[philip 2014 9/10 6:26pm]

The first thing I wanted to discuss was a question about what Bill wrote in LCSIII on page 91 while discussing the hierarchical control model of the mass and spring. On the last line of the page, Bill says, “It’s important to realize that the force adjustments, while logical and reasonable, are not generated either by logic or by reason; they arise naturally out of the two-level arrangement of control systems”. Why would Bill say this? Looking at the figures plotting position, velocity and force, there is plenty of information to discuss about logical patterns in changing force, changing position, and changing velocity. But what Bill said makes no sense…how could something be logical if it’s not generated by logic? After all, Bill logically programmed the simulation - there must be an intuitive reason for why the math adds up.

There is unrelenting discussion of the concept of “control of behavior” vs “control of perception”.

Given #1: Behavior is defined as the control (of perception). Thus control (of behavior) would be interpreted as:

• the control (of a perception (of [behavior])). or,

• the control (of a perception (of [the control of perception])). or,

• the control (of a perception (of the control (of a perception (of perception))))***.

If behavior is defined as a controlled perceptual input (i.e. an input matched to a reference), then

there are two ways to interpret the meaning of “the control of behavior”. The first is the direct perceptual control of the reference value of a behaving system.

As always, there are benevolent and maleficent examples.

Benevolent example: you’re a general contractor giving an employed subcontractor instructions about how to complete a task (basically, you’re communicating your reference value). Then you ask him to repeat the instructions back and compare his ‘projected’ reference value to your’s. Bob Hintz hinted at the importance of understanding the concept of language and reference information sharing. Notice that if you or another person explicitly states your reference value, then this is a measure of reducing uncertainty in the outcome of behavior. Only through language can somebody directly control a perception of another’s reference value - thus ‘controlling’ their behavior (in the sense that you’re preventing unintended intentions by matching a perceptual representation of a reference value to a reference value for the reference value).

Maleficent example: you’re torturing a messenger harboring a national secret. you can assume the messenger doesn’t want to give you the information. therefore, you apply pain until the consciousness of the messenger is saturated and he can no longer maintain his concentration. In this special case, the reference assumes a binary value - speak or don’t speak. Of course, the messenger could, in theory, be tortured to death without uttering a word. This would be the most vivid example if we put ourselves in the skin of the person getting tortured and imagine ourselves controlling a perception of not speaking. The torturer would initially try to interrogate us and provide logical reasons why giving up the information would serve our goals at a higher level of the hierarchy (for example, if we’re keeping the secret to protect our family first, and country second, then the torturer would threaten to go after family - standard procedures, we might imagine). If this fails, then the ‘logic’ turns to a lower level of the hierarchy: the application of pain - for instance, by altering a joint angle beyond its natural limit. Obviously, we usually control our joint angles to control higher level perceptions, but in this case the higher level reference is artificially altered to allow for a joint angle to match its reference.

So…

The control of behavior, as one may describe it, is the control of the reference value of the behaving system itself, which, by Bill’s definition, is NOT an input from the environment, in contrast to as it is for an output controller. If we choose to step outside the bounds of PCT and insist that we can apply “disturbances” to another person’s reference value, then I think we shall lose the concentration of the entire PCT crowd. If we adhere to PCT according to Bill, there is no logical method of proving that the behavior of one PCS can be controlled by the behavior of another. Unless, of course, we consider the first example above, in which a control system is establishing a direct perception of the reference value (through language) and comparing it with a referential reference. This is more like a student-teacher relationship than a random_person_in_society–random_person_in_society relationship.

This brings us to the discussion of protocols, specifically, Figure 14 referenced by Martin Taylor 3 days ago in the thread, “Protocols”. The figure caption reads:

Alan and Beth each control some perception, oblivious to the presence of the other. A side-effect of Alan’s control disturbs the perception Beth is controlling, and a side-effect of Beth’s control disturbs a perception Alan is controlling. If those side-effects happen to be beneficial, one or both of them might change their output action so that they deliberately disturb the other.

Ok, this works completely fine for a baby crying. But as soon as we talk about the side-effects of behavior happening to be ‘beneficial’ we have to immediately realize the error. The mere usage of the phrase “side-effects of behavior happening to be beneficial” is loaded. Side-effects are a part of output. As PCT says, output happens to control references, and that’s ALL we care about with respect to the output (i.e. we don’t care about how it affects others’ references). To analyze the feedback effect of side-effects as in Figure 14 would be similar to saying that if random ‘beneficial’ disturbances (side-effects of another system), for any reason at all, cause a system’s perception to approach a reference, then the behavior of this system might be reorganized to specifically elicit more beneficial side effects from the other system. In my opinion, this is absurd. Co-reorganization is perfectly fine for a baby crying because of the lack of language capabilities. But it would take us dangerously astray to consider how these interactions might play out with adults in society. In his example, Martin said (paraphrased):

MT (part 1 of example):

Actions have side-effects which could influence someone else’s controlled perception. A farmer who is controlling for having more money opts to grow crops and sell them to some agent, and that’s the end of it. A side effect of his growing crops is that food appears on supermarket shelves.

PY:

I must insist that you reconsider your labeling of this as a “side-effect”. You are confusing the distinction between a “side-effect” and an “uncontrolled perception”. I believe what you are trying to say is that the farmer could not care less what the agent does with his crops, as long as he gets the money for it. Of course. But surely, the food appearing on the supermarket shelf must be a controlled perception of the agent who bought the crops, as opposed to a direct side-effect of the farmer’s actions. In fact, I’m sure the farmer wants the food to appear on the supermarket shelf, because if the crops aren’t purchased then there’ll be no money for the agent (or supermarket) to spend on a second round of crops. So in reality, the PCS-farmer should logically be controlling this perception too (for example, by referring people to the supermarket). Back to Martin

MT (part 2 of example):

At the same time as the farmer is controlling his perception of making income, a woman is controlling a perception of her family being not hungry, by giving money to the supermarket in exchange for food. A side-effect of her control is that the farmer gets a little more money. The woman and the farmer neither know nor care that the other exists.

PY:

The woman wants the food. Obviously, the woman doesn’t care about how much money the farmer makes, and so this variable is an uncontrolled perception on her part - but again, NOT a side effect of her behavior. We should realize that neither the farmer’s effect on the appearance of food in the store, nor the woman’s effect on the farmer’s pay is a side effect of any behavior. Rather, both of these variables are actually controlled perceptions (of the agent/supermarket) and nobody else. The reason that this example generates such a calamity is because you are propagating the “I don’t care (uncontrolled perception) attitude” of the farmer and woman through the behavior of the intermediary agent (as though the agent is also sharing these uncontrolled perceptions). Of course, this is false - the agent is specifically controlling the perceptions which the farmer and woman aren’t. The agent indeed DOES care about the existence of both the farmer and the woman, and if there is any talk about side-effects of uncontrolled perceptions on the part of woman or farmer, it could not survive being “passed through” the agent.

Given #2: behavior is defined as belonging only to a perceptual control system. Obviously, therefore, the position of a mouse on a computer screen is not “behavior” because the computer which outputs the position of the mouse does not control a perception of the effect of the output. Thus, it is highly improper to think you are controlling the perception of the behavior of the mouse. If, however, we take a look at the code of a computer program which controls the position of a mouse, then we’ll realize that that the program gets a reading of the mouse position and moves it toward the reference. Obviously, you can control the position of the mouse by controlling the reference value in this program. Now, take a look at the 3rd interpretation of the meaning of control of behavior I stated above:

• the control (of a perception (of the control (of a perception (of perception))))***

This one, if it makes any logical sense, is especially hard to wrap your head around. Step 1: first, there’s a perception of a perception being controlled. Think of looking at the computer code of the mouse-controlling program and seeing it controls the x and y coordinates of mouse position to a particular reference value. Now, step 2: the perception of the perception being controlled is itself being controlled. Think of rewriting the computer code you’re looking at, altering the reference value or type of the variable being controlled by the computer program. For instance, you either change the reference position of the mouse in the program, or have the program control something as a function of this position, or even something else entirely. This is the control of behavior - corresponding simply to the control of a direct perception of the reference value itself.

makes fine sense for a computer. But could you imagine doing it to a person?

Imagine: You walk up to the person and temporarily occupy their entire consciousness - feeling all of their thoughts and emotions and gaining information about variables they’ve been controlling for years and years. In short, you become the other person in the same manner as you are yourself. This would be equivalent to possession.

There is one final (advanced) method of controlling behavior. But this will be the topic of a further discussion.

Regarding the third interpretation of the definition of behavior presented, I really should have written,

• the control (of a perception (of the control (of a perception (of the control (of perception)))))***,

• the control (of a perception (of the control (of a perception (of perception))))***.

Notice the difference in the nature of what is controlled at the “end”. Rather than referring to an indeterminate “perception (of a perception)” to control, I have instead referred to a perception of the actual definition of the control of perception, which is defined as behavior. What seems to be the import is that a controlled perception IS behavior, and moreover, the perception we are all trying to control is the perception of the perceptual control system itself. A perception of a perceptual control system is controlled by the the control of perception. The recursive definition naturally follows from the apparent repetition in the language. Now, common sense entails that if you want to feel in control of anything, then you should definitely be looking at something you can control. Well sometimes people ask, what if you don’t want to control anything? What if you didn’t want to control your perceptions? …what if you didn’t want your behavior to have a purpose? But it’s not a choice - it’s impossible. Anything you do amounts to nothing more and nothing less than a controlled perception. Even if you don’t want to be alive, there’s no way out. We are slaves to a conscious simultaneously reorganizing experience.

Now, I expounded on the first definition in regards to coding/bugging a computer. I stressed a “perceiving of the perception of the perception” in the source code of a computer perceptual control system program. However, we should refer, more completely, to “a perception of the perception of control.” To understand the “goal” of reorganization, we look at page 130 of LCSIII and see that, as computer programmers, we are trying to understand and control a variable which isn’t seen in the source code itself, but is generated during the runtime of the program.

···

On Wed, Sep 10, 2014 at 9:23 PM, PHILIP JERAIR YERANOSIAN pyeranos@ucla.edu wrote:

[philip 2014 9/10 6:26pm]

The first thing I wanted to discuss was a question about what Bill wrote in LCSIII on page 91 while discussing the hierarchical control model of the mass and spring. On the last line of the page, Bill says, “It’s important to realize that the force adjustments, while logical and reasonable, are not generated either by logic or by reason; they arise naturally out of the two-level arrangement of control systems”. Why would Bill say this? Looking at the figures plotting position, velocity and force, there is plenty of information to discuss about logical patterns in changing force, changing position, and changing velocity. But what Bill said makes no sense…how could something be logical if it’s not generated by logic? After all, Bill logically programmed the simulation - there must be an intuitive reason for why the math adds up.

There is unrelenting discussion of the concept of “control of behavior” vs “control of perception”.

Given #1: Behavior is defined as the control (of perception). Thus control (of behavior) would be interpreted as:

• the control (of a perception (of [behavior])). or,
• the control (of a perception (of [the control of perception])). or,
• the control (of a perception (of the control (of a perception (of perception))))***.

If behavior is defined as a controlled perceptual input (i.e. an input matched to a reference), then

there are two ways to interpret the meaning of “the control of behavior”. The first is the direct perceptual control of the reference value of a behaving system.

As always, there are benevolent and maleficent examples.

Benevolent example: you’re a general contractor giving an employed subcontractor instructions about how to complete a task (basically, you’re communicating your reference value). Then you ask him to repeat the instructions back and compare his ‘projected’ reference value to your’s. Bob Hintz hinted at the importance of understanding the concept of language and reference information sharing. Notice that if you or another person explicitly states your reference value, then this is a measure of reducing uncertainty in the outcome of behavior. Only through language can somebody directly control a perception of another’s reference value - thus ‘controlling’ their behavior (in the sense that you’re preventing unintended intentions by matching a perceptual representation of a reference value to a reference value for the reference value).

Maleficent example: you’re torturing a messenger harboring a national secret. you can assume the messenger doesn’t want to give you the information. therefore, you apply pain until the consciousness of the messenger is saturated and he can no longer maintain his concentration. In this special case, the reference assumes a binary value - speak or don’t speak. Of course, the messenger could, in theory, be tortured to death without uttering a word. This would be the most vivid example if we put ourselves in the skin of the person getting tortured and imagine ourselves controlling a perception of not speaking. The torturer would initially try to interrogate us and provide logical reasons why giving up the information would serve our goals at a higher level of the hierarchy (for example, if we’re keeping the secret to protect our family first, and country second, then the torturer would threaten to go after family - standard procedures, we might imagine). If this fails, then the ‘logic’ turns to a lower level of the hierarchy: the application of pain - for instance, by altering a joint angle beyond its natural limit. Obviously, we usually control our joint angles to control higher level perceptions, but in this case the higher level reference is artificially altered to allow for a joint angle to match its reference.

So…

The control of behavior, as one may describe it, is the control of the reference value of the behaving system itself, which, by Bill’s definition, is NOT an input from the environment, in contrast to as it is for an output controller. If we choose to step outside the bounds of PCT and insist that we can apply “disturbances” to another person’s reference value, then I think we shall lose the concentration of the entire PCT crowd. If we adhere to PCT according to Bill, there is no logical method of proving that the behavior of one PCS can be controlled by the behavior of another. Unless, of course, we consider the first example above, in which a control system is establishing a direct perception of the reference value (through language) and comparing it with a referential reference. This is more like a student-teacher relationship than a random_person_in_society–random_person_in_society relationship.

This brings us to the discussion of protocols, specifically, Figure 14 referenced by Martin Taylor 3 days ago in the thread, “Protocols”. The figure caption reads:

Alan and Beth each control some perception, oblivious to the presence of the other. A side-effect of Alan’s control disturbs the perception Beth is controlling, and a side-effect of Beth’s control disturbs a perception Alan is controlling. If those side-effects happen to be beneficial, one or both of them might change their output action so that they deliberately disturb the other.

Ok, this works completely fine for a baby crying. But as soon as we talk about the side-effects of behavior happening to be ‘beneficial’ we have to immediately realize the error. The mere usage of the phrase “side-effects of behavior happening to be beneficial” is loaded. Side-effects are a part of output. As PCT says, output happens to control references, and that’s ALL we care about with respect to the output (i.e. we don’t care about how it affects others’ references). To analyze the feedback effect of side-effects as in Figure 14 would be similar to saying that if random ‘beneficial’ disturbances (side-effects of another system), for any reason at all, cause a system’s perception to approach a reference, then the behavior of this system might be reorganized to specifically elicit more beneficial side effects from the other system. In my opinion, this is absurd. Co-reorganization is perfectly fine for a baby crying because of the lack of language capabilities. But it would take us dangerously astray to consider how these interactions might play out with adults in society. In his example, Martin said (paraphrased):

MT (part 1 of example):

Actions have side-effects which could influence someone else’s controlled perception. A farmer who is controlling for having more money opts to grow crops and sell them to some agent, and that’s the end of it. A side effect of his growing crops is that food appears on supermarket shelves.

PY:

I must insist that you reconsider your labeling of this as a “side-effect”. You are confusing the distinction between a “side-effect” and an “uncontrolled perception”. I believe what you are trying to say is that the farmer could not care less what the agent does with his crops, as long as he gets the money for it. Of course. But surely, the food appearing on the supermarket shelf must be a controlled perception of the agent who bought the crops, as opposed to a direct side-effect of the farmer’s actions. In fact, I’m sure the farmer wants the food to appear on the supermarket shelf, because if the crops aren’t purchased then there’ll be no money for the agent (or supermarket) to spend on a second round of crops. So in reality, the PCS-farmer should logically be controlling this perception too (for example, by referring people to the supermarket). Back to Martin

MT (part 2 of example):

At the same time as the farmer is controlling his perception of making income, a woman is controlling a perception of her family being not hungry, by giving money to the supermarket in exchange for food. A side-effect of her control is that the farmer gets a little more money. The woman and the farmer neither know nor care that the other exists.

PY:

The woman wants the food. Obviously, the woman doesn’t care about how much money the farmer makes, and so this variable is an uncontrolled perception on her part - but again, NOT a side effect of her behavior. We should realize that neither the farmer’s effect on the appearance of food in the store, nor the woman’s effect on the farmer’s pay is a side effect of any behavior. Rather, both of these variables are actually controlled perceptions (of the agent/supermarket) and nobody else. The reason that this example generates such a calamity is because you are propagating the “I don’t care (uncontrolled perception) attitude” of the farmer and woman through the behavior of the intermediary agent (as though the agent is also sharing these uncontrolled perceptions). Of course, this is false - the agent is specifically controlling the perceptions which the farmer and woman aren’t. The agent indeed DOES care about the existence of both the farmer and the woman, and if there is any talk about side-effects of uncontrolled perceptions on the part of woman or farmer, it could not survive being “passed through” the agent.

Given #2: behavior is defined as belonging only to a perceptual control system. Obviously, therefore, the position of a mouse on a computer screen is not “behavior” because the computer which outputs the position of the mouse does not control a perception of the effect of the output. Thus, it is highly improper to think you are controlling the perception of the behavior of the mouse. If, however, we take a look at the code of a computer program which controls the position of a mouse, then we’ll realize that that the program gets a reading of the mouse position and moves it toward the reference. Obviously, you can control the position of the mouse by controlling the reference value in this program. Now, take a look at the 3rd interpretation of the meaning of control of behavior I stated above:

• the control (of a perception (of the control (of a perception (of perception))))***

This one, if it makes any logical sense, is especially hard to wrap your head around. Step 1: first, there’s a perception of a perception being controlled. Think of looking at the computer code of the mouse-controlling program and seeing it controls the x and y coordinates of mouse position to a particular reference value. Now, step 2: the perception of the perception being controlled is itself being controlled. Think of rewriting the computer code you’re looking at, altering the reference value or type of the variable being controlled by the computer program. For instance, you either change the reference position of the mouse in the program, or have the program control something as a function of this position, or even something else entirely. This is the control of behavior - corresponding simply to the control of a direct perception of the reference value itself.

makes fine sense for a computer. But could you imagine doing it to a person?

Imagine: You walk up to the person and temporarily occupy their entire consciousness - feeling all of their thoughts and emotions and gaining information about variables they’ve been controlling for years and years. In short, you become the other person in the same manner as you are yourself. This would be equivalent to possession.

There is one final (advanced) method of controlling behavior. But this will be the topic of a further discussion.