Hi Rick,
RM: Exactly, the dog controls the behavior of the sheep by disturbing a variable that the sheep control by grouping themselves closer together. Boris disagrees and says that the sheep are not being controlled by the dog, which is demonstrably wrong. Without the dog there, the gaps between the sheep would on average be quite wide.
HB :
First of all nobody can control any behavior. Neither dog or sheep. Both control their own perception.
So if I see right your »attack on me« that I’m wrong, you are trying to say that dog and sheep are controlling for the same perception : the gap between sheep and so it seems to you that dog controls the behavior of sheep. And that’s by your oppinion proof that sheeps are controlled by dog. Or in other words : because sheeps behave as dog wants, sheeps are controlled. Did I missed something ?
The problem here is that sheep could control the perception of the gap between sheep, for safety as you said. But when you say safety what does it mean. If we say just control for safety we don’t know what is really controlled. I asssume that safety could be higher order perception, composed of at least two lower level controlled perceptions : distance between ship and distance between dog and sheep. Both controlled perceptions are contributing to safety, specialy the last one.
But sheep do control at least one more life-important perception : eating grass. When they control such a perception they spread on great teritory seeking for better grass, not seeking for control of gap between them. That’s what owners of sheep wants. So owner train a dog to group sheep so not to be spread arround.
Sheeps are not controlling only for the gap between sheep, they control for their life-important perception. The owner of course also control that sheep could eat best grass. Dog and people who trained dog, do control for sheep being togetheras as much as possible to eat grass, but also not to lose any of sheep.
So I think you are right that dog control for »the gap between sheep« as that’s what owner wants. But I think you are not right that dog controls the behavior of sheep by controlling the gap among them.  Sheep can control for different perception among which is also perception of grouping together for safety. But behavior of sheep is not determined with presence of the dog, but depends from sheep’s internal control by which relevant behavior (output) for the sheep’s control of perception is established. Â
Sheep beside very important control of eating grass, control also perception for the distance to the predator (in our case it’s dog).  It’s probably some inborn survival behavior. So sheep control for the perception of distance to dog. Here is video, which show how group of sheeps are moving in accordance with distance to the dog. When dogs are at safe diastance (probably low »error« in sheep control) sheeps are not moving. When distance minimize sheep move (run) away from dog. But the distance is varying in accordance to control of sheep and dog. When sheep control differently – run away from group, than we can see that sheep in control of their perception. Â
https://www.youtube.com/watch?v=wJUVULU7GDM
So we can conclude that there are many »survivial behaviors« or sheep can control for different perceptions. One of them and very important is : the gap between dog (predator) and sheep. With output sheep will control perception of being enough far from dog. We can say that sheep controls the distance to the dog (predator) and varying distance will in some proportion (not linear) cause »errors« in control. The closer dog is, more »error« and more distance to the dog is produced, less »error« no activity or perception of eating grass is controlled.
Dog’s controlled perception is : the more the sheep are together – less »error«, the more they are appart – €“ more »error«. So again varying distance, varying »error« of dog control. Also seen in video when one sheep ran away from group. Dogs ran after it.
So both : dog and sheep are controllig what they are supposed to control.
Now Rick you can make simulation of both LCS each controling for own perception – varying perception of distance. We have only to establish, how these diferent controls with different goals are coordinated. I would say that dog run in different directions trying to change the distance to sheep so to move sheep in wanted direction.
In this way sheep is stil controlling for distance and less »error« in the direction of group of other sheep mean that is more possibility for sheep to move to the group. So in this case sheep is tending to encrease the gap to the dog and dog somehow enable that. But this is not always the case.
So I concluded that dog behavior is not always implaying sheep behavior to increase distance to the dog only by running away. It can be also different. In next video you can see that sheeps choose another behavior, which has something to do with the dog, but in opposite direction. Instead of increasing the gap to the dog, with running away, they choose the control of perception of moving the dog away. So they attack dog. Â
https://www.youtube.com/watch?v=RY_7U56R0AM
Your term Rick about controling other living being somehow sounds to me as S-R logic. Dog behavior (stimulus, disturbance) to sheep perceptual input, and then directly to sheep behavior is somehow determined sheep behavior by it’s input (environment). It’s denying Bill’s aymetry of control : LCS can control environment, and environment can’t control LCS.
By your control of behavior of others can happen in the opposite  way.  So it sounds like absolutely determined behavior by environment : sheep input (dogs behavior) - sheep output (induced behavior by dog behavior). Nothing in between. No process inside »controlled« LCS counts, no references, no reorganization, just stream from input to output (control of others).
LCS are already controlling, so nobody can overtake that control. Sheep all the time control their perception. Â Dogs all the time control their percpetion.
I didn’t find any example to show how sheep can attack alone sheep dog, not just ran away form the dog, but I found this video, which shows how sheep attacked wolf, when he comes to close.
https://www.youtube.com/watch?v=F29dMtxXCkE
If we think that sheep are inocent little animals harmly eating grass, we can made a hudge mistake : here is video which show how nasty can be sheep attack.
https://www.youtube.com/watch?v=GJrlo2VuUR8
Maybe this video will help to see, how sheep control the distance to dog (predator) :
https://www.youtube.com/watch?v=uXkHDiuadTI
And maybe one more video which shows that dog is not controlling the behavior of the ram.
https://www.youtube.com/watch?v=cluHOW_EnAQ
And maybe the last video that shows that perceptual control is choosen in the ram not in environment and can be very unpredictable and dangerous.
https://www.youtube.com/watch?v=6jgjk3DsN7c
All in all I think that there is enough proof that dogs are not controlling behavior of the sheep, but sheep control it’s own perception that it chooses  and dog control it’s perception that it chooses. But neither of them can control behavior. Nor own, neither others. As far as I understand there is no such a thing in PCT. Â
Best,
Boris
···
From: csgnet-request@lists.illinois.edu [mailto:csgnet-request@lists.illinois.edu] On Behalf Of Richard Marken
Sent: Tuesday, September 09, 2014 2:45 AM
To: csgnet@lists.illinois.edu
Subject: Re: Sheepdog and flock behavior
[From Rick Marken (2014.09.08.1545)]
Kent McClelland (2014.09.07.2050)
KM: I see that my name has come up in this exchange, so I’ll see if I can say anything to help move it along. As I look at what you’ve written, I don’t see much in either person’s position to disagree with, but you seem to be at an impasse. The difficulty seems to be semantic, revolving around what it means to control another person’s behavior, which Rick says can be done and Boris says can’t. In my view, you’re both partly right.
RM: I think it’s more than semantic. I think the difficulty turns on knowing what control is, in fact, not in theory. Once you know what control is then there is no controversy at all: behavior can be controlled. That’s what we are seeing with the sheepdogs (the dogs controlling the gaps between the sheep) and with the E and S in the rubber band demo (E controlling the finger position of S). A variable (the gap, the finger position) is being kept in a pre-selected state (0 gap, finger on target dot), protected from disturbance (autonomously produced changes in the closeness of the sheep to one another, the closeness of the finger to the target dot) by varying actions appropriately (moving towards or away from the sheep, increasing or reducing the pull on the rubber band) .
KM: The problem, as I see it, is that a person’s behavior (or that of any living control system with hierarchical control of perceptions) is never just one thing. A person controls lots of different perceptions simultaneously, at different perceptual levels. And when a person observes another person’s physical actions, the observer can see “behaviors” at a lot of different perceptual levels.
RM: Yes, but this is true of all of our experience, not just our experience of other people’s behavior. We control perceptual variables of many different types (hypothetically 11 different types of perceptual variable). I don’t see what this has to do with whether or not people can control other people’s behavior.
KM: Because a person controls lots of different things at once, under the right circumstances another person can control SOME of these behaviors (as the controller perceives them), but not ALL of them, at least not all of them simultaneously. The trick in controlling a particular behavior by another person (call it behavior A) is to get the person to focus on controlling behavior B, and then to remove all the other ways for the person to keep controlling behavior B except by doing behavior A.
RM: Yes, this is all I claimed was true: people can control the behavior of other people. Of course, they can only control some of these behaviors – the one’s that correct for disturbances to the variable the controllee is controlling. This is why I gave the example of controlling the position of S’s finger in the rubber band demo. Since S must vary the position of the finger in order to compensate for E- produced disturbances to the position of the knot, the variable S is controlling, E can control S’s finger position by disturbing the position of the knot. S’s finger position is the only aspect of S’s behavior that E can control in this situation. But that aspect of S’s behavior can certainly be controlled.
KM: That’s the classic pattern for manipulation of a person’s behavior.
RM: Right! This is how you can manipulate (control) another person’s behavior.
KM: The pattern also applies to the dog and sheep example.
RM: Exactly, the dog controls the behavior of the sheep by disturbing a variable that the sheep control by grouping themselves closer together. Boris disagrees and says that the sheep are not being controlled by the dog, which is demonstrably wrong. Without the dog there, the gaps between the sheep would on average be quite wide.
KM: Thus, the control of one person’s behavior by another person can and does happen
RM: My point exactly.
RM: One example: When I give a cashier $10 for a $5 item I am controlling for the cashier giving me $5 change. The cashier is also controlling for me paying $5 for the item. Typically, we both get the behaviors from each other than we want: a nice example of mutual control, and no one gets hurt (or oppressed). This kind of agreed on mutual control is the basis of civilization.
KM: Martin Taylor, in the chapter he’s writing for the LCS IV book edited by Warren Mansell (a preliminary draft of which I’ve been lucky enough to see), calls this kind of exchange transaction a “protocol,”
RM: I looked at Martin’s reply and what he describes as a “protocol” is not the same as the mutual controlling that occurs in the customer/cashier interaction.
KM: and, if I understand his concept correctly, protocols provide a way for two people acting together to control two different perceptions. The side-effects of one person’s behavior in controlling his or her own perception allow the other person to control a different perception (and vice-versa). In your example, the cashier controls the perception of completing a sale by getting your money and giving you the item and $5 change, while you control the perception of buying the item by handing over the money and getting the item and change back.
RM: The difference between the customer/cashier interaction and the interaction between two control systems described by Martin is that there is no control of one control system by another in the situation Martin describes. In the customer/cashier situation, successful control by each control system requires successful control of each control system’s behavior by the other control system. In Martin’s “protocol”, shown below, Alan’s ability to control PA does not depend on controlling Beth’s behavior and vice versa. The actions Alan uses to control PA are a disturbance to PB which Beth will oppose along with any other disturbances to PB, and vice versa. But Alan’s ability to control PA does not depend on his controlling Beth’s behavior. INdeed, Alan can control PA just fine whether Beth is controlling PB or not.
![Inline image 1]()
This is quite different from the customer/cashier situation. The customer is controlling for paying the exact cost for the product; the cashier is controlling for collecting the exact payment. If the product costs $5 and the customer hands $10 to the cashier, the customer will achieve the “pay the exact cost” goal only if the cashier returns the exact change. So the customer is controlling for the cashier returning $5. The cashier is also controlling for collecting the cost of the customer’s item and so will allow the customer to leave with the product only after payment has been received. If the cashier were not controlling for getting only the exact amount for the product the customer would not successfully control for getting the correct change behavior from the cashier; if the customer were not controlling for paying for the product the cashier would not successfully control for getting paid by the customer.
KM: I don’t think I would describe this exchange situation as control of another person’s behavior
RM: Apparently Bob Hintz doesn’t think so either. But, again, I think it can be easily demonstrated that both the customer and cashier are, in fact, controlling each other’s behavior by introducing disturbances to the hypothetical behavior variables that are being controlled. For example, have the cashier give the customer $4 rather $5 dollars in change – a disturbance to the customer’s perception of the cashier returning the correct change. I think you will see the customer take action – like arguing with the cashier – to bring the cashier’s behavior to the desired state: handing over the $5 change. Similarly, have the customer hand the cashier $4 rather than $5. I think you will see the cashier take action – like calling over the manager – to bring the customer’s behavior to the desired state: pay for the product. So the customer is controlling for the “returning correct change” behavior of the cashier and the cashier is controlling for the “paying for the product” behavior of the customer.
KM: unless you were buying the item from someone who had a monopoly, and the only way for you to get the item (behavior B) would be to pay the asking price (behavior A). Even then, you could refuse to buy the item or just get something else, unless the item were something you couldn’t live without.
RM: I hope my little explanation shows that this is not necessary. The customer and cashier are controlling each other’s behavior but it doesn’t look like they are controlling each other’s behavior (and they certainly don’t feel like they are controlling each other) until one or the other doesn’t behave as desired. This is the way it is with all our controlling, though, isn’t it? We don’t feel like we are controlling when we walk down the street or talk on the phone or swim in the lake – that is, we don’t notice our controlling until we start to lose it .Skillful control does’t feel like anything special; it just feels like we are doing things. And this is true of skillful controlling of our own limbs, of the physical consequences of limb movements (like walking and swimming) and interpersonal consequences of these limb movements (like getting change from a cashier).
RM: I think I understand why people don’t want to believe that they can be controlled and why they do want to believe that they never control other people. It fits with our idea that we are “free” and “good”. But I think it’s better to recognize when we are controlling, especially when we are controlling (and being controlled by) other control systems. Understanding how to recognize this is, I think, one of the most important things to learn from PCT. It’s only when you know that you are controlling (and being controlled) that you can adjust your own controlling to reduce inter (and intra) personal conflict and work to organize society in a way that allows controllers to do their controlling with the least amount of conflict and the greatest amount of cooperation with other controllers. Making believe that we can’t be controlled or that we are not controlled by and/or don’t control other controllers is not the road to a better world, I think.
Best
Rick
Hope these thoughts help you to clarify the questions you were discussing.
Best,
Kent
On Sep 7, 2014, at 12:54 PM, Richard Marken wrote:
[From Rick Marken (2014.09.07.1100)]
boris.hartman@masicom.net wrote:
BH : I don’t think that dogs were controlling gaps by »controlling sheeps behavior«. Sheeps were controlling their behavior on the bases of disturbances dog produced to sheep perceptual control.
RM: Yes, the sheep were controlling a perception (I called it safety) to which the approach of the dog was a disturbance. The sheep compensated for this disturbance by moving closer to other sheep. This is the behavior that the dog wanted to see (were controlling for) because it closed the gaps that were a disturbance to the perception the dog was controlling. So the dog was able to control the behavior of the sheep, getting them to bunch closer together, because the sheep were controlling for safety by bunching together.
RM : So any model of this behavior would have to model the dogs’ “gap” control system and the sheep’s safety control system. Such a simulation would show that control systems can both control and be controlled.
BH : Well this is a good idea to make simultaneous simulation. Did you try it ?
RM: Not yet.
BH: But this is the n-th time that we are trying to solve the problem how LCS can be “controlled”. In the book : Making sense of behavior Bill talks about “attempts of control” not “control”…
BH: Maybe Kent could explain what’s wrong with your position about »LCS can be controlled".
RM: Yes, I think that’s a good idea, though I would be surprised if Kent thought that there is something wrong with my “position”. My position is simply that people can clearly control the behavior other people and their own behavior can be controlled as well. This is an observable fact that is explained by PCT. I leave the explanation as an exercise.
RM : It can also be easily demonstrated with humans using the rubber band demo; the E in this experiment can control the finger position of the S once S has agreed to control the position of the knot.
BH : Yes. Watch your wording in rubber band demo. “Once S has agreed” or as Kent said “S chooses”. . So it’s obviously that S agreed to be controlled.
RM: Not quite. S has agreed to control a particular perception (the location of the knot relative to the dot, in this case). S has not agreed to be controlled; indeed, S is typically unaware of being controlled after agreeing to control that perception.
HB : Aplying distrubance to a »controlled variable« does not mean that control is established. It seems that Kent and I agree that disturbances are aplyed to perceptual control and target person chooses.
RM: I would be interested to hear what Kent has to say about that. I would be very surprised if he agreed with your analysis.
RM : Nor does control theory show that controlling organisms (particularly humans) is necessarily a bad thing to do.
BH : Now you are exaggerating Rick. You can’t control organisms. I’d really like to see how you are doing that in everyday life ?
RM: One example: When I give a cashier $10 for a $5 item I am controlling for the cashier giving me $5 change. The cashier is also controlling for me paying $5 for the item. Typically, we both get the behaviors from each other than we want: a nice example of mutual control, and no one gets hurt (or oppressed). This kind of agreed on mutual control is the basis of civilization.
RM : What control theory does show is that arbitrary control, particularly of humans by other humans, will almost certainly lead to conflict. Arbitrary control is exerting control without considering the fact that living control systems are controlling many variables at the same time and when you arbitrarily decide to have a person do something (by disturbing a controlling variable) what you have them do may conflict with other things they are controlling.
HB : It makes some sense, but I don’t understand what you meant by »arbitrary control« ?
RM: It means controlling without taking a persons wants and needs (the references for the many different perceptions that the person is controlling for) into account. In non-PCT terms, it is controlling another person without respecting that person’s humanity (and autonomy).
BH: If you are thinking like this one : »Controlle behavior match my wanted perception of behvior and thus it is »controlled« it’s wrong.
RM: That’s exactly how I am thinking. And I think you are wrong about this being wrong.
BH: You have PCT which helps you understand what’s happening inside organisms. Human control and behavior are quite unpredictable, because the references are formed inside organism, where most of control is done.
RM: Once you have correctly identified a controlled variable behavior (the actions that protect that variable from disturbance) can be predicted with very high accuracy. See the “Basic Control demo” at http://www.mindreadings.com/ControlDemo/BasicTrack.html to see what I mean.
BH: Just observable »facts« of behavioral event are not prove that people control each other. Although it’s not excluded. The controller’s behavior is just a disturbance to controlee perceptual control. From what happens in controlee comparator (error) will probably decide whether controlee behavior will resemble to something controller wanted or not. But never behavior of controlee will be just exact »copy« of controller’s wanted behavior, because controlee is in control.
RM: It may not be an “exact” copy of what the controller wants, but it can be very close (like within 1% of the desired value).
RM: Anyway, this behavior – managing flocks – is a very interesting demonstration of controlling a perception (of gaps between sheep) via disturbance of a perception being controlled by the control systems that are being controlled.
BH: This one makes some PCT sense. J
RM: Well, that’s progress!
Best
Rick
But as usual I’m living space for not understanding something right. And as always, sorry for my language.
Best,
Boris
From: csgnet-request@lists.illinois.edu [mailto:csgnet-request@lists.illinois.edu] On Behalf Of Richard Marken
Sent: Sunday, August 31, 2014 3:23 AM
To: csgnet@lists.illinois.edu
Subject: Re: Sheepdog and flock behavior
[From Rick Marken (2014.08.30.1820)]
[From Bruce Abbott (2014.08.30.0820 EDT)]
Researchers have investigated how sheepdogs manage their flocks by fitting both the dogs and the sheep with highly accurate GPS devices, allowing the researchers to track their movements. The research is presented in a BBC article at http://www.bbc.com/news/science-environment-28936251 .
RM: Great find, Bruce! There are two things that make this report particularly interesting to me. The first is the quote by Andrew King, which is a great description of the PCT approach to research. To paraphrase King: in order to understand the behavior of organisms you have to try to look at their behavior from the point of view of the organism (behaving system) Itself. I make this same point in the chapter on “Looking at Behavior through Control Theory Glasses” in “Doing Research on Purpose” (why hasn’t that become a best seller yet?). I do it in the section on trying to understand the apparent “fixed action pattern” of the greylag goose. The goose is seen to continue to make the movements that would pull an egg back into its nest even when the egg is no longer present. Looking at an organism’s behavior from the organism’s perspective helps you come up with good ideas about what perceptual variables the organism is controlling (the first step in the Test for the Controlled Variable). By looking at the goose’s egg rolling behavior from the goose’s perspective I was able to come up with the hypothesis that the goose is trying to control the pressure of the egg against the back of its bill and when the egg is removed the continued efforts to move the non-existent egg into the nest (the apparent fixed action pattern) is just the efforts of the pressure control system to restore the pressure of the egg against the back of the bill. By looking at the sheepdogs herding behavior from the dogs’ perspective King came up with the reasonable hypothesis that the dogs were controlling their perception of the gaps between patches of white (the sheep), trying to keep those gaps at zero.
RM: The other thing that’s interesting about this report is that the dogs were clearly controlling their perception of the gaps by controlling the behavior of the sheep. And they did this by becoming a disturbance to a perception that the sheep control by getting closer to other sheep: the perception of safety. So any model of this behavior would have to model the dogs’ “gap” control system and the sheep’s safety control system. Such a simulation would show that control systems can both control and be controlled. This kind of simulation would help dispel what I think is a common misconception about the control theory model of organisms – particularly humans. It is a misconception that I myself labored under until just a few years ago. It is the idea that because organisms are autonomous control systems – autonomous in the sense that they set their own references for the states of their own perception – they cannot be controlled. But autonomous control systems can be controlled, as is demonstrated by the sheepdogs controlling the sheep. It can also be easily demonstrated with humans using the rubber band demo; the E in this experiment can control the finger position of the S once S has agreed to control the position of the knot. And thanks to Bruce Abbott I demonstrated to myself that E can still exert this control even if S continuously – and autonomously – varies his or her reference for the position of the knot.
RM: Control theory doesn’t say that organisms (particularly humans) can’t be controlled; indeed, it shows that they can be controlled, mainly by disturbance to a controlled variable. Nor does control theory show that controlling organisms (particularly humans) is necessarily a bad thing to do. What control theory does show is that arbitrary control, particularly of humans by other humans, will almost certainly lead to conflict. Arbitrary control is exerting control without considering the fact that living control systems are controlling many variables at the same time and when you arbitrarily decide to have a person do something (by disturbing a controlling variable) what you have them do may conflict with other things they are controlling. So to take an example that Bill used (somewhere), if E decides to place S’s finger against a hot soldering iron while controlling Ss finger position in the rubber band game that will clearly lconflict with another goal S has (not getting burned). Non-arbitrary control is control that is done with the consent (often implicit but sometimes explicit) of the would-be controllee. Non- arbitrary control is, I think, essential when humans control other humans. To see why, think about what happens when people are arbitrarily controlled (herded) in the same way that the sheep were. Hint: They don’t like it. Why do you think not?
RM: Anyway, this behavior – managing flocks – is a very interesting demonstration of controlling a perception (of gaps between sheep) via disturbance of a perception being controlled by the control systems that are being controlled.
Best
Rick
The researchers found that the behavior could be described by two simple rules, but more interesting from a PCT perspective, they found that to understand the behavior they needed to view the action from the animal’s perspectives. According to researcher Dr. Andrew King:
“At the beginning we had lots of different ideas. We started out looking from a birds eye view, but then we realised we needed to see what the dog sees. It sees white, fluffy things. If there are gaps between them or the gaps get bigger, the dogs needs to bring them together.”
According to Dr King, sheepdogs are making the most of the “selfish herd theory” to bring the animals close together and move them where they want.
“One of the things that sheep are really good at is responding to a threat by working with their neighbours. It’s the selfish herd theory: put something between the threat and you. Individuals try to minimise the chance of anything happening to them, so they move towards the centre of a group.”
The article continues as follows:
A colleague, Dr Daniel Strombom from Uppsala University in Sweden, used the GPS data from the collars to develop computer simulations. This enabled them to develop a mathematical shepherding model.
The algorithm displays the same weaving pattern exhibited by sheepdogs. It helps to solve what has been called the ‘the shepherding problem’: how one agent can control a large number of unwilling agents.
I’d be interested to know whether the computer algorithm models each individual’s control systems or operates by some other method.
Bruce
–
Richard S. Marken, Ph.D.
Author of Doing Research on Purpose.
Now available from Amazon or Barnes & Noble
–
Richard S. Marken, Ph.D.
Author of Doing Research on Purpose.
Now available from Amazon or Barnes & Noble
–
Richard S. Marken, Ph.D.
Author of Doing Research on Purpose.
Now available from Amazon or Barnes & Noble
