Behavior as Consistently Produced Results

[From Rick Marken (2015.11.21.1400)]

Martin Taylor (2015.11.20/16/36)--

RM: I'm posting my answer to your post under a different subject heading because, as usual, this discussion with you has given me new insight into the reason people are having difficulty understanding what Bill is trying to explain in the section of LCS I that I am persistently pointing to (pp. 171-176): that what we see as "behavior" is a process of control. The insight came about while I was watering the plants in our garden (following Voltaire's advice) and thinking about the following reply to my question about whether we are seeing the reference states of controlled variables when we see "behavior".

RM: So do you agree that "extension of the leg" is a controlled variable with a reference state "fully extended"? And that the fact that this is a controlled variable can be objectively (experimentally) determined. A simple "yes" or "no" will do. And if the answer is "yes" then why not include them in the definition of "behavior"?

RM: I would really like to hear what your answer is to this.

MT: The answer is NO.

MT: Now let's see why. Firstly because you ask two separate questions. The answer to the first is "An observer might so assume, but there's no way for the observer to know". Only if the observer knows what perception was being controlled can the observer even see the state of the environmental referent of the controlled perception, and only if control is perfect and the observer knew that control was perfect would the observer be able to see its reference state. Bill covered the first point in the paragraph to which you referred when he said: "once behaviour has been defined in terms of an appropriate variable". As soon as he needed that precursor condition, he was into theory, not observation.

RM: This is what jogged my insight. Apparently you take the phrase "appropriate variable" to refer to "perceptual variable". And if this were what Bill meant by "appropriate variable" then, indeed, he would have been "into theory" because the perceptual variable is a theoretical construct in PCT. But that would be conflating theory and observation and I think Bill was way too smart to make that kind of mistake. To see why this would be a mistake, let's start by taking a look at the whole sentence in which this phrase occurs:

WTP: The existence of reference states is not conjectural; once behavior has been defined in terms of an appropriate variable, such reference states always exist.

RM: So the existence of reference states is not conjectural; they exist, they are not theoretical. And we know from the beginning of the next paragraph that these reference states are "..the problem to which control theory is addressed" -- that is, they are the phenomenon that control theory explains. We also know that controlled perceptions are part of control theory -- the theory that explains the existence of reference states of controlled variables. A theory is a conjecture about how things work; controlled perceptions, being part of control theory, are a conjecture about why the reference states of controlled variables exist.
RM: So if what Bill meant by "appropriate variable" in the sentence above was "perceptual variable", then what he would have been saying is that the non-conjectural existence of reference states is established once they have been defined in terms of conjectures about the reason for their existence. In other words, he would be saying that the existence of the reference state -- the phenomenon to be explained by control theory -- can only be observed using the theory that explains the phenomenon. This is clearly circular. It's like saying that the existence of linear acceleration -- a phenomenon explained by Newton's laws of motion -- could only be observed using the laws of motion that explain this phenomenon (yet Galileo managed to observe it 50 years before Newton was born).
RM: While I think it's pretty clear that Bill was not talking about "perceptual variables" when he referred to "appropriate variables" -- Bill knew very well the difference between a theory and the phenomenon the theory explains -- Martin's confusion led me to the realization that Bill's way of describing the phenomenon of reference states in terms of the "appropriate variables" might encourage confusion between theory and phenomenon. It might also suggest a level of analysis of the phenomenon of behavior that is far deeper than what is implied by the informal term "behavior". This eventually led to my insight, which was how the fact that behavior is control fits into research aimed at testing the PCT model of behavior -- a fact I hadn't really considered carefully before.
RM: The insight is that research on behavior from a control theory perspective has to start with a far more informal description of the behavior under study than what is implied in Table 1, p. 172 of LCS I. That table was created to show that behavior could be analyzed into a controlled variable, the reference state of that variable and the means used to control that variable. But that analysis was done after it was observed that the behavior being analyzed was a process of control. (The analysis could also be confusing because the terms used in the analysis are very similar (even identical) to those used in the theory that explains the results of the analysis. For example, using the term "controlled variable" to refer to the observed variable that is under control is often the same term we use to refer to the perceptual variable that is part of the theory that explains the observed control).
RM: So what I propose is a simpler analysis of behavior than the one shown in Table 1, p. 172 (and in my "Behavior is control" spreadsheet) that can be used as the starting point for PCT-based research on purposeful behavior. The evidence that behavior is control is that the events we call "behaviors" are consistently produced results of an organisms actions -- results that we know to be produced under quite different circumstances each time. So when we see people produce consistent results we know they are controlling. So I propose having a table for the analysis of behavior with the following columns:
Behavior | Results | Possible Disturbances | Possible CVs | Possible Reference States | Possible Means of Control
RM: The "Behavior" column would have the name of a behavior, such as "catching a ball", "riding a bike", "watering a plant", etc. The results column would be a description of a consistently produced result that is referred to by the behavior name. For example, one consistently produced result of "catching a ball" is "a caught ball";one consistently produced result of "riding a bike" is "a ridden bike"; one consistently produced result of "watering a plant" is "a watered plant'. These are pretty general descriptions of the results that are consistently produced but I think they get the point across.
RM: The fact that these consistently produced results are controlled results is deduced from knowing that they are produced under different circumstances -- what we call "varying disturbances" -- each time. These disturbances are not always obvious but we know they're there. The next column let's you speculate about what those varying disturbances might be. So one disturbance to catching a ball is the trajectory of the ball. The next three columns let you speculate in more detail about the results that are being controlled and the means by which they are controlled. This involves speculating about what variable might be controlled, what the reference state of that variable might be and, finally, what means might be used to control that variable. This is where we are getting into the realm of theory since more precise ideas about a controlled result requires knowing that such a result is an aspect of the environment seen from the point of view of the organism. In theory, the aspect of the aspect of the environment that is under control is a perception. But it's really not necessary to know this in order to come up with a more precise idea of the controlled variable.An example of this comes from attempts to determine the variable controlled when catching a ball. One of the first attempts to get a better description of the "result" that was being consistently produced was made by a physicist named Chapman who calculated what a person would see when a fly ball was hit straight at them in different trajectories. Based on a simple trigonometric analysis Chapman saw that a person would end up right under the ball if they kept the tangent of the optical angle made by the height of he ball relative to the eye increasing .
RM: Anyway, these are just preliminary thoughts but I think they are important things to consider for people who want to do research on the controlling done by living systems. The first step in such research is to describe the behavior you want to study in terms of consistently produced results -- that is, as control. The next step is to see this controlling in terms of possible controlled variables and their reference states. This step is aimed at getting a more precise idea of what results are being controlled. You do this by coming up with hypotheses about the aspects of the system's environment that correspond to the results being controlled. The next step is to test these hypotheses using some version of the TCV.
RM: All of these steps can be carried out without knowing PCT, although knowing PCT does help you come up with the right hypotheses. This is because PCT directs you to the fact that it is aspects of the environment, as seen from the point of view of the behaving system, that are under control. In the PCT model, these aspects of the environment that are controlled are called perceptions -- controlled perceptions.
RM: All this should have been the first chapter of my book "Doing Research on Purpose". But I think I will incorporate it into something like a little manual that will be about how to actually do research on control (purposeful behavior). I just don't know if I can think of a title for it that could possibly be as good as "Doing Research on Purpose".
Best
Rick

···

--
Richard S. Marken
<https://urldefense.proofpoint.com/v2/url?u=http-3A__www.mindreadings.com&d=BQMFaQ&c=8hUWFZcy2Z-Za5rBPlktOQ&r=-dJBNItYEMOLt6aj_KjGi2LMO_Q8QB-ZzxIZIF8DGyQ&m=0vvZtKeJRsZAgqQh8fUeqU9e9fKzyKqZeknR4Ge5Qx4&s=ASgJhzHkgYQliErkLME77XF7cRQCBJRr1uj6Dc3oiSo&e=>www.mindreadings.com
Author of <https://urldefense.proofpoint.com/v2/url?u=http-3A__www.amazon.com_Doing-2DResearch-2DPurpose-2DExperimental-2DPsychology_dp_0944337554_ref-3Dsr-5F1-5F1-3Fie-3DUTF8-26qid-3D1407342866-26sr-3D8-2D1-26keywords-3Ddoing-2Bresearch-2Bon-2Bpurpose&d=BQMFaQ&c=8hUWFZcy2Z-Za5rBPlktOQ&r=-dJBNItYEMOLt6aj_KjGi2LMO_Q8QB-ZzxIZIF8DGyQ&m=0vvZtKeJRsZAgqQh8fUeqU9e9fKzyKqZeknR4Ge5Qx4&s=WTxxthF_fHs988hxWU6OL21AUF4Fpe29iCBD95BeiFk&e=>Doing Research on Purpose.
Now available from Amazon or Barnes & Noble

[Martin Taylor 2015.11.21.23.39]

I suppose you will have read my response to Bruce Nevin, and now

realize that this is not so.

Or so Bill asserts, always with the important proviso that "an

appropriate variable" has been defined (not “observed”). What does
“appropriate” mean in this context? It’s at best ambiguous, and one
of the possible meanings brings us right into the circularity of
which you complain. That meaning is: “appropriate variable” – a
variable for which it is possible to observe consistent states
brought about by the action of the presumed actor despite variations
in context. (Or something similar).

A version of "appropriate" less problematic on its surface is

something along the lines of "a variable that can be identified with
a class of actions that have sufficient in common to be given the
same label in everyday speech, such as “watering the flowers’.” But
this only hides the same circularity, because being given the same
label implied that the variables are given a label because they
bring about consistent results through the action of an actor.

I don't have a meaning for "appropriate variable" in this context

that avoids the circularity, which is the same issue that is
involved when a scientist “cherry-picks” data that will conform to
the preferred hypothesis.

All of which is quite fair. There do exist situations in which it

looks as though an actor is bringing a variable to a predetermined
value, and perceptual control theory is a very good theory to
explain what is happening in those situations. And like any good
theory, it explains other situations, as well, in this case
observable behaviours that do not seem to bring any observable
variable to any particular state. According to PCT there are at
least two reasons this could happen; the observer might not have
“defined” an “appropriate variable”, or the actor’s behaviour may
not be effective in controlling the perceptual variable it is
supposed to influence. The observer may quite correctly observe
someone trying to lift a heavy rock and observe that the
“appropriate variable” is the rock’s position, but since the rock
does not move, is the observer justified in saying that the rock is
where the actor has a reference for where it should be?

Your misapprehension about my understanding...

True. And when they don't, we (or I) still believe that they are

controlling, but either we haven’t been able to see what perception
they are controlling or they aren’t very good at controlling it. A
skilled darts player doesn’t put the dart into the same part of the
dartboard every time he throws, but if you ask him to throw at the
bullseye every time, his darts will be clustered very close to it.
When he is playing a darts game, the observer unacquainted with the
scoring rules might see no difference between the pro’s control and
that of an occasional amateur player, since both scatter their hits
all over the board, and would not observe that the skilled player
threw at, and hit, triple 20 in this occasion and double 3 on that,
whereas the amateur successfully hit the board but seldom the
reference place on the board.

The point is that we extrapolate from "events we call "behaviors"

[which] are * consistently produced results of an organisms
actions"* to events we call behaviours that don’t produce any
consistent results that we can observe. That’s the theory part.
Using the same theory that explains the consistent results, we can
explain the inconsistent results.

And one consistent result of "voting for a candidate" is an X on a

piece of paper, perhaps, but even that isn’t assured, and it’s a lot
lower level than the perception (ineffectively) controlled using the
voting behaviour. Some people use guns to control that perception
rather more effectively.

And since PCT allows for the possibility that control may sometimes

not be perfect, or even very good, what aspects of the environment
are then called “perceptions”?

Martin
···

[From Rick Marken (2015.11.21.1400)]

            Martin Taylor

(2015.11.20/16/36)–


Bill covered the
first point in the paragraph to which you referred when he
said: “once behaviour has been defined in terms of an
appropriate variable”. As soon as he needed that precursor
condition, he was into theory, not observation.

          RM: This is what jogged my insight. Apparently you take

the phrase “appropriate variable” to refer to “perceptual
variable”.

          And if this were what Bill meant by "appropriate

variable" then, indeed, he would have been “into theory”
because the perceptual variable is a theoretical construct
in PCT. But that would be conflating theory and
observation and I think Bill was way too smart to make
that kind of mistake. To see why this would be a mistake,
let’s start by taking a look at the whole sentence in
which this phrase occurs:

          WTP:

The existence of reference states is not conjectural; once
behavior has been defined in terms of an appropriate
variable, such reference states always exist.

RM: So the existence of reference states is * not
conjectural*; they exist, they are * not
theoretical*.

          And we know from the beginning of the next paragraph

that these reference states are “…the problem to which
control theory is addressed” – that is, they are the
phenomenon that control theory explains. We also know
that controlled perceptions are part of control theory –
the theory that explains the existence of reference states
of controlled variables. A theory is a conjecture about
how things work; controlled perceptions, being part of
control theory, are a conjecture about why the reference
states of controlled variables exist.

          RM: While I think it's pretty clear that Bill was not

talking about “perceptual variables” when he referred to
“appropriate variables” – Bill knew very well the
difference between a theory and the phenomenon the theory
explains – Martin’s confusion


RM: So what I propose is a simpler analysis of behavior
than the one shown in Table 1, p. 172 (and in my “Behavior
is control” spreadsheet) that can be used as the starting
point for PCT-based research on purposeful behavior. The
evidence that behavior is control is that the events we
call “behaviors” are * consistently produced results of
an organisms actions* – results that we know to be
produced under quite different circumstances each time. So
when we see people produce consistent results we know they
are controlling.

          So I propose having a table for the analysis of

behavior with the following columns:

          Behavior | Results |  Possible

Disturbances | Possible CVs | Possible Reference States |
Possible Means of Control

          RM: The "Behavior" column would have the name of a

behavior, such as “catching a ball”, “riding a bike”,
“watering a plant”, etc. The results column would be a
description of a consistently produced result that is
referred to by the behavior name. For example, one
consistently produced result of “catching a ball” is “a
caught ball”;one consistently produced result of “riding a
bike” is “a ridden bike”; one consistently produced
result of “watering a plant” is "a watered plant’. These
are pretty general descriptions of the results that are
consistently produced but I think they get the point
across.

          RM: All of these steps can be carried out without

knowing PCT, although knowing PCT does help you come up
with the right hypotheses. This is because PCT directs you
to the fact that it is aspects of the environment, as seen
from the point of view of the behaving system, that are
under control. In the PCT model, these aspects of the
environment that are controlled are called perceptions –
controlled perceptions.

[From Rick Marken (2015.11.22.1215)]

···

Martin Taylor (2015.11.21.23.39)_-

MT: I suppose you will have read my response to Bruce Nevin, and now

realize that this is not so.

RM: There are so many threads going on that I’ve lost track.

MT: Or so Bill asserts, always with the important proviso that "an

appropriate variable" has been defined (not “observed”).

RM: So now it’s a “proviso”. This doesn;t help. What you take as a proviso I take as a foreshadowing of the TCV. It’s true that you don’t really have a good idea of what the reference state for a controlled variable is when you observe some behavior until you’ve done the TCV. The TCV is used do determine the most appropriate definition of the controlled variable and, thus, of the reference state thereof. But that doesn’t proviso the fact that reference states are real; that they exist.

RM: That was really the basis of my insight. When you look at behavior you can see that controlling is going on – that there exists a, possibly varying, reference state for a controlled variable – even if you don’t know exactly what variable is being controlled (you don’t know the correct definition of the “appropriate variable” that is being controlled). For example, you can see that a two-wheeled balancing robot is controlling its balance since you know that the robot would quickly fall over if balance were not being controlled. So research on control can start with a rather general observation that control is going on – that there is a reference state for a controlled variable – you just don’t know what it is. And that’s the goal of PCT research – figuring out what variable is actually under control. In the case of the balancing robot, for example, there are several possibilities: the controlled variable might be the location of its center of gravity, its optical orientation relative to environmental vertical, it’s angular momentum, etc. Now the job is to design tests to determine which of these hypotheses is the best definition of the controlled variable and its reference state.

RM: So I’ll try again to explain my insight. What I realized is that you don’t have to know what the reference state of a controlled variable is in order to know that one exists. You know that one exists when you see evidence that controlling is going on. This evidence can be pretty informal. The evidence is seeing that a consistent (or stable) result is being produced in a situation where such consistency is not expected. It does take some understanding of physics and physiology to know when a result is or is not expected – that’s why Bill was able to see that what we see as behavior is a process of control – but we don’t have to be completely accurate at identifying examples of control. If we are wrong – if an apparent example of control doesn’t actually involve control – then this will be quickly revealed when you do the TCV. For example, a friend sent me this video,assuming it was an astonishing example of control:

https://33.media.tumblr.com/aaafe5a3377ad150b19b95d0161298f0/tumblr_ni7bi8rcRt1sqkwlno1_400.gif

RM: I thought it might be the real deal at first myself. But it turns out to be a fake. But that would have been shown to be true very soon after you applied the first disturbance or discovered the support holding up the glass.

RM: So, again, I thacnk you for helping me get to this insight. I’ll try to write up a more detailed description of how this insight relates to “doing research on purpose” as soon as I get a chance.

Best regards

Rick

What does

“appropriate” mean in this context? It’s at best ambiguous, and one
of the possible meanings brings us right into the circularity of
which you complain. That meaning is: “appropriate variable” – a
variable for which it is possible to observe consistent states
brought about by the action of the presumed actor despite variations
in context. (Or something similar).

A version of "appropriate" less problematic on its surface is

something along the lines of "a variable that can be identified with
a class of actions that have sufficient in common to be given the
same label in everyday speech, such as “watering the flowers’.” But
this only hides the same circularity, because being given the same
label implied that the variables are given a label because they
bring about consistent results through the action of an actor.

I don't have a meaning for "appropriate variable" in this context

that avoids the circularity, which is the same issue that is
involved when a scientist “cherry-picks” data that will conform to
the preferred hypothesis.

All of which is quite fair. There do exist situations in which it

looks as though an actor is bringing a variable to a predetermined
value, and perceptual control theory is a very good theory to
explain what is happening in those situations. And like any good
theory, it explains other situations, as well, in this case
observable behaviours that do not seem to bring any observable
variable to any particular state. According to PCT there are at
least two reasons this could happen; the observer might not have
“defined” an “appropriate variable”, or the actor’s behaviour may
not be effective in controlling the perceptual variable it is
supposed to influence. The observer may quite correctly observe
someone trying to lift a heavy rock and observe that the
“appropriate variable” is the rock’s position, but since the rock
does not move, is the observer justified in saying that the rock is
where the actor has a reference for where it should be?

Your misapprehension about my understanding...
True. And when they don't, we (or I) still believe that they are

controlling, but either we haven’t been able to see what perception
they are controlling or they aren’t very good at controlling it. A
skilled darts player doesn’t put the dart into the same part of the
dartboard every time he throws, but if you ask him to throw at the
bullseye every time, his darts will be clustered very close to it.
When he is playing a darts game, the observer unacquainted with the
scoring rules might see no difference between the pro’s control and
that of an occasional amateur player, since both scatter their hits
all over the board, and would not observe that the skilled player
threw at, and hit, triple 20 in this occasion and double 3 on that,
whereas the amateur successfully hit the board but seldom the
reference place on the board.

The point is that we extrapolate from "events we call "behaviors"

[which] are * consistently produced results of an organisms
actions"* to events we call behaviours that don’t produce any
consistent results that we can observe. That’s the theory part.
Using the same theory that explains the consistent results, we can
explain the inconsistent results.

And one consistent result of "voting for a candidate" is an X on a

piece of paper, perhaps, but even that isn’t assured, and it’s a lot
lower level than the perception (ineffectively) controlled using the
voting behaviour. Some people use guns to control that perception
rather more effectively.

And since PCT allows for the possibility that control may sometimes

not be perfect, or even very good, what aspects of the environment
are then called “perceptions”?

Martin


Richard S. Marken

www.mindreadings.com
Author of Doing Research on Purpose.
Now available from Amazon or Barnes & Noble

          RM: This is what jogged my insight. Apparently you take

the phrase “appropriate variable” to refer to “perceptual
variable”.

RM: So the existence of reference states is * not
conjectural*; they exist, they are * not
theoretical*.

          And we know from the beginning of the next paragraph

that these reference states are “…the problem to which
control theory is addressed” – that is, they are the
phenomenon that control theory explains. We also know
that controlled perceptions are part of control theory –
the theory that explains the existence of reference states
of controlled variables. A theory is a conjecture about
how things work; controlled perceptions, being part of
control theory, are a conjecture about why the reference
states of controlled variables exist.

          RM: While I think it's pretty clear that Bill was not

talking about “perceptual variables” when he referred to
“appropriate variables” – Bill knew very well the
difference between a theory and the phenomenon the theory
explains – Martin’s confusion


RM: So what I propose is a simpler analysis of behavior
than the one shown in Table 1, p. 172 (and in my “Behavior
is control” spreadsheet) that can be used as the starting
point for PCT-based research on purposeful behavior. The
evidence that behavior is control is that the events we
call “behaviors” are * consistently produced results of
an organisms actions* – results that we know to be
produced under quite different circumstances each time. So
when we see people produce consistent results we know they
are controlling.

          So I propose having a table for the analysis of

behavior with the following columns:

          Behavior | Results |  Possible

Disturbances | Possible CVs | Possible Reference States |
Possible Means of Control

          RM: The "Behavior" column would have the name of a

behavior, such as “catching a ball”, “riding a bike”,
“watering a plant”, etc. The results column would be a
description of a consistently produced result that is
referred to by the behavior name. For example, one
consistently produced result of “catching a ball” is “a
caught ball”;one consistently produced result of “riding a
bike” is “a ridden bike”; one consistently produced
result of “watering a plant” is "a watered plant’. These
are pretty general descriptions of the results that are
consistently produced but I think they get the point
across.

          RM: All of these steps can be carried out without

knowing PCT, although knowing PCT does help you come up
with the right hypotheses. This is because PCT directs you
to the fact that it is aspects of the environment, as seen
from the point of view of the behaving system, that are
under control. In the PCT model, these aspects of the
environment that are controlled are called perceptions –
controlled perceptions.