavioral illusion. The plea only appears to be a stimulus when it is a disturbance to a controlled variable, such as helping someone. This is evidenced by the fact that the plea for help no longer has any stimulating power – it is no longer a disturbance-- if the helper is not controlling for helping. What constitutes a disturbance is defined by what variable is being controlled. Very important to understand this.

And on the point about the helper becoming part of the feedback function between the helpee and the variable the helpee is controlling, in order to be part of that function the helper must not only actually help, but her helping behavior must be directed by the helpee’s outputs . So if I am the helpee needing help jacking up my car (the controlled variable being a lifted wheel) a helper would be in the feedback function between my output and that controlled variable only if all of her effects on the controlled variable were directed by me. I believe that few would-be helpers could (or would) yield to this level of micromanagement since it would be a gross disturbance to their own controlling. This is another reason why I think people being part of the feedback functions of other people would be extremely rare.

So, while Eetu’s description of the helper as part of the helpee’s feedback function is accurate, it describes a rather unrealistic situation. In reality, when someone helps you do something, they are actually in the position of a disturbance – an independent effect on the variable you are controlling – rather than a part of the feedback connection between your outputs and the CV. In the jacking up the car situation, the helper will be a “good” disturbance --one that has effects on the CV that move it to its reference state-- if she knows how to jack up a car to the proper level, otherwise the helper is likely to be a bad disturbance – and an active one at that, so there would be a lot of conflict.

I couldn’t agree more. So what are all of these “measures and methods” that can be used for studying those “abstract ‘symbolic’ CVs”.

I think it’s better to say that in the TCV you hypothesize, based on observation of the subject’s overt behavior, the perceptual variable the subject is controlling – the CV. The hypothesized CV is not necessarily a variable or level of a variable that you can perceive yourself but one that you know can be perceived by some means, such as the special acoustic sensors that are used to perceive the high frequency acoustic pulses of the bat. You don’t have to assume that the hypothetical CV is being kept at a particular reference state; the TCV works even when the subject’s reference for the CV is varying (see my Mind Reading demo).

You test your hypothesis about the CV by disturbing the hypothesized CV and looking for lack of effect of the disturbance. If the disturbance is effective, your hypothesis about the CV is wrong and you try another hypothesis. You keep trying different hypotheses about the CV until you find one that is least affected by disturbances or gives the best fit to observed behavior by a control system model that includes that variable as the CV.

How do you know that? And how do you know it’s true for all the different kinds of collective behavior?

I’m sorry, this kind of psychologizing is not my “cup of tea”, so to speak. And the fMRI study strikes me as an example of pop-psychology at its worst.

But I would like to know about the measures and methods you have developed for studying abstract ‘symbolic’ CVs.

Best, Rick

No one here is likely to call it a stimulus.

But this is not in the first of the two types of collective control that you distinguished, namely, conflict arising from disturbance. Here, each participant is controlling part of a complex CV, and in that way they each are “components of the feedback functions of other controllers.”

Asking for help may indeed be a disturbance (or not), but that is not how it constitutes a request. It constitutes a request because the asker controls variables perceptible to the other which the other recognizes as a request for help.

Asking is not mere disturbance of what the other is controlling, it’s control of the social conventions for asking for and receiving help. Both parties know how to play either part in the transaction, and have done so countless times in the course of their lives. They can do so because each has learned the protocol during many experiences of observation, instruction, and participation. (These learning processes resulting in shared or homologous input functions is another aspect of collective control.) The details of what constitutes asking for help and what constitutes giving help are context dependent and quite variable, but the turn-taking form of the transaction is consistent. The transaction can also be broken, interrupted, abandoned, or mistakenly perceived, but these accidents can happen to control of any sequence.

Indeed. That is likely why that kind of micromanagement is in fact exceptional. The norm is more like the TCV, by which the helper narrows down a perception of the CV and participates in controlling it. If the helper doesn’t accurately perceive the CV, or or its current value, or its reference value, or aspects of the environmental feedback function, their participation may introduce disturbances which the assisted person then resists. The resistance may include communication by gesture and language, and may even constitute explicit instruction, but it is transitory, since in general resistance to disturbance persists only as long as the disturbance persists. (There are seeming exceptions involving imagination.)

First, identify phenomena before identifying data and how to measure it by what methods. We’ve made a start with conventionalized social transactions. These transactions are observable phenomena. A handshake is one example. It is usually part of a more complex transaction.

A fundamental characteristic of this class of conventionalized social transactions is that they are sequence perceptions. In general, the completion of step n (perhaps a perception of a product of that step) must be included in perceptual input for starting to control step n+1. Since you did raise the point, it may be necessary to affirm that this does not mean that the conclusion of step n is a stimulus for performing step n+1. Skinner thought so, and others have, but that’s not part of this discussion.

In ordinary sequence control, one person performs the successive steps. In collective control of a conventionalized social transaction, the participants alternate in performing the successive steps of the sequence. This is part of what makes it collectively controlled. Another aspect is how all the participants have come to know the structure of the transaction and know how to participate in it. That learning process and the resulting standardization is also accomplished and maintained by collective control. But we’ll leave that out of the discussion for now. Keep it simple.

Each step in a conventionally structured transaction can be controlled by a variety of means. A raised eyebrow can constitute a question. A slightly explosive release of breath and a dropping of hands away from a task can constitute a request for help.

A characteristic of symbolic culture is that control of one CV can constitute display of an abstract CV for others to control. I don’t mean something so narrowly coded as when the agent in Graham Greene’s novel The Quiet American goes to the window holding a book of poetry an his visitor is later assassinated. Collective control there, yes, but only two people were in on that convention, and I’m talking here of the body of conventions within a community.

So we’re talking about relationship or categorial perceptions (as in a raised eyebrow can constitute a question) and structurings of these (as in the temporal sequence in a transaction such as a handshake, or asking-answering, or asking-helping).

Methods for collecting data on these have been developed over a long time by many people investigating language, kinesics, symbolic culture, and sociology, among other fields.

To make use of those methods and data, or other methods and data if need be, we have to devise ways of making such data suitable for modeling the control systems that are interacting. While controlling the CV for one step in a sequence may be amenable to quantitative measurement, the sequence itself is not. A step is completed or not, a binary choice, as condition for controlling the next step.

When your teachers and parents and others taught you that the earth is round and not flat then they took part into collective control of beliefs.

Rick: That’s an assertion, not an explanation; it doesn’t tell ne HOW my belief that the earth is round was collectively controlled. What were the control systems involved; what were their inputs, outputs, controlled variables, disturbances?

Eetu: I try to keep my hypothetical explanation or description as simple as possible. We need three people, two adults who both control for a belief that earth is round. That they control a common CV, makes that control collective. (Simple cooperative control.) The third person (a child perhaps) believes that earth is flat. Her expression of that belief, e.g. saying that earth is flat, is a disturbance to the control systems of those other two and causes an error in them. Their outputs can be verbal explanations, physical demonstrations, or even hitting to the head by a book as in the old school they used to do. I think this is enough. Whether their control is successful or not is a different matter.

Then to the other issue which Bruce has already considered in more detailed manner.

Rick: I would just suggest adding that your plea for help is a disturbance to the helper only if the helper is controlling for helping; if helping is a perceptual variable controlled by the helper.

Eetu: that is probably not so. Some (but not many) of us may be like scouters just searching for someone to help. But in that case a plea for help is of course not a disturbance for their controlling. A more safe assumption is that we control for perceiving fellow people OK and contented. Perceiving unhappy and suffering or help begging neighbors is a disturbance which can be canceled in principle by two kind of output: either helping or turning the blind eye. Only in this phase of the process the helper is hopefully controlling for helping.

Rick: And the helper becomes part of the feedback function between you and what you are controlling only if she does more than just offer but actually helps.

Eetu: Independently of what the helper is actually controlling for she can become part of the feedback function (a feedback link or atenfel) of the helpee if she only happens to cause - as a side effect or main effect of her controlling - something which helps the helpee control what ever the helpee is controlling for. For example, if you want just to be alone for a while you can ask your partner to go and by you something from a shop - something which you really don’t need.

T. Eetu Pikkarainen
(Kännykästä / From mobile phone)

I presume you will be presenting your ideas about collective control to an audience that includes people who are not as well versed in the PCT perspective on behavior as you are. Most conventional behavioral scientists (including linguists) would see the helper’s behavior as a response to the helpee’s plea. They would see the plea as a stimulus or cause of the helping rather than as a disturbance to a CV that produces an error that causes the helping that is BOTH an effect (a response caused by the error) and a cause (of the CV moving toward its reference state). It’s a control process, the central feature of which is the CONTROLLED VARIABLE!

Actually, each person is controlling different CVs that are brought to their reference state by their actions – pleading and helping – that compensate for the disturbance – pleading and helping – produced by the other person. Neither person has control over the actions of the other – the helpee can’t make the helper help and the helper can’t make the helpee plead for help. So the the helper’s help and the helpee’s plea are disturbances – not part of the feedback functions of helper or helpee – because they have effects on the CVs that are independent of the outputs of both the helper and helpee.

Again, this is an incorrect analysis. If the helper is controlling for helping when requested, then a request for help will be a disturbance to that perception that is corrected by the helper helping. Of course the asker has to control variables perceptible to the helper which the helper recognizes as a request for help, but that’s just another way of saying that the asker has to ask; speaking – saying “help” – is, of course, a control process.

Before I address the rest of your comments I think we should find out if we are on the same page in our understanding of what a disturbance and feedback function are in the PCT model.

Best, Rick

No, you are still not on the right page.

Asking and receiving are two parts of one communication protocol, a sequence perception like any sequence perception except that the participants alternate in executing the steps.

Independently on Bruce’s idea of a collective sequence perception, Rick’s concept of disturbance seems a little strange to me.
I have understood that disturbance means the sum of the forces in the environment which cause the controlled variable (CV) away from its preference state. Is this right? If it is then every force which causes the CV towards its preference state is deducted away from the sum of disturbance. Am I on a right track?

You are right. I am still not on the right page. The helper and helpee are in each other’s feedback functions. They are controlling , via each other, for getting what they want— to be helped in the case of the helpee and to help in the case of the helper — by disturbing what each hopes or knows that the other is controlling for; the helpee hopes that the helper is controlling for helping so that his pleas will be a disturbance corrected by the helper giving help and once the helpee pleads for help the helper knows that the helpee is controlling for getting help and she can compensate for that disturbance by helping. Of course, the helpee and helper are successful in their controlling if they are actually controlling for what the other needs them to be controlling for.

Best Rick

Still haven’t got it right. The helper is in the helpers feedback function, but the healthy is not in the helpers feedback function. I leave it as an exercise for you to figure out why the difference.

Sorry Rick but this makes not much sense. Could you please repeat?

T. Eetu Pikkarainen
(Kännykästä / From mobile phone)

Rick, I guess you meant to write: “The helper is in the helpee’s feedback function, but the helpee is not in the helper’s feedback function.” Right?
I think that is wrong because they both are in each other’s feedback function. The trick is first that the helper is controlling on two levels. On the higher level she’s controlling for the satisfaction of the helpee and only in the lower level for helping (doing something to help). Secondly there is a the third phase in the process (protocol of helping), and in that phase the helpee thanks or seems happy or something like that. Here the helpee is in the feedback function of the helper. (Of course there can be many more sub phases in any realization os s protocol of helping.)

T. Eetu Pikkarainen
(Kännykästä / From mobile phone)

Yes, that’s what I meant to say. I’ve been traveling and was posting via my phone, which made several corrections that weren’t correct so I’m glad you got my meaning, Eetu.

Yes, in your example they are in each other’s feedback functions. The helper is in the helpee’s feedback function because the helpee is controlling for getting help from the helper and acts to get it by asking the helper for help. And the helpee is in the helper’s feedback function because you assumed that the helper is controlling for hearing the helpee say “thank you” and acts to get the helpee to say it by helping.

In my scenario, I assumed that the helper was controlling for helping people who ask for help so that helpee’s plea for help was a disturbance to that controlled variable that the helper corrected by helping.

So I think of scenarios where the helpee is either in the helper’s feedback function (if the helper is controlling for seeing gratitude from the helpee) of just a disturbance to a variable controlled by the helper (where the helper is controlling for helping when asked).

But I can also think of a scenario were the helper is in the helpee’s feedback function (if the helpee is controlling for getting help from the helper) but it’s hard for me to think of a sensible scenario there where helper is just a disturbance to a variable the helpee is controlling. The helpee is defined as someone who is controlling for getting help. So if the helpee asks for help from the helper is automatically in the helpee’s feedback function. The only way I can see for the helper to be just a disturbance to the helpee would be to change the variable the helpee is controlling for. For example, if the helpee is just controlling for finding out if the helper can provide help of a particular kind. So instead of just pleading for help the helpee might ask “Can you fix my drain”, and that question would just be a disturbance to the variable the helper is controlling – telling people whether he can fix certain things.

The point is that, in interactions between two (or more) people, whether a person is in the feedback function of just a disturbance to a variable controlled by the other person depends on how each person’s behavior (controlling) fits into the control loop of the other.

Best, Rick

The belief in the existence of God (collective)
Vs the belief in the current sensing and presence of God (perceptual/experiential)

Please explain. I don’t understand this at all.

Again, (I asked this already once) I can’t understand your use of the concept of disturbance. I have learned that disturbance is one or more environmental forces or effects which cause the controlled variable away from its preference state. Is this so? If it is then your scenario does not make sense. Let’s say the person A is controlling for helping other people. Then if the person B asks for help that asking is NOT a disturbance for A. A refusal of B: “Go away, I don’t need any help.” would be a disturbance for A’s controlling.

T. Eetu Pikkarainen
(Kännykästä / From mobile phone)

Are you saying that belief in the existence of God is an example of collective control while belief in the current sensing and presence of God is an example of individual control?

Beliefs are imaginations that can be disturbed by other people – for example, parents teaching the child to believe and scientists teaching the child not to – and other people can try to control these imaginings, a process known as brain washing. Is this what you mean when you say belief in the existence of God is collective control, that people try to influence or control what other people believe?

If so, I don’t see why " belief in the current sensing and presence of God" is any less an example of collective control “than belief in the existence of God”. I’ve seen people convinced (or apparently convinced) by Pentecostal preachers that what the are currently sensing/perceiving is the presence of God.

In discussions of collective control, control theory remains unchanged. Nothing is added to or taken from the theory.

What is different is the properties of the environment in which a control system operates. We like to point out that placing a ball on one of Galileo’s inclined planes is very different from placing a mouse or a pigeon on it. In a parallel manner, an environment devoid of control systems is much simpler than an environment populated by interacting control systems. Practically all PCT computer simulations model an environment devoid of other control systems. To model interactions of control systems the environment of the PCT model for each control system must include the other control systems.

An environment populated by other control systems is unpredictable in way that an unpopulated environment, subject only to physical and chemical laws, is not. The observed phenomena of control in a populated environment importantly includes ways in which the participating control systems make their behavior more predictable. This is the subject matter of Martin Taylor’s lifelong study of communication and communication protocols.

Protocols are not additions to the theory. They are additions to the list of controllable perceptions.

This also has consequences for a PCT theory of learning. The foundation of learning in PCT is reorganization when persistent error in a control loop cannot be reduced by higher-level systems controlling by means of alternative perceptions in their perceptual input — e.g. using the stairs when the elevator is busy or disabled. Important among the observable phenomena in a populated environment is learning by example and the more tightly coupled form of it, learning by instruction.

To learn by example, L (the learner) perceives that E (the exemplar) is controlling a variable V. At a higher level, L has a reference for control of V but can control it only in imagination because no lower loops are functionally connected to that output. By observing how E controls V, L controls in imagination the lower-level loops which would produce those environmental appearance aspects of L’s own behavior. If the lower loops are already familiar from use for other purposes, L is good to go. If active (non-imaginary) control results in error controlling V, reorganization may begin.

This is how children learn protocols and other controlled perceptions of how interactions are structured. These perceptions are simply public means for control of other perceptions, like stairs and elevators. They are perceptions of recurrent phenomena observable in the environment and usable to facilitate control. Other such instrumental phenomena have more durative physical properties — stairs, elevators, roads, bridges, universities, grocery stores, the groceries organized and replenished daily in them, and so on. These are like any objects in the unpopulated world of the typical PCT model, except that they exist and are organized as they are by other control systems whose activities are mostly not perceived by the modeled control system which is using them. The driver maintaining the course of the car within the boundaries of the correct lane of the road has no concern for those who maintain the road, but has some concern for those who enforce the correctness of one lane vs. the other, though the presence of oncoming traffic covers that pretty well — oncoming traffic driven by other control systems.

So phenomena first. There’s a lot of data about the populated world, data for phenomena of a different kind than those of physics and chemistry. When you’re modeling a control system in a populated environment, some collective phenomena are physical properties of variables controlled by other control systems; some are observable appearances of the behavior (control outputs) of other control systems.

1 Like