Scholar Alert - [ "perceptual control theory" ]

[From Rick Marken (2016.01.09.1715)]

···

On Thu, Dec 29, 2016 at 3:53 AM, Warren Mansell wmansell@gmail.com wrote:

WM: Jeff Vancouver has just produced a wonderfully elegant computational model of the complex effects of self-efficacy based on PCT. It is also particularly exciting because it includes an imagination loop for the first time!Â

RM: According to Vancouver and Purl, their model is motivated by a paper by Powers (1991) Commentary on Bandura’s “Human Agency”, American Psychologist. Specifically, they say: "…Powers suggested that computational modeling, which PCT facilitates, might help clarify self-efficacy’s roles in human behaviorâ€?. I got a copy of the Powers’ 1991 comment and the closest thing to a suggestion that computational modeling might help clarify self efficacy’s roles in human behavior was the following:  “Control theory is best used and tested through the method
of modeling or simulation. To devise working models, however, and especially to
test them, one needs experimental data that are more reliable than the
customary results of research in the social sciencesâ€?.Â

RM: Since computational models are a form of working model, Vancouver and Purl are correct to say that Powers suggested the use of computational models, though he suggested this as the way to test control theory, not self-efficacy’s roles in human behavior. But leaving that aside, Vancouver and Purl seem to have ignored the last part of Bill’s recommendation about modeling: "To devise working models, however, and especially to test them, one needs experimental data that are more reliable than the customary results of research in the social sciencesâ€?.  Instead of testing their model against highly reliable experimental data, Vancouver and Purl test their model against the data obtained in an experiment reported by Schmidt and DeShon (2010)The Moderating Effects of Performance Ambiguity on the Relationship Between Self-Efficacy and Performance, J applied psychology which provides data that are anything but reliable. Schmidt and DeShon conducted a typical psychology experiment – a factorial experiment – where the goal was to see  whether the effect of one independent variable (self-efficacy) on some dependent variables (performance, effort) depends on the value of another variable (ambiguity).Â

RM: There are two problems with testing a model against the data in the Schmidt and DeShon (2010) study. First, the results are averages over many subjects; PCT models are typically tested by comparing them to the behavior of one individual at a time. Second, and more important, is the fact that the results in these experiments are highly unreliable. The predicted interaction between ambiguity and self-efficacy is statistically significant (p<.01) but the reliability of this prediction (measured as R^2) is .03; only 3% of the variance in either dependent variable was accounted for by the interaction between ambiguity and self-efficacy. So the PCT model is being used to explain a very weak effect (indeed, essentially random noise) and one that is not seen in a large proportion of the subjects tested; and in the subjects where the effect is seen, it is must usually be a very small effect, if it’s there at all.Â

RM: Moreover, Vancouver and Purl don’t provide any measure of goodness of fit of the model to the Schmidt and DeShon data. What they show is that their model produces a qualitative match of the average results of the Schmidt and DeShon. We don’t consider a PCT model to be a successful model of behavior unless it accounts for about 90% of the variance in the observed behavior and comes within about 3% of that behavior of each individual tested.

RM: There are also problems with the Vancouver and Purl PCT model itself (Figure 2 in their diagram). One of the main problems is that it is not clear what the controlled variables. One is “Task Progress”  which is apparently a combination of information ambiguity, task state, effort and self - efficacy. But I didn’t find this variable defined in the paper; that is, there is no description of how  these inputs combined to produce a perception of “task progress”. This is a big omission since this variable seems to be central to the behavior of the model. The other controlled variable is not labeled but the input to the perceptual function that defines this variable is “dynamic expected utility of the task”. This input is the output of the other control system; therefore it’s not clear to me how this “dynamic expected utility” variable is controlled.Â

RM: There are other problems with the model but the basic problem with the application of PCT in this situation is that it is not being used to explain controlling. The term “controlled variable” is nowhere to be found in either the Vancouver and Purl or the Schmidt and DeShon paper. Once again, the problem for PCT comes down to the fact that PCT explains a phenomenon – control, also known as purposeful behavior – to which conventional psychological researchers are essentially blind.Â

RM: Apparently J. Applied Psychology will occasionally publish Commentaries on published papers. I’m thinking of writing one on the Vancouver and Purl paper. Should I or would someone else like to do it.

BestÂ

Rick


Richard S. MarkenÂ

"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
                --Antoine de Saint-Exupery

RM: I appreciate Jeff’s enthusiasm for PCT but I think his model (described in Vancouver, J.D. & Â Purl , A. (2016) Computational Model of
Self-Efficacy’s Various Effects on Performance: Moving the Debate Forward, The
Journal of applied psychology
) gives a misleading impression of what PCT is about.Â

[From: Jeff Vancouver (2017.01.10 EST)]

Ref: [From: Richard Pfau (2016.01.10 10:21 EST)] – assume year wrong

Thanks for your endorsement. I would be interested in learning what the technical issues are.

···

[From: Richard Pfau (2016.01.10 10:21 EST)]

Ref: [Rick Marken (2016.01.09.1715)] and [Mansell (2016.12.29 6:53 am)]

I found the Vancouver article (“A Computational Model of Self-Efficacy’s Various Effects on Performance: Moving the Debate Forward”) to be very worthwhile in helping to highlight PCT and bring it to broader attention, showing how it complements and helps explain
self-efficiency, and helping to shed light on one of PCT’s “black boxes” (i.e., that of the Output Function; and yes, PCT has its “back boxes”, just as behaviorist thinking had its black box between the S and the R).

Anyhow, I found the article to be well written, articulate, and helpful in showing how thinking about PCT and self-efficiency can be integrated (to the benefit of PCT and its promotion among psychologists).

Although there may be te chnical issues, to me the overall article is commendable and well worth inclusion in the PCT marketing arsenal.

[From Rick Marken (2017.01.10.1335)]

···

David Goldstein sent this to me privately but I thought it should go to the net and David approved so here it is:Â

On Tue, Jan 10, 2017 at 7:13 AM, davidmg@verizon.net wrote:

(From David Goldstein (2017.01.09.1000))
[Re: Rick Marken (2016.01.09.1715)]

Rick,

All your points are well taken.

However, if you make a comment, I would suggest that you do it in the most positive way which shows a different way that Jeff could have carried out the study. The purpose would be to educate the reader to the more PCT research methodology that you advocate.

I would not want to discourage Jeff from pursuing PCT research or to feel that his PCT friend is attacking him. Suggesting an alternative method to address the research target would not do this, I think.

David

David

-----Original Message-----
From: davidmg@verizon.net
Sent: 2017-01-10 01:15:37 +0000
To: Richard Marken
Subject: RE: Re: Scholar Alert - [ “perceptual control theory” ]
[From Rick Marken (2016.01.09.1715)]

On Thu, Dec 29, 2016 at 3:53 AM, Warren Mansell wmansell@gmail.com wrote:

WM: Jeff Vancouver has just produced a wonderfully elegant computational model of the complex effects of self-efficacy based on PCT. It is also particularly exciting because it includes an imagination loop for the first time!Â

RM: According to Vancouver and Purl, their model is motivated by a paper by Powers (1991) Commentary on Bandura’s “Human Agency”, American Psychologist. Specifically, they say: "…Powers suggested that computational modeling, which PCT facilitates, might help clarify self-efficacy’s roles in human behaviorâ€?. I got a copy of the Powers’ 1991 comment and the closest thing to a suggestion that computational modeling might help clarify self efficacy’s roles in human behavior was the following:  “Control theory is best used and tested through the method of modeling or simulation. To devise working models, however, and especially to test them, one needs experimental data that are more reliable than the customary results of research in the social sciencesâ€?.Â

RM: Since computational models are a form of working model, Vancouver and Purl are correct to say that Powers suggested the use of computational models, though he suggested this as the way to test control theory, not self-efficacy’s roles in human behavior. But leaving that aside, Vancouver and Purl seem to have ignored the last part of Bill’s recommendation about modeling: "To devise working models, however, and especially to test them, one needs experimental data that are more reliable than the customary results of research in the social sciencesâ€?.  Instead of testing their model against highly reliable experimental data, Vancouver and Purl test their model against the data obtained in an experiment reported by Schmidt and DeShon (2010)The Moderating Effects of Performance Ambiguity on the Relationship Between Self-Efficacy and Performance, J applied psychology which provides data that are anything but reliable. Schmidt and DeShon conducted a typical psychology experiment – a factorial experiment – where the goal was to see  whether the effect of one independent variable (self-efficacy) on some dependent variables (performance, effort) depends on the value of another variable (ambiguity).Â

RM: There are two problems with testing a model against the data in the Schmidt and DeShon (2010) study. First, the results are averages over many subjects; PCT models are typically tested by comparing them to the behavior of one individual at a time. Second, and more important, is the fact that the results in these experiments are highly unreliable. The predicted interaction between ambiguity and self-efficacy is statistically significant (p<.01) but the reliability of this prediction (measured as R^2) is .03; only 3% of the variance in either dependent variable was accounted for by the interaction between ambiguity and self-efficacy. So the PCT model is being used to explain a very weak effect (indeed, essentially random noise) and one that is not seen in a large proportion of the subjects tested; and in the subjects where the effect is seen, it is must usually be a very small effect, if it’s there at all.Â

RM: Moreover, Vancouver and Purl don’t provide any measure of goodness of fit of the model to the Schmidt and DeShon data. What they show is that their model produces a qualitative match of the average results of the Schmidt and DeShon. We don’t consider a PCT model to be a successful model of behavior unless it accounts for about 90% of the variance in the observed behavior and comes within about 3% of that behavior of each individual tested.

RM: There are also problems with the Vancouver and Purl PCT model itself (Figure 2 in their diagram). One of the main problems is that it is not clear what the controlled variables. One is “Task Progress”  which is apparently a combination of information ambiguity, task state, effort and self - efficacy. But I didn’t find this variable defined in the paper; that is, there is no description of how  these inputs combined to produce a perception of “task progress”. This is a big omission since this variable seems to be central to the behavior of the model. The other controlled variable is not labeled but the input to the perceptual function that defines this variable is “dynamic expected utility of the task”. This input is the output of the other control system; therefore it’s not clear to me how this “dynamic expected utility” variable is controlled.Â

RM: There are other problems with the model but the basic problem with the application of PCT in this situation is that it is not being used to explain controlling. The term “controlled variable” is nowhere to be found in either the Vancouver and Purl or the Schmidt and DeShon paper. Once again, the problem for PCT comes down to the fact that PCT explains a phenomenon – control, also known as purposeful behavior – to which conventional psychological researchers are essentially blind.Â

RM: Apparently J. Applied Psychology will occasionally publish Commentaries on published papers. I’m thinking of writing one on the Vancouver and Purl paper. Should I or would someone else like to do it.

BestÂ

Rick

Richard S. MarkenÂ

"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
                --Antoine de Saint-Exupery

RM: I appreciate Jeff’s enthusiasm for PCT but I think his model (described in Vancouver, J.D. %26 Â Purl , A. (2016) Computational Model of Self-Efficacy’s Various Effects on Performance: Moving the Debate Forward, The Journal of applied psychology) gives a misleading impression of what PCT is about.Â

Richard S. MarkenÂ

"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
                --Antoine de Saint-Exupery

[From Rick Marken (2016.01.10.1520)]

···

Richard Pfau (2016.01.10 10:21 EST)–

RP: I found the Vancouver article (“A Computational Model of Self-Efficacy’s Various Effects on Performance: Moving the Debate Forward”) to be very worthwhile in helping to highlight PCT and bring it to broader attention, showing how it complements and helps explain self-efficiency…

RM: Hi Richard. I presume you mean “self efficacy”, not “self-efficiency”. I would be interested in hearing what it was about self-efficacy that you think is explained by Jeff’s model. My own take on it is that the model explains “self-efficacy” in the same way that a group of physicians explained the sleep inducing properties of opium in Molière’s *The Imaginary Invalid *-- as the result of a “virtus dormitiva” or “dormitive principle”. That is, the phenomenon to be explained (sleep inducing properties of opium, self-efficacy) is explained in terms of the phenomenon to be explained (the sleep inducing properties of opium, self-efficacy).Â

RM: Usually, the "dormitive principle" explanation is couched in different words than those used to describe the phenomenon to be explained (so the explanation of the “sleep inducing properties of opium” is referred to as a “dormitive principle” or, even better,  a “virtus dormitiva”, rather than the “sleep inducing properties of opium”). But in the case of “self-efficacy” the explanation is couched in the same words as the phenomenon to be explained. This can be seen in Jeff’s model:

Â

RM: Note that “self-efficacy” is a variable in the model of a person displaying “self-efficacy”. So this is a model that clearly employs a “dormitive principle” approach to explaining “self-efficacy”: “self-efficacy” is a result of a “self-efficacy principle” that is a property of people displaying “self-efficacy”.Â

RM: PCT models specifically aim to avoid this kind of tautological approach to explanation, which is very common in non-PCT models of behavioral phenomena. For example, here’s a nice post from Bill Powers (from back in 1994) that explains how the "dormitive principle" shows up in reinforcement models of “operant” behavior. Note that the only thing Bill got wrong here is the idea of a “dormitive principle” comes from Voltaire rather than Molière. But he was in the right country.

[From Bill Powers (941122.0730 MST)]Â

Bruce Abbott: Where on Earth do you get these ideas? Dormative principle? Â
I don’t think I can name a single person doing work in the experimental
analysis of behavior who would subscribe to your conception of THEIR
conception of reinforcement. You are, as I and, apparently, quite a
number of others have tried to tell you, constructing a straw-person to
demolish.Â

Bill Powers: Of course they wouldn’t. They think their approach is perfectly
reasonable, otherwise they wouldn’t use it. I’m not saying they’re
stupid. But from my outsider’s point of view, they are saying that the
reason a food pellet alters behavior rates is that the food pellet
contains the ability to reinforce behavior, which means it affects a
probability of producing the behavior, which means it alters the
behavior rate. This probability is itself unobservable: all we observe
are changes in mean behavior rates (or inter-response intervals), not
changes in the presumed probabilities that bring them about or the
imagined effects of the food pellet on those probabilities.
So the parallel to Bateson’s “dormitive principle” (someone has
mentioned that the real source was Voltaire) is quite exact.
Reinforcement alters behavior rates because the events said to be
reinforcing contain some unknown property that alters a probability and
altering that probability alters the behavior rate. What is that
property? It is the property of altering the probability of emitting a
behavior. This is an exact example of a dormitive principle: the
explanation of why an effect results from a cause is that there is
something about the cause capable of producing the effect. In short, the
explanation is completely devoid of content. The statement boils down to
" there is a relationship between mean behavior rates and mean rates of
delivery of food pellets because there is a relationship between mean
behavior rates and mean rates of delivery of food pellets."Â

This is the mode of explanation of mysterious phenomena that was used
before Galileo. It is prescientific. Before you object that EAB people
would not be so dumb as to employ a prescientific mode of explanation,
let me remind you that the people who lived before Galileo were no less
intelligent than those who lived afterward; evolution doesn’t work that
fast. The only difference between the pre- and post-Galilean modes of
explanation was that after Galileo we began demanding to know the
mechanisms behind apparent cause-effect relationships. Our brains didn’t
get any better; only our conceptions did.

Best regards

Rick


Richard S. MarkenÂ

"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
                --Antoine de Saint-Exupery

I definitely agree with you here Rick - a PCT model should have no ‘dormitive principles’. Most words we use in psychology should have meaning only from the outside as we try to explain what a working model is doing - where the components of the PCT model are strictly defined and internal and therefore cannot have direct counterparts to the observer…
Warren

···

On Tue, Jan 10, 2017 at 11:21 PM, Richard Marken rsmarken@gmail.com wrote:

[From Rick Marken (2016.01.10.1520)]

Richard Pfau (2016.01.10 10:21 EST)–

RP: I found the Vancouver article (“A Computational Model of Self-Efficacy’s Various Effects on Performance: Moving the Debate Forward”) to be very worthwhile in helping to highlight PCT and bring it to broader attention, showing how it complements and helps explain self-efficiency…

RM: Hi Richard. I presume you mean “self efficacy”, not “self-efficiency”. I would be interested in hearing what it was about self-efficacy that you think is explained by Jeff’s model. My own take on it is that the model explains “self-efficacy” in the same way that a group of physicians explained the sleep inducing properties of opium in Molière’s *The Imaginary Invalid *-- as the result of a “virtus dormitiva” or “dormitive principle”. That is, the phenomenon to be explained (sleep inducing properties of opium, self-efficacy) is explained in terms of the phenomenon to be explained (the sleep inducing properties of opium, self-efficacy).Â

RM: Usually, the "dormitive principle" explanation is couched in different words than those used to describe the phenomenon to be explained (so the explanation of the “sleep inducing properties of opium” is referred to as a “dormitive principle” or, even better,  a “virtus dormitiva”, rather than the “sleep inducing properties of opium”). But in the case of “self-efficacy” the explanation is couched in the same words as the phenomenon to be explained. This can be seen in Jeff’s model:

Â

RM: Note that “self-efficacy” is a variable in the model of a person displaying “self-efficacy”. So this is a model that clearly employs a “dormitive principle” approach to explaining “self-efficacy”: “self-efficacy” is a result of a “self-efficacy principle” that is a property of people displaying “self-efficacy”.Â

RM: PCT models specifically aim to avoid this kind of tautological approach to explanation, which is very common in non-PCT models of behavioral phenomena. For example, here’s a nice post from Bill Powers (from back in 1994) that explains how the "dormitive principle" shows up in reinforcement models of “operant” behavior. Note that the only thing Bill got wrong here is the idea of a “dormitive principle” comes from Voltaire rather than Molière. But he was in the right country.

[From Bill Powers (941122.0730 MST)]Â

Bruce Abbott: Where on Earth do you get these ideas? Dormative principle? Â
I don’t think I can name a single person doing work in the experimental
analysis of behavior who would subscribe to your conception of THEIR
conception of reinforcement. You are, as I and, apparently, quite a
number of others have tried to tell you, constructing a straw-person to
demolish.Â

Bill Powers: Of course they wouldn’t. They think their approach is perfectly
reasonable, otherwise they wouldn’t use it. I’m not saying they’re
stupid. But from my outsider’s point of view, they are saying that the
reason a food pellet alters behavior rates is that the food pellet
contains the ability to reinforce behavior, which means it affects a
probability of producing the behavior, which means it alters the
behavior rate. This probability is itself unobservable: all we observe
are changes in mean behavior rates (or inter-response intervals), not
changes in the presumed probabilities that bring them about or the
imagined effects of the food pellet on those probabilities.
So the parallel to Bateson’s “dormitive principle” (someone has
mentioned that the real source was Voltaire) is quite exact.
Reinforcement alters behavior rates because the events said to be
reinforcing contain some unknown property that alters a probability and
altering that probability alters the behavior rate. What is that
property? It is the property of altering the probability of emitting a
behavior. This is an exact example of a dormitive principle: the
explanation of why an effect results from a cause is that there is
something about the cause capable of producing the effect. In short, the
explanation is completely devoid of content. The statement boils down to
" there is a relationship between mean behavior rates and mean rates of
delivery of food pellets because there is a relationship between mean
behavior rates and mean rates of delivery of food pellets."Â

This is the mode of explanation of mysterious phenomena that was used
before Galileo. It is prescientific. Before you object that EAB people
would not be so dumb as to employ a prescientific mode of explanation,
let me remind you that the people who lived before Galileo were no less
intelligent than those who lived afterward; evolution doesn’t work that
fast. The only difference between the pre- and post-Galilean modes of
explanation was that after Galileo we began demanding to know the
mechanisms behind apparent cause-effect relationships. Our brains didn’t
get any better; only our conceptions did.

Best regards

Rick


Richard S. MarkenÂ

"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
                --Antoine de Saint-Exupery

Dr Warren Mansell
Reader in Clinical Psychology

School of Health Sciences
2nd Floor Zochonis Building
University of Manchester
Oxford Road
Manchester M13 9PL
Email: warren.mansell@manchester.ac.uk
Â
Tel: +44 (0) 161 275 8589
Â
Website: http://www.psych-sci.manchester.ac.uk/staff/131406
Â
Advanced notice of a new transdiagnostic therapy manual, authored by Carey, Mansell & Tai - Principles-Based Counselling and Psychotherapy: A Method of Levels Approach

Available Now

Check www.pctweb.org for further information on Perceptual Control Theory

[From Rick Marken (2017.01.11.0930)]

···

RM: One of the main problems is that I’m not really sure what Jeff and the other “self-efficacy” researchers are trying to find out that is relevant to how people control. But a hint comes from Bill’s 1991 Comment where he says the following:

BP: A belief interpreted as an optimistic goal for effectiveness
would lead to an increase of effort and, if the effort were successful, an
increase of the perceived and actual effectiveness of action. A belief that works
by increasing the optimism of perceptions- by representing the same actual consequences
of behavior more favorably-- reduces the apparent shortfall that is driving
behavior, and so decreases behavior. And belief as imagination—subsstituting imagined
success for a perception that reports something less than success-- can put an
end to effective action.

 RM: I take “belief” here to refer to the belief in one’s self-efficacy. In that case, Bill has described three possible ways self-efficacy might be implemented in a control model: as an increased reference for achieving a controlled result, as an increased perception of the controlled result or as imagined successful control. I would test these ideas in a simple tracking task where we can accurately measure how well a person is controlling. After having the participant practice the task until she was fairly skilled at it I would test the participant in a series of tracking trials. At the end of each trial I would provide a measure of how well the participant did on the prior trial and ask for an estimate of how much better or worse she expected to do on the current trial. Then I would see if there was a relationship between these estimates of “self- efficacy” and the measures of control. If the correlation between these variables were anything less that .90 I would conclude that estimates/beliefs about “self-efficacy”  have little to do with one’s actual efficacy (ability to control). If, on the other hand, there is a strong relationship between estimates of and actual self-efficacy then I would look to see what parameters of a control model seem to be affected by these estimates – references, perceptions, gain, etc.

RM: This is just one possibility. I’m sure there are other ways to approach it. But the essential point is to start by keeping it simple. And, of course, the research should be done one participant at a time – what’s called a “completely within subjects design” – since it might be true that beliefs about self efficacy have different effects for different people.Â

Â

RM: If you have to worry about a person being driven away from PCT when you correct them about misconceptions about PCT then I don’t think PCT (at least as I understand it) was really in the cards for that person. Jeff has endured plenty of “correcting” from both Bill and myself and, while he hasn’t accepted many of these corrections, he has certainly stuck with PCT, as he understands it. So I wouldn’t worry about Jeff remaining with PCT, at least, as he understands it. And, finally, I don’t think people should be immune from criticism of their ideas simply because they are a devotee of PCT. PCT is a science and there are right and wrong answers; it’s not a matter of shades of gray. I do think that  people who are new to PCT should given more leeway and corrected gently, if at all. But after a certain amount of time (like 5 years) I think you can feel free to be somewhat more strict;-)

Best regards

Rick

Â


Richard S. MarkenÂ

"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
                --Antoine de Saint-Exupery

David Goldstein (2017.01.09.1000)–

DG: However, if you make a comment, I would suggest that you do it in the most positive way which shows a different way that Jeff could have carried out the study. The purpose would be to educate the reader to the more PCT research methodology that you advocate.

DG: I would not want to discourage Jeff from pursuing PCT research or to feel that his PCT friend is attacking him. Suggesting an alternative method to address the research target would not do this, I think.

From Bruce Abbott (2017.01.11.1500 EST)]

Rick Marken (2016.01.10.1520) –

Richard Pfau (2016.01.10 10:21 EST)

RP: I found the Vancouver article (“A Computational Model of Self-Efficacy’s Various Effects on Performance: Moving the Debate Forward”) to be very worthwhile in helping to highlight PCT and bring it to broader attention, showing how it complements and helps explain self-efficiency…

RM: Hi Richard. I presume you mean “self efficacy”, not “self-efficiency”. I would be interested in hearing what it was about self-efficacy that you think is explained by Jeff’s model. My own take on it is that the model explains “self-efficacy” in the same way that a group of physicians explained the sleep inducing properties of opium in Molière’s *The Imaginary Invalid *-- as the result of a “virtus dormitiva” or “dormitive principle”. That is, the phenomenon to be explained (sleep inducing properties of opium, self-efficacy) is explained in terms of the phenomenon to be explained (the sleep inducing properties of opium, self-efficacy).

RM: Usually, the “dormitive principle” explanation is couched in different words than those used to describe the phenomenon to be explained (so the explanation of the “sleep inducing properties of opium” is referred to as a “dormitive principle” or, even better, a “virtus dormitiva”, rather than the “sleep inducing properties of opium”). But in the case of “self-efficacy” the explanation is couched in the same words as the phenomenon to be explained. This can be seen in Jeff’s model:

image00281.png

RM: Note that “self-efficacy” is a variable in the model of a person displaying “self-efficacy”. So this is a model that clearly employs a “dormitive principle” approach to explaining “self-efficacy”: “self-efficacy” is a result of a “self-efficacy principle” that is a property of people displaying “self-efficacy”.

BA: I don’t see how you arrive at this conclusion. Self-efficacy is not being used to explain why the person is displaying self-efficacy, any more than a perceptual variable in PCT is being used to explain why a person is displaying a perception.  Look at the model: Self-efficacy is a variable that depends on two inputs: ability and bias. It acts both as an input to the task input perceptual function, whose output is described as “perception of task progress.� Together with “task importance,� self-efficacy (along with “task importance�) also affects the output function.

I’m not defending the model and its constructs, but clearly the model isn’t stating that self-efficacy is being used both as a concept to be explained and its own explanation (i.e., dormitive principle).

Dormitive principles are circular explanations and as such they are easily confused with definitions. In the exchange you quoted between Bill and me, Bill asserted that the reinforcement principle is a dormitive principle: Food serves as a reinforcer of the behavior that produces it (under certain conditions) because it contains the dormitive principle of the ability to reinforce. But reinforcers are labeled as such when, under given conditions, their production by a behavior leads to an increase in the rate of the behavior.  In other words, the observed ability to increase the rate of response defines the consequent delivery of the food (or whatever) as a reinforcer. Consequently, Bill’s assertion that “reinforcement� is a dormitive principle was wrong; he misconstrued a definition as an explanation.

In the present case, your assertion that self-efficacy functions as a dormitive principle does not appear to have arisen by misconstruing definition as explanation.  Self-efficacy is not defined in the diagram (and I do not have access to the article so I must rely on the diagram), but it is clear in the diagram that self-efficacy is not being used there to explain self-efficacy. Instead, it is shown in the diagram to derive from other variables and to have effects on yet other variables in a dynamic system.

Bruce

[From Jeff Vancouver (2017.01.20.1130 EST)]

···

Hi Rick (and all),

Fist, Rick, you are have not updated the year of your posts yet, FYI.

[From Rick Marken (2016.01.09.1715)]

WM: Jeff Vancouver has just produced a wonderfully elegant computational model of the complex effects of self-efficacy based on PCT. It is also particularly exciting because it includes an imagination loop for the first time!

JV: Thank you Warren.

RM: I appreciate Jeff’s enthusiasm for PCT but I think his model (described in Vancouver, J.D. & Purl , A. (2016) Computational Model of Self-Efficacy’s Various Effects on Performance: Moving the Debate Forward,
The Journal of applied psychology) gives a misleading impression of what PCT is about.

RM: According to Vancouver and Purl, their model is motivated by a paper by Powers (1991) Commentary on Bandura’s “Human Agency”,
American Psychologist . Specifically, they say: "…Powers suggested that computational modeling, which PCT facilitates, might help clarify self-efficacy’s roles in human behaviorâ€?. I got a copy of the Powers’ 1991 comment and the closest thing to a suggestion
that computational modeling might help clarify self efficacy’s roles in human behavior was the following: “Control theory is best used and tested through the method of modeling or simulation. To devise working models, however, and especially to test them,
one needs experimental data that are more reliable than the customary results of research in the social sciences�.

RM: Since computational models are a form of working model, Vancouver and Purl are correct to say that Powers suggested the use of computational models, though he suggested this as the way to test control theory, not self-efficacy’s roles
in human behavior. But leaving that aside, Vancouver and Purl seem to have ignored the last part of Bill’s recommendation about modeling: "To devise working models, however, and especially to test them,
one needs experimental data that are more reliable than the customary results of research in the social sciences �. Instead of testing their model against highly reliable experimental data, Vancouver and Purl test their model against the data
obtained in an experiment reported by Schmidt and DeShon (2010)The Moderating Effects of Performance Ambiguity on the Relationship Between Self-Efficacy and Performance, J applied psychology which provides data that are anything but reliable. Schmidt
and DeShon conducted a typical psychology experiment – a factorial experiment – where the goal was to see whether the effect of one independent variable (self-efficacy) on some dependent variables (performance, effort) depends on the value of another variable
(ambiguity).

JV: Respectively disagree. Bill seemed to be saying that PCT and modeling could help clarify social cognitive theory, which includes self-efficacy as a key concept,
but that is not important anyway so I too will put that away.

Regarding the way one tests their models I will say as I have to Rick before (he, like me, has much trouble listening to lessons – its okay DG, Rick and I have
a long history, but I will get back to your point) that tackling explanations behavior one can look at it well enough is like the person looking for their wallet where the light is better as opposed to where one thinks they lost the wallet. I have fit data
like Rick wants (Vancouver, Weinhardt, & Schmidt, 2010 – see references below) and to tthe degree he wants it (Vancouver, Putka, and Scherbaum, 2005). But there is more than one way to validate models and I am disinclined to consider only the criteria laid
out by some who are working on problems that can obtain that level of fit. Indeed, the 2005 paper gets great fit, but is of less relevance to making a scientific point because it is largely uncontested.

RM: There are two problems with testing a model against the data in the Schmidt and DeShon (2010) study. First, the results are averages over many subjects; PCT models are typically tested by comparing them to the behavior of one individual
at a time. Second, and more important, is the fact that the results in these experiments are highly unreliable. The predicted interaction between ambiguity and self-efficacy is statistically significant (p<.01) but the reliability of this prediction (measured
as R^2) is .03; only 3% of the variance in either dependent variable was accounted for by the interaction between ambiguity and self-efficacy. So the PCT model is being used to explain a very weak effect (indeed, essentially random noise) and one that is not
seen in a large proportion of the subjects tested; and in the subjects where the effect is seen, it is must usually be a very small effect, if it’s there at all.

JV: Yes, the phenomenon explained is very weak, so the effect is very small. Some phenomena is weak.

RM: Moreover, Vancouver and Purl don’t provide any measure of goodness of fit of the model to the Schmidt and DeShon data. What they show is that their model produces a qualitative match of the average results of the Schmidt and DeShon.
We don’t consider a PCT model to be a successful model of behavior unless it accounts for about 90% of the variance in the observed behavior and comes within about 3% of that behavior of each individual tested.

JV: I did not have the data, so I could not fit individual data. Even if I did, there is a lot of latent processes (and variables) involved such that fitting
the data well would be difficult. The point here was could I fit it at all.

RM: There are also problems with the Vancouver and Purl PCT model itself (Figure 2 in their diagram). One of the main problems is that it is not clear what the controlled variables. One is “Task Progress” which is apparently a combination
of information ambiguity, task state, effort and self - efficacy. But I didn’t find this variable defined in the paper; that is, there is no description of how these inputs combined to produce a perception of “task progress”. This is a big omission since
this variable seems to be central to the behavior of the model. The other controlled variable is not labeled but the input to the perceptual function that defines this variable is “dynamic expected utility of the task”. This input is the output of the other
control system; therefore it’s not clear to me how this “dynamic expected utility” variable is controlled.

JV: The math used to represent the task progress input function is in the paper. Moreover, the variable in the environment called task state is defined. All the
variables used to explain the inputs in to the input function are explained. This comment seems to come down to you not reading the paper well enough.

JV: On the issue of the controlled variable for the dynamic expected utility variable I would agree that what is controlled is much more ambiguous. Indeed, I
would say it is not an accurate way of talking about that that subsystem does, though I am not completely convinced I have abstracted it fully. Does that mean that my model is not PCT? Not sure. It is inspired by PCT, but did it go outside of PCT, maybe. See
next response.

RM: There are other problems with the model but the basic problem with the application of PCT in this situation is that it is not being used to explain controlling. The term “controlled variable” is nowhere to be found in either the Vancouver
and Purl or the Schmidt and DeShon paper. Once again, the problem for PCT comes down to the fact that PCT explains a phenomenon – control, also known as purposeful behavior – to which conventional psychological researchers are essentially blind.

JV: First, I have decided that the word control is very problematic awhile back, so I avoid it. That is, engineers tell me that the action the control takes is
the controlled variable. That action is to affect the system state so it gets to or remains at the referent. I can see how one could say that the system take is the controlled variable as Powers does, but it has created a sematic battle that seems unproductive
to wage.

RM: Apparently J. Applied Psychology will occasionally publish Commentaries on published papers. I’m thinking of writing one on the Vancouver and Purl paper. Should I or would someone else like to do it.

JV: If you make a comment on this, it seems that what would be productive is to pose the alternative. So along the lines of DG’s comment, if you have a protocol
in mind that would be better (or could further test the model), then present it. Likewise, if you think the model is not what explains data like Schmidt and DeShon’s, then make an alternative model, perhaps with the protocol that could distinguish them. Of
course, that is a lot to do, so getting it into one paper is a challenge. Indeed, I take baby steps because I learned that too much in one paper overwhelms the reviewers. I suspect that you do not think that the Schmidt and DeShon data is worthy of explanation
and that is fine. But what I think you want to do is say my model is not a PCT model. That is understandable (i.e., the choice agent is not a PCT system by Powers’ definition), but such a complaint misses a, if not the main, lesson of the paper: We should
not care what the theory is called that explains human behavior, we should care about how accurate it is. In this case, are the kinds of agents used in this model necessary to explain the phenomena observed?

JV: Rick, if you recall we went through this with my 2010/2014 model. You thought it was not true to the PCT core. You thought you could make a model that was
true to the PCT core and thus different than mine. I was excited about the possibility. We talked on the phone. You ended up not being able to envision a different model than mine once you understood my model. Shall we play that game again?

Rick Marken (2016.01.10.1520) & From Bruce Abbott (2017.01.11.1500 EST)]

JV: On the issue of the explanation of self-efficacy, I think BA defended me well. I would also add that, as noted in the paper, the way self-efficacy belief develops in the mind is described in our 2014 paper, as noted in this paper. That
is, in this paper we took a short cut by making it a function of ability and bias whereas in the 2014 we should how it could develop and be a function of ability and where the bias might come from (though more work can clearly be done there). Because each
paper is constrained in terms of space, on can only do so much in a single paper.

Somewhat yours always,

Jeff Vancouver

References

Vancouver, J. B., Putka, D. J., & Scherbaum, C. A. (2005). Testing a computational model of the goal-level effect: An example of a neglected methodology. Organizational Research Methods, 8, 100-127.

Vancouver, J. B., Weinhardt, J. M., & Schmidt, A. M. (2010). A formal, computational theory of multiple-goal pursuit: Integrating goal-choice and goal-striving processes. Journal of Applied Psychology, 95, 985-1008.

Vancouver, J. B., Weinhardt, J.M., Vigo, R (2014). Change one can believe in: Adding learning to a computational model of self-regulation.
Organizational Behavior and Human Decision Processing, 124, 56-74.

[From Rick Marken (2017.01.20.1745)]

···

Richard Pfau (2017.01.20 10:15 EST)

RP: Ref: [Rick Marken (2016.01.10 1520)]Â Â “I would be interested in hearing what it was about self-efficacy that you think is explained by Jeff’s model.”

RP:Â Actually, I was referring to Jeff and Purl’s total article rather than to the model per se.

RM: Then could you tell me what it was about self-efficacy that you thought was explained by the whole article?

Â

RP: By way of explanation, I include the attached (in part, for the benefit of other csg readers). As indicated by that attachment, when explaining behavior, we can focus on any of several levels of analysis – ranging from the broad macro-environment down to the cell and molecular level.Â

RM: I think Jeff and Purl’s modeling is at the same level of explanation as PCT models - the functional level. They just introduce functions that are not implementable by the neurophysiology (functions like “self efficacy”). The functional components of the PCT model are designed to be consistent with what we know about the nervous system (See Chapter 3, “Premises” in B:CP).

Â

RP: Concepts such as self-efficacy, belief, and purpose are at “The Person/Individual Level” while PCT is at "The Sub-Personal Level."Â

RM: And, unfortunately, they are mixed together inappropriately in Jeff and Purl’s model.

Â

RP: One reason that I appreciate Jeff’s article is that he and Purl indicate how self-efficacy and PCT (at their two different levels of analysis and explanation) can be viewed as complementary and related to one another.

RM: It seem that what you view as “complementary” is what I view as confusion.Â

Â

RP: The article shows how self-efficacy might be modeled (at The Sub-Personal Level) and its various effects on performance (at The Person/Individual Level) explained using the PCT framework.Â

RM: Could you explain their model of self- efficacy? It looked to me like their model doesn’t explain self-efficacy; it just says that there is this function in model called “self-efficacy” that affects other functions in the model.

RP: By showing how PCT can account for aspects of self-efficacy, a phenomenon of interest to many psychologists, Vancouver and Purl also draw more attention to PCT.Â

RM: I’m far less interested in drawing more attention to PCT than in presenting an accurate description of PCT and examples of how to study the controlling done by living systems that is explained by PCT. I don’t care for inaccurate presentations of PCT and incorrect examples of how to test the model even if it draws attention to PCT – indeed, especially if it draws attention to PCT.Â

Â

RP: Bridging the gap between explanatory levels as Vancouver and Purl are
doing can help bring psychologists interested in other levels into the PCT fold.

RM: This assumes that psychologists have not entered the PCT fold because they are interested in other levels of analysis. I think it’s for other reasons. But, again, I’m not interested in bringing psychologists into the PCT fold; I want psychologists to come into the fold of their own volition because they have a correct understanding of PCT, see its significance and are willing and able to start a science of life based on an understanding of the fact that organisms are control (purposeful) systems.Â

RP: In doing so, the article also begins to shed light on possible mechanisms and workings within one of PCT’s “black boxes” – the Output Function.  In this way, consideration of self-efficacy seems to enrich modeling and thinking about PCT functions, processes, and mechanisms.

RM: The possible mechanisms of the black boxes are neural mechanisms; some possibilities are described in the “Premises” chapter of B:CP

Â

RP: As with the Output Function, the Input Function (another of PCT’s black boxes) would seemingly benefit from more thinking and research aimed at clarifying its processes.Â

RM: The actual hardware mechanisms of the perceptual and output functions are being studied by neurophysiologists; PCT models must include functional descriptions of these components of the control loopt. These descriptions can serve as specifications for the kind functions that the neural mechanisms must be able to carry out.Â

Â

RP: A focus on controlled variables, perception, and the modeling of control processes by PCTers is important, of course, and at the heart of PCT. But it seems as if opening up PCT’s black boxes (of the Input Function and Output Function) and bringing more psychologists having interests in areas such as self-efficacy into the PCT fold through efforts such as Vancouver’s and Purl’s is also important – and largely neglected to date.

RM: What you see when you open PCT’s black boxes are the neural (and other physiological) mechanisms that implement the functions they carry out. You certainly won’t see “self-efficacy” or anything like that in any of them;-)

BestÂ

Rick


Richard S. MarkenÂ

"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
                --Antoine de Saint-Exupery

[Martin Taylor 2017.01.20.23.10]

[From Rick Marken (2017.01.20.1745)]

Could you tell the rest of us why you think there should be

something in the article that explains self-efficacy, or why you
want Richard to invent something that isn’t in the article? I can’t
find anything in the article that suggests an intent or attempt to
explain self-efficacy. After all, that’s not what the article claims
to be about, either in its title or, so far as I can see, in its
text. Here’s the first half of the abstract again:

···
                                                Richard

Pfau (2017.01.20
10:15 EST)

                                                RP: Ref: [Rick

Marken (2016.01.10
1520)]Â Â “I would be
interested in
hearing what it was
about self-efficacy
that you think is
explained by Jeff’s
model.”

                                                RP:Â  Actually, I was

referring to Jeff
and Purl’s total
article rather than
to the model per se.

          RM: Then could you tell me what it was about

self-efficacy that you thought was explained by the whole
article?

Â

[From Rick Marken (2017.01.21.1200)]

···

Jeff Vancouver (2017.01.20.1130 EST)–

Â

RM: Since computational models are a form of working model, Vancouver and Purl are correct to say that Powers suggested the use of computational models, though he suggested this as the way to test control theory, not self-efficacy’s roles
in human behavior. But leaving that aside, Vancouver and Purl seem to have ignored the last part of Bill’s recommendation about modeling: "To devise working models, however, and especially to test them,
one needs experimental data that are more reliable than the customary results of research in the social sciencesâ€?.Â

Â

JV: Respectively disagree. Bill seemed to be saying that PCT and modeling could help clarify social cognitive theory,

RM: But you can’t do that when the data is random noise (unreliable), as is the Schmidt/DeShon data.Â

RM: There are two problems with testing a model against the data in the Schmidt and DeShon (2010) study. First, the results are averages over many subjects; PCT models are typically tested by comparing them to the behavior of one individual
at a time. Second, and more important, is the fact that the results in these experiments are highly unreliable. The predicted interaction between ambiguity and self-efficacy is statistically significant (p<.01) but the reliability of this prediction (measured
as R^2) is .03; only 3% of the variance in either dependent variable was accounted for by the interaction between ambiguity and self-efficacy. So the PCT model is being used to explain a very weak effect (indeed, essentially random noise) and one that is not
seen in a large proportion of the subjects tested; and in the subjects where the effect is seen, it is must usually be a very small effect, if it’s there at all.Â

Â

JV: Yes, the phenomenon explained is very weak, so the effect is very small. Some phenomena is weak.

Â

RM: Or it’s non-existent.Â

RM: Moreover, Vancouver and Purl don’t provide any measure of goodness of fit of the model to the Schmidt and DeShon data.

Â

JV: I did not have the data, so I could not fit individual data.

RM: Then there is no way of evaluating whether you “PCT” model is any better than any other model that could produce the same qualitative match to the data.Â

Â

RM: There are also problems with the Vancouver and Purl PCT model itself (Figure 2 in their diagram). One of the main problems is that it is not clear what the controlled variables.Â

Â

JV: The math used to represent the task progress input function is in the paper. Moreover, the variable in the environment called task state is defined. All the
variables used to explain the inputs in to the input function are explained. This comment seems to come down to you not reading the paper well enough.

RM: Why make it difficult for the reader to find what is the central feature of a PCT model – the proposal regarding the perceptual variable that is controlled?Â

JV: On the issue of the controlled variable for the dynamic expected utility variable I would agree that what is controlled is much more ambiguous. Indeed, I
would say it is not an accurate way of talking about that that subsystem does, though I am not completely convinced I have abstracted it fully. Does that mean that my model is not PCT? Not sure. It is inspired by PCT, but did it go outside of PCT, maybe. See
next response.

RM: I think the ambiguity of the assumed controlled variable is not a reason for saying that you model is not a PCT model. I think the non-PCT aspects of your model are the failure to clearly point out the variable(s) being controlled by the model (after all, PCT is a model that explains control) and the inclusion of model components  (like self-efficacy) that are not part of the model. As I noted in reply to Richard Pfau, the PCT model is designed to be consistent with what is known of how the nervous system works. So a PCT model will only contain functions that we know a nervous system can perform: perception, comparison, and output. The trick is to put those pieces together so that they produce behavior (controlling) that matches the controlling that is observed.

Â

 RM: There are other problems with the model but the basic problem with the application of PCT in this situation is that it is not being used to explain controlling. The term “controlled variable” is nowhere to be found in either the Vancouver
and Purl or the Schmidt and DeShon paper.Â

Â

JV: First, I have decided that the word control is very problematic awhile back, so I avoid it.

RM: Well that’s kind of too bad; it’s like avoiding the word “force” in physics or the word “evolution” in biology. Control is the phenomenon that PCT explains; it is Powers brilliant observation that behavior is control that made PCT necessary; it is the reason I (and Tim Carey) wrote “Controlling People”.Â

JV: That is, engineers tell me that the action the control takes is
the controlled variable. That action is to affect the system state so it gets to or remains at the referent. I can see how one could say that the system take is the controlled variable as Powers does, but it has created a sematic battle that seems unproductive
to wage.

RM: It’s not a semantic battle; it’s a scientific battle and it is very productive to wage it.

Â

 RM: Apparently J. Applied Psychology will occasionally publish Commentaries on published papers. I’m thinking of writing one on the Vancouver and Purl paper. Should I or would someone else like to do it.

Â

JV: If you make a comment on this, it seems that what would be productive is to pose the alternative.

RM: Will do. I don’t know if I will do it, but maybe.

JV: Rick, if you recall we went through this with my 2010/2014 model. You thought it was not true to the PCT core.

RM: I was weak:-)

Â

JV: You thought you could make a model that was
true to the PCT core and thus different than mine. I was excited about the possibility. We talked on the phone. You ended up not being able to envision a different model than mine once you understood my model. Shall we play that game again?

 RM: As I recall that model was, indeed, a reasonable PCT model. I think it had some incorrect things, like an external variable determining a reference. But I don’t recall that it had things in it like “self efficacy”. But my problem now is just the model presented in your paper, the one that Warren and Richard liked so much. I just wanted to express my dissent and explain why I don’t like it. Again, it’s because it contains components that are examples of “dormitive principles” rather than PCT and the data are essentially random noise.

Best

Rick

JV: On the issue of the explanation of self-efficacy, I think BA defended me well. I would also add that, as noted in the paper, the way self-efficacy belief develops in the mind is described in our 2014 paper, as noted in this paper. That
is, in this paper we took a short cut by making it a function of ability and bias whereas in the 2014 we should how it could develop and be a function of ability and where the bias might come from (though more work can clearly be done there). Because each
paper is constrained in terms of space, on can only do so much in a single paper.

Â

Somewhat yours always,

Â

Jeff Vancouver

Â

References

Â

Vancouver, J. B., Putka, D. J., & Scherbaum, C. A. (2005). Testing a computational model of the goal-level effect: An example of a neglected methodology. Organizational Research Methods, 8, 100-127.

Vancouver, J. B., Weinhardt, J. M., & Schmidt, A. M. (2010). A formal, computational theory of multiple-goal pursuit: Integrating goal-choice and goal-striving processes. Journal of Applied Psychology, 95, 985-1008.

Vancouver, J. B., Weinhardt, J.M., Vigo, R (2014). Change one can believe in: Adding learning to a computational model of self-regulation.
Organizational Behavior and Human Decision Processing, 124, 56-74.

Â

Â


Richard S. MarkenÂ

"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
                --Antoine de Saint-Exupery

[From Rick Marken (2017.01.21.1730)]

···

Martin Taylor (2017.01.20.23.10)–

MT: Could you tell the rest of us why you think there should be

something in the article that explains self-efficacy, or why you
want Richard to invent something that isn’t in the article?

RM: Actually, Richard Pfau is the one you should be asking. He’s the one who said “I found the Vancouver article…to be very worthwhile in helping to highlight PCT and bring it to broader attention, showing how it complements and helps explain self-efficiency.” I just asked what it was in the article that helped explain self-efficacy.Â

BestÂ

Rick

Â

I can't

find anything in the article that suggests an intent or attempt to
explain self-efficacy. After all, that’s not what the article claims
to be about, either in its title or, so far as I can see, in its
text. Here’s the first half of the abstract again:

-----------

Self-efficacy, which is one’s belief in one’s capacity, has been

found to both positively and negatively influence effort and
performance. The reasons for these different effects have been a
major topic of debate among social–cognitive and perceptual conntrol
theorists. In particular, the findings of various self-efficacy
effects has been motivated by a perceptual control theory view of
self-regulation that social–cognitive theorists’ questiion. To
provide more clarity to the theoretical arguments, a computational
model of the multiple processes presumed to create the positive,
negative, and null effects for self-efficacy is presented.

----------



I suppose though, if someone wrote an article suggesting that PCT

might explain why one keeps one’s had away from a hot oven burner,
you would be asking what about oven burner heating was explained by
the article. The lack of such an explanation would then be
sufficient reason to say the article was worthless, or worse, that
it was a misrepresentation of PCT. After all, every unfamiliar use
of PCT is by definition a misrepresentation of PCT, isn’t it?

Martin


Richard S. MarkenÂ

"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
                --Antoine de Saint-Exupery

Â

[From Jeff Vancouver (2017.01.23.12:45)]

···

[From Rick Marken (2017.01.21.1200)]

RM: But you can’t do that when the data is random noise (unreliable), as is the Schmidt/DeShon data.

JV: You know it is not random noise or it would not be statistically significant. You do not like that little variance is accounted for. My sense is this is the
cost of moving into the imagination mode modeling.

RM: Moreover, Vancouver and Purl don’t provide any measure of goodness of fit of the model to the Schmidt and DeShon data.

JV: I did not have the data, so I could not fit individual data.

RM: Then there is no way of evaluating whether you “PCT” model is any better than any other model that could produce the same qualitative match to the data.

JV: Though this is not true (fitting data quantitatively is not the only way to evaluate a model), it is why an alternative model would be welcome.

RM: There are also problems with the Vancouver and Purl PCT model itself (Figure 2 in their diagram). One of the main problems is that it is not clear what the controlled variables.

JV: The math used to represent the task progress input function is in the paper. Moreover, the variable
in the environment called task state is defined. All the variables used to explain the inputs in to the input function are explained. This comment seems to come down to you not reading the paper well enough.

RM: Why make it difficult for the reader to find what is the central feature of a PCT model – the proposal regarding the perceptual variable that is controlled?

JV: Because it is how I imagine the imagination mode works. Sorry for the inconvenience to your orthodoxy.

JV: On the issue of the controlled variable for the dynamic expected utility variable I would agree
that what is controlled is much more ambiguous. Indeed, I would say it is not an accurate way of talking about that that subsystem does, though I am not completely convinced I have abstracted it fully. Does that mean that my model is not PCT? Not sure. It
is inspired by PCT, but did it go outside of PCT, maybe. See next response.

RM: I think the ambiguity of the assumed controlled variable is not a reason for saying that you model is not a PCT model. I think the non-PCT aspects of your model are the failure to clearly point out the variable(s) being controlled by
the model (after all, PCT is a model that explains control) and the inclusion of model components (like self-efficacy) that are not part of the model. As I noted in reply to Richard Pfau, the PCT model is designed to be consistent with what is known of how
the nervous system works. So a PCT model will only contain functions that we know a nervous system can perform: perception, comparison, and output. The trick is to put those pieces together so that they produce behavior (controlling) that matches the controlling
that is observed.

JV: Perception, comparison, and output are not functions. They are the labels for the functions that describe a control system. They get their name from the role
they play in the system. The functions are additive, multiplicative, time integrator, etc. I did not include functions not describe in Chapter 3. Now here you seem to be saying that it is not okay to label a signal that is presumably used in forming an imagined
perception. I am not sure why that would be prohibited. That is, what self-efficacy represents is the weight that scales the output signal to the imagined perception. Would not such a weight be present in any imagination model system unless the output signal
perfectly aligned with the perceptual signal? Just because that weight has several implications that have led system-level psychologists to label and talk about it does not mean it cannot be in a PCT model or that I should label is something else (though I
have been tempted).

RM: There are other problems with the model but the basic problem with the application of PCT in this situation is that it is not being used to explain controlling. The term “controlled
variable” is nowhere to be found in either the Vancouver and Purl or the Schmidt and DeShon paper.

JV: First, I have decided that the word control is very problematic awhile back, so I avoid it.

RM: Well that’s kind of too bad; it’s like avoiding the word “force” in physics or the word “evolution” in biology. Control is the phenomenon that PCT explains; it is Powers brilliant observation that behavior is control that made PCT necessary;
it is the reason I (and Tim Carey) wrote “Controlling People”.

JV: That is, engineers tell me that the action the control takes is the controlled variable. That action
is to affect the system state so it gets to or remains at the referent. I can see how one could say that the system take is the controlled variable as Powers does, but it has created a sematic battle that seems unproductive to wage.

RM: It’s not a semantic battle; it’s a scientific battle and it is very productive to wage it.

JV: is the controlled variable the action the individual is doing to control the state of the variable or the variable whose state the individual controls via
their behavior? In PCT it is the latter whereas in engineering it is the former. These are not the same thing, should not be mixed up, and thus I get why you think it is a problem. My problem is two disciplines use the term for these two different things and
so I want to move away from using it at all.

JV: If you make a comment on this, it seems that what would be productive is to pose the alternative.

RM: Will do. I don’t know if I will do it, but maybe.

JV: I cannot decipher those two sentences.

RM: As I recall that model was, indeed, a reasonable PCT model. I think it had some incorrect things, like an external variable determining a reference. But I don’t recall that it had things in it like “self efficacy”. But my problem now
is just the model presented in your paper, the one that Warren and Richard liked so much. I just wanted to express my dissent and explain why I don’t like it. Again, it’s because it contains components that are examples of “dormitive principles” rather than
PCT and the data are essentially random noise.

JV: It did indeed have things in it like self-efficacy.

Jeff Vancouver

[From Rick Marken (2017.01.24.1530)]

···

Jeff Vancouver (2017.01.23.12:45)–

RM: But you can’t do that when the data is random noise (unreliable), as is the Schmidt/DeShon data.Â

Â

JV: You know it is not random noise or it would not be statistically significant. You do not like that little variance is accounted for. My sense is this is the
cost of moving into the imagination mode modeling.

RM: “Random noise” was probably a tad hyperbolic. But the the Schmidt/DeShon data is well below the level of quality suitable for testing PCT models. I mentioned the poor quality of the Schmidt/DeShon data because you had used a quote from Bill Powers to support your contention that your PCT model was superior to Bandura and Locke’s social-cognitive model because it was computational. The following was the only quote from the Powers (1991) article that I thought could possibly have served as the basis for your claim:

 BP: “Control theory is best used and tested through the method of modeling or simulation. To devise working models, however, and especially to test them, one needs experimental data that are more reliable than the customary results of research in the social sciencesâ€?.Â

RM: So while Bill did say that working (ie. computational) models were the best ones to use to test PCT, he also said that this is only true when the models are tested against data that is more reliable than the data that is customarily obtained in the social sciences. The Schmidt/DeShon data is certainly no more reliable – indeed, notably less reliable  – than that customarily obtained in research in the social sciences (I did a survey of journal articles and found the average R^2 value for behavioral data to be ~ .36; it looks like the average R^2 value in the Schmidt/DeShon data to be ~.05 ) . So using Bill’s quote as an endorsement of your approach was quite misleading.

Â

RM: Then there is no way of evaluating whether you “PCT” model is any better than any other model that could produce the same qualitative match to the data.Â

Â

JV: Though this is not true (fitting data quantitatively is not the only way to evaluate a model), it is why an alternative model would be welcome.

RM: I could easily come up with a computational model that is an alternative to yours. But the data are so noisy there would be no way to tell which is better.Â

RM: Why make it difficult for the reader to find what is the central feature of a PCT model – the proposal regarding the perceptual variable that is controlled?Â

JV: Because it is how I imagine the imagination mode works. Sorry for the inconvenience to your orthodoxy.

RM: If all the controlled variables in your model are controlled in imagination mode then how does the model perform the task? And even if controlled variables are imagined perceptions they can still be described.Â

JV: On the issue of the controlled variable for the dynamic expected utility variable I would agree
that what is controlled is much more ambiguous. Indeed, I would say it is not an accurate way of talking about that that subsystem does, though I am not completely convinced I have abstracted it fully. Does that mean that my model is not PCT? Not sure. It
is inspired by PCT, but did it go outside of PCT, maybe. See next response.

JV: Perception, comparison, and output are not functions. They are the labels for the functions that describe a control system. They get their name from the role
they play in the system. The functions are additive, multiplicative, time integrator, etc. I did not include functions not describe in Chapter 3. Now here you seem to be saying that it is not okay to label a signal that is presumably used in forming an imagined
perception. I am not sure why that would be prohibited. That is, what self-efficacy represents is the weight that scales the output signal to the imagined perception. Would not such a weight be present in any imagination model system unless the output signal
perfectly aligned with the perceptual signal? Just because that weight has several implications that have led system-level psychologists to label and talk about it does not mean it cannot be in a PCT model or that I should label is something else (though I
have been tempted).Â

RM: If self-efficacy is a weight that scales the output signal then “self-efficacy” is simply part of the output function. But it looks like the function of self-efficacy in the model is to adjust the weight of the output function. And it does this based on two inputs, bias and ability. So self-efficacy is a function itself that produces output (that weights the task output) based on these two inputs. So what are bias and ability? What part of a control system are they? Bias seems to depend on the trail block so it looks like a perceptual function. And ability is just sitting in the organism as a number generator apparently. So your model has a bunch of places to put in free parameters; as they say, with some effort you could probably get the model to produce behaviors that look like an elephant wagging its tail.Â

Also, the “perception of task progress” is a non-imaginary controlled variable that is a function of four inputs: information ambiguity, task state, time searching and task efficacy. That’s one heck of a variable. You really should explain why you picked that as the controlled variable and describe the evidence that that variable is actually controlled (by subjects and the model).Â

Â

RM: There are other problems with the model but the basic problem with the application of PCT in this situation is that it is not being used to explain controlling. The term “controlled
variable” is nowhere to be found in either the Vancouver and Purl or the Schmidt and DeShon paper.Â

JV: First, I have decided that the word control is very problematic awhile back, so I avoid it.

RM: What word do you use instead?

 RM: Well that’s kind of too bad; it’s like avoiding the word “force” in physics or the word “evolution” in biology. Control is the phenomenon that PCT explains; it is Powers brilliant observation that behavior is control that made PCT necessary;
it is the reason I (and Tim Carey) wrote “Controlling People”.

JV: That is, engineers tell me that the action the control takes is the controlled variable. That action
is to affect the system state so it gets to or remains at the referent. I can see how one could say that the system take is the controlled variable as Powers does, but it has created a sematic battle that seems unproductive to wage.

RM: So what do you call the controlled variable? Â

Â

RM: It’s not a semantic battle; it’s a scientific battle and it is very productive to wage it.

Â

JV: is the controlled variable the action the individual is doing to control the state of the variable or the variable whose state the individual controls via
their behavior? In PCT it is the latter whereas in engineering it is the former. These are not the same thing, should not be mixed up, and thus I get why you think it is a problem. My problem is two disciplines use the term for these two different things and
so I want to move away from using it at all.

RM: How do you talk about the PCT controlled variable then? Â

Â

JV: If you make a comment on this, it seems that what would be productive is to pose the alternative.

Â

RM: Will do. I don’t know if I will do it, but maybe.

Â

JV: I cannot decipher those two sentences.

RM: The “will do” referred to the fact that if I write the comment I will “pose the alternative”, though the alternative may simply be no alternative model since there doesn’t seem to be a control phenomenon here to model. The “I don’t know” referred to the fact that didn’t know whether I would actually write it. But now I’m pretty sure I will. Â

Best

Rick

Â

 RM: As I recall that model was, indeed, a reasonable PCT model. I think it had some incorrect things, like an external variable determining a reference. But I don’t recall that it had things in it like “self efficacy”. But my problem now
is just the model presented in your paper, the one that Warren and Richard liked so much. I just wanted to express my dissent and explain why I don’t like it. Again, it’s because it contains components that are examples of “dormitive principles” rather than
PCT and the data are essentially random noise.

Â

JV: It did indeed have things in it like self-efficacy.

Â

Jeff Vancouver


Richard S. MarkenÂ

"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
                --Antoine de Saint-Exupery

[From Jeff Vancouver (2017.01.25.0915)]

···

[From Rick Marken (2017.01.24.1530)]

RM: “Random noise” was probably a tad hyperbolic. But the the Schmidt/DeShon data is well below the level of quality suitable for testing PCT models. I mentioned the poor quality of the Schmidt/DeShon data because you had used a quote from
Bill Powers to support your contention that your PCT model was superior to Bandura and Locke’s social-cognitive model because it was computational. The following was the only quote from the Powers (1991) article that I thought could possibly have served as
the basis for your claim:

BP: “Control theory is best used and tested through the method of modeling or simulation. To devise working models, however, and especially to test them, one needs experimental data that are more reliable
than the customary results of research in the social sciences�.

RM: So while Bill did say that working (ie. computational) models were the best ones to use to test PCT, he also said that this is only true when the models are tested against data that is more reliable than
the data that is customarily obtained in the social sciences. The Schmidt/DeShon data is certainly no more reliable – indeed, notably less reliable – than that customarily obtained in research in the social sciences (I
did a survey of journal articles and found the average R^2 value for behavioral data to be ~ .36; it looks like the average R^2 value in the Schmidt/DeShon data to be ~.05 ) . So using Bill’s quote as an endorsement of your approach was quite misleading.

JV: There are many layers to testing a model. The first layer is simply showing that the model can produce the behavior it is intended to explain. This was the
primary objective of the Vancouver & Purl paper. That is, Bandura argued that the theory I was promoting could not possibly work to explain the kind of data that Schmidt and DeShon produced. The counter to that claim is a model that produces the kind of data
Schmidt and DeShon had. As we noted in the paper, I did not suggest that the data from one study was sufficient to prove the validity of the model. Clearly other models could be created that also reproduced the data. I would also suspect that the noisier the
data, the greater the number of models that could be built, but not sure about that. Meanwhile, models are also evaluated on their plausibility. This is where you have an issue, though I think it is you not understanding the model, which is troubling because
another main advantage of the modeling is the transparency of it. Personally, I wonder if this is like that case where the engineer showed other engineers the diagram of some functional system which the engineers could not identify by the diagram, but largely
reproduced with asked to diagram the function. I get that you are not very interested in producing a different model because you do not find the data used to confirm it very compelling. But I am engaged in a debate and the model I produced countered an argument
from the other side. So you can always complain about the quality of data from an experiment and I am not going to argue that more empirical work is needed. But to say one cannot present a model without a set of quality studies to back it up is to not recognize
the process of science. If you think my model is not plausible because of some element of it (as I presume you are good with much of it), then make the alternative part and defend it. That is what we do in science. So far, however, all I am hearing is that
you do not understand it.

RM: I could easily come up with a computational model that is an alternative to yours. But the data are so noisy there would be no way to tell which is better.

JV: Like I said the last time we had this debate, that is a stupid excuse. What is typically done is that someone comes up with an alternative model that explains
the same thing that the original model explains, and then someone (one of the protagonists or a third party) comes up with an empirical way to distinguish them. The quality of some extant data is not really all that relevant. However, without the two models,
it is difficult to determine what empirical protocol would distinguish them. So as I have said before, put your model where your mouth is.

RM: Why make it difficult for the reader to find what is the central feature of a PCT model – the proposal regarding the perceptual variable that is controlled?

JV: Because it is how I imagine the imagination mode works. Sorry for the inconvenience to your orthodoxy.

RM: If all the controlled variables in your model are controlled in imagination mode then how does the model perform the task? And even if controlled variables are imagined perceptions they can still be described.

JV: Who said all the control variables are in imagination mode? Imagination mode for the task agent is evoked with information about the current state of the
CV is blocked. This is what the information ambiguity parameter is for. Meanwhile, the choice agent is never in imagination mode though the quantity it is controlling might be from an agent in imagination mode.

JV: On the issue of the controlled variable for the dynamic expected utility variable I would agree
that what is controlled is much more ambiguous. Indeed, I would say it is not an accurate way of talking about that that subsystem does, though I am not completely convinced I have abstracted it fully. Does that mean that my model is not PCT? Not sure. It
is inspired by PCT, but did it go outside of PCT, maybe. See next response.

JV: Perception, comparison, and output are not functions. They are the labels for the functions that
describe a control system. They get their name from the role they play in the system. The functions are additive, multiplicative, time integrator, etc. I did not include functions not describe in Chapter 3. Now here you seem to be saying that it is not okay
to label a signal that is presumably used in forming an imagined perception. I am not sure why that would be prohibited. That is, what self-efficacy represents is the weight that scales the output signal to the imagined perception. Would not such a weight
be present in any imagination model system unless the output signal perfectly aligned with the perceptual signal? Just because that weight has several implications that have led system-level psychologists to label and talk about it does not mean it cannot
be in a PCT model or that I should label is something else (though I have been tempted).

RM: If self-efficacy is a weight that scales the output signal then “self-efficacy” is simply part of the output function.

JV: Not sure how you figure that. It is scaling the output to match the input. Scaling within the output function would mean that the output signal is scaled
to the input signal. Yet, they are very different signals given one is affecting the referents of lower level agents (control systems) and one is representing the state of a controlled variable.

RM: But it looks like the function of self-efficacy in the model is to
adjust the weight of the output function. And it does this based on two inputs, bias and ability. So self-efficacy is a function itself that produces output (that weights the task output) based on these two inputs. So what are bias and ability? What
part of a control system are they? Bias seems to depend on the trail block so it looks like a perceptual function. And ability is just sitting in the organism as a number generator apparently. So your model has a bunch of places to put in free parameters;
as they say, with some effort you could probably get the model to produce behaviors that look like an elephant wagging its tail.

JV: As I said before (and in the paper), consult our 2014 paper to see how self-efficacy comes about. I was trying to keep the modeling in this paper simple for
the audience. I could see how you might have some additional questions about this (and particularly bias) after consulting that paper. But I will not address this until you do your due diligence.

Vancouver, J. B.,
Weinhardt, J.M., Vigo, R (2014). Change one can believe in: Adding learning to a computational model of self-regulation.
Organizational Behavior and Human Decision Processing, 124, 56-74.

RM: Also, the “perception of task progress” is a non-imaginary controlled variable that is a function of four inputs: information ambiguity, task state, time searching and task efficacy. That’s one heck
of a variable. You really should explain why you picked that as the controlled variable and describe the evidence that that variable is actually controlled (by subjects and the model).

JV: So you are not reading the paper. Perception of task progress is a non-imaginary controlled variable where the input function is simply task progress. However,
when the task progress is unavailable (information is ambiguous), then it uses imagination mode to generate the perception. That is what the other inputs are about. Again, if imagination mode has been modeled in a different way elsewhere, or you have a different
way of modeling it, then we might have something to talk about.

RM: There are other problems with the model but the basic problem with the application of PCT in this situation is that it is not being used to explain controlling. The term “controlled
variable” is nowhere to be found in either the Vancouver and Purl or the Schmidt and DeShon paper.

JV: First, I have decided that the word control is very problematic awhile back, so I avoid it.

RM: What word do you use instead?

JV: regulate.

RM: Well that’s kind of too bad; it’s like avoiding the word “force” in physics or the word “evolution” in biology. Control is the phenomenon that PCT explains; it is Powers brilliant
observation that behavior is control that made PCT necessary; it is the reason I (and Tim Carey) wrote “Controlling People”.

JV: That is, engineers tell me that the action the control takes is the controlled variable. That action
is to affect the system state so it gets to or remains at the referent. I can see how one could say that the system take is the controlled variable as Powers does, but it has created a sematic battle that seems unproductive to wage.

RM: So what do you call the controlled variable?

JV: The state variable.

RM: It’s not a semantic battle; it’s a scientific battle and it is very productive to wage it.

JV: is the controlled variable the action the individual is doing to control the state of the variable
or the variable whose state the individual controls via their behavior? In PCT it is the latter whereas in engineering it is the former. These are not the same thing, should not be mixed up, and thus I get why you think it is a problem. My problem is two disciplines
use the term for these two different things and so I want to move away from using it at all.

RM: How do you talk about the PCT controlled variable then?

JV: As a model of self-regulation. You have seen my paper on this. In particular, I pitted a model of self-regulation of action against a model of self-regulation
of perception and showed how the self-regulation of perception model was superior.

Vancouver, J. B. & Scherbaum, C. A. (2008). Do We Self-Regulate Actions or Perceptions? A Test of Two Computational
Models. Computational and Mathematical Organizational Theory, 14, 1-22.

JV: If you make a comment on this, it seems that what would be productive is to pose the alternative.

RM: Will do. I don’t know if I will do it, but maybe.

JV: I cannot decipher those two sentences.

RM: The “will do” referred to the fact that if I write the comment I will “pose the alternative”, though the alternative may simply be no alternative model since there doesn’t seem to be a control phenomenon here to model. The “I
don’t know” referred to the fact that didn’t know whether I would actually write it. But now I’m pretty sure I will.

JV: By no control phenomenon you must mean no phenomenon, because behavior, unless random, is the control of perceptions. So you seem to be returning to your
original position that the Schmidt and DeShon data is nothing but noise.

Jeff

Down in text….

···

From: Martin Taylor [mailto:mmt-csg@mmtaylor.net]
Sent: Saturday, January 21, 2017 6:20 PM
To: csgnet@lists.illinois.edu
Subject: Re: Scholar Alert - [ “perceptual control theory” ]

[Martin Taylor 2017.01.20.23.10]

[From Rick Marken (2017.01.20.1745)]

Richard Pfau (2017.01.20 10:15 EST)

RP: Ref: [Rick Marken (2016.01.10 1520)] “I would be interested in hearing what it was about self-efficacy that you think is explained by Jeff’s model.”

RP: Actually, I was referring to Jeff and Purl’s total article rather than to the model per se.

RM: Then could you tell me what it was about self-efficacy that you thought was explained by the whole article?

Could you tell the rest of us why you think there should be something in the article that explains self-efficacy, or why you want Richard to invent something that isn’t in the article? I can’t find anything in the article that suggests an intent or attempt to explain self-efficacy. After all, that’s not what the article claims to be about, either in its title or, so far as I can see, in its text. Here’s the first half of the abstract again:

Self-efficacy, which is one’s belief in one’s capacity, has been found to both positively and negatively influence effort and performance. The reasons for these different effects have been a major topic of debate among social–cognitive and perceptual control theeorists. In particular, the findings of various self-efficacy effects has been motivated by a perceptual control theory view of self-regulation that social–cognitive theorists’ question. To provide more clarityy to the theoretical arguments, a computational model of the multiple processes presumed to create the positive, negative, and null effects for self-efficacy is presented.

HB : I didn’t read the article but the statement :

JV : In particular, the findings of various self-efficacy effects has been motivated by a perceptual control theory view of self-regulation that social–cognitive theorists’ question.

HB : What is PCT view of self-regulation ? Where can we find this in Bills’ literature ?

Who are social–cognitive theorists and what are they questioning about PCT viiew of self-regulation ? What is »Social Cognition« ???

When I and Jeff had a quite long talk some years behind Jeff wanted to sort PCT into self-regulation theories. I warned him that self-regulation theories spring from PCT : relations Powers-Carver, what Mary Powers described very clearly. But Jeff is still spreading wrong infromations around that there is some PCT view of self-regulation. PCT and self-regulation has little in common except that Carver made a critical distortion to PCT. It’s just a »branched blind street« on the PCT highway.

JV : To provide more clarity to the theoretical arguments, a computational model of the multiple processes presumed to create the positive, negative, and null effects for self-efficacy is presented.

HB : To provide clarity to what »theoretical arguments« ??? What could be a »computational model« of the »multiple processes« ??? Which »multiple processes« ??? And what could be positive, negative, and null effects for »self-efficacy« from the PCT view of self-regulation ???

I’m sorry to say but it seems a mess to me. I’ve read many Jeffs’ articles and we established through our long talk that his »human« model can’t survive. So it’s wrong. We also established that Jeff is very weak in physiological knowledge and he can’t support his modeling with substancial proves. And on that "weak« bases Jeff is trying to make a computational model for a long time now. Many basic assumptions about organisms functioning are wrong in »self-efficacy« or »self-regulation« (Carver) model. So although I hate being on Rick side, he could be right this time.

Computational model is probably one of solutions in clarifying theoretical problems. But not for any price. I think that first firm qualitative theoretical ground has to be made to provide the bases for quantitative, computational model. And I don’t see that stable theoretical ground. First modeling organism has to be desingned in such way that we can theoretically assume that organism can survive.Â

Boris

P.S. for more detailed critics I should read article. I’m sorry if I misjudged something on the bases of previous Jeffs’ articles.

I suppose though, if someone wrote an article suggesting that PCT might explain why one keeps one’s had away from a hot oven burner, you would be asking what about oven burner heating was explained by the article. The lack of such an explanation would then be sufficient reason to say the article was worthless, or worse, that it was a misrepresentation of PCT. After all, every unfamiliar use of PCT is by definition a misrepresentation of PCT, isn’t it?

Martin

[From Rick Marken (2017.01.25.1120 PST)]

···

Jeff Vancouver (2017.01.25.0915)

Â

RM: So while Bill did say that working (ie. computational) models were the best ones to use to test PCT, he also said that this is only true when the models are tested against data that is more reliable than
the data that is customarily obtained in the social sciences.Â

Â

JV: There are many layers to testing a model. The first layer is simply showing that the model can produce the behavior it is intended to explain.

RM: Actually, the first step is to show that the behavior it is intended to explain actually happens.Â

JV: This was the
primary objective of the Vancouver & Purl paper. That is, Bandura argued that the theory I was promoting could not possibly work to explain the kind of data that Schmidt and DeShon produced.

RM: That’s like getting in an argument with Trump about the source of the 3 million fraudulent votes for Clinton. I’m not interested in battling with people who are trying to explain non-existent phenomena. The evidence for an effect of belief in one’s self-efficacy on performance is extremely weak. PCT is a theory for the fact-based community, which is usually people who are considered liberals, because facts have a distinctly liberal bias. I’m pretty sure Bandura and Locke are right wingers so they would be expected to be comfortable in the world of “alternative facts”.

JV: Â But I am engaged in a debate and the model I produced countered an argument
from the other side.

RM: You were joining the wrong debate. You should have debated their facts, not their theory.Â

Â

RM: I could easily come up with a computational model that is an alternative to yours. But the data are so noisy there would be no way to tell which is better.Â

Â

JV: Like I said the last time we had this debate, that is a stupid excuse.

RM: I think it’s a very good “excuse” but, then, maybe I’m just being stupid. I just find it to be enormously more satisfying to try to develop models that explain what I know to be facts. That was one of the things that was most attractive to me about PCT; it was about “phenomena first”. Inventing models to explain random noise is not my idea of a good time (unless I’m in Vegas, but I never go since gambling is the one vice I never really got into). Examples of how to test alternative models are described in my book “Doing Research on Purpose”. A demonstration of the PCT approach to testing alternative models that is not in that book is in this paper:Â

https://www.dropbox.com/s/eymkj4bxuorpyuy/Chasin%27Choppers.pdf?dl=0

RM: If self-efficacy is a weight that scales the output signal then “self-efficacy” is simply part of the output function.

JV: Not sure how you figure that. It is scaling the output to match the input.

RM: Then the self-efficacy component of the model is a control system controlling the perception of something like the ratio of the output to the input variable. But neither the output nor the input variable is shown as an input to self-efficacy. If a system cannot perceive the variable it is controlling (in this case the degree of match of output to input) then it can’t control it; it can’t act so as to maintain the match in the reference state, “matched”.

Â

RM: What word do you use instead [of ‘control’]?

Â

JV: regulate.Â

Â

RM: So what do you call the controlled variable? Â

Â

JV: The state variable.

RM: I think it would make a lot more sense, then, to call the controlled variable the “regulated variable”. Â

RM: How do you talk about the PCT controlled variable then? Â

Â

JV: As a model of self-regulation. You have seen my paper on this. In particular, I pitted a model of self-regulation of action against a model of self-regulation
of perception and showed how the self-regulation of perception model was superior.

Â

Vancouver, J. B. & Scherbaum, C. A. (2008). Do We Self-Regulate Actions or Perceptions? A Test of Two Computational
Models. Computational and Mathematical Organizational Theory, 14, 1-22.

RM: I must have misplaced it. Could you send me a copy, or a link to a copy. I’d appreciate it. Â

Â

JV: By no control phenomenon you must mean no phenomenon, because behavior, unless random, is the control of perceptions.

RM: No, what I mean is that you have not determined what variable is being controlled (regulated) and, thus, have no way of determining how well that variable is being controlled.Â

Â

JV: So you seem to be returning to your
original position that the Schmidt and DeShon data is nothing but noise.

RM: Not “nothing but” but so noisy as to be worthless for model testing.Â

RM: I think you could make this research a lot less noisy by using a simple control task, like a tracking task, as the example of control (regulation) behavior. We know how to model it extremely well because we have a very good idea of what variable is being controlled. Then you could see whether beliefs in self-efficacy have any effect on how well people control using a procedure I suggested in an earlier post. Once a participant is trained up on the task then go through a sequence of trials where, before each trial, the person is asked how well she believes she will do; then see if there is any relationship between the estimate of self-efficacy and how well she controls the cursor/target relationship. I suggest that you start simple (like Galileo and Powers) and see if there is any there there before coming up with these Rube Goldberg models to account for rather questionable results.Â

Best regards

Rick


Richard S. MarkenÂ

"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
                --Antoine de Saint-Exupery

[Martin Taylor 2017.01.25.14.26]

Sorry. You are quite right. My mistake.

Martin

···

On 2017/01/21 8:28 PM, Richard Marken
wrote:

[From Rick Marken (2017.01.21.1730)]

Martin Taylor (2017.01.20.23.10)–

            MT: Could you tell the rest of us why you think there

should be something in the article that explains
self-efficacy, or why you want Richard to invent
something that isn’t in the article?

          RM: Actually, Richard Pfau is the one you should be

asking. He’s the one who said " I
found the Vancouver article…to be very worthwhile in
helping to highlight PCT and bring it to broader
attention, showing how it complements and helps explain
self-efficiency." I just asked what it was in the
article that helped explain self-efficacy.

Hi there, things are moving quickly! See the first reference…

Warren

···

Begin forwarded message:

From: Google Scholar Alerts scholaralerts-noreply@google.com
Date: 15 March 2015 23:13:57 GMT
To: wmansell@gmail.com
Subject: Scholar Alert - [ “perceptual control theory” ]

Scholar Alert: [ “perceptual control theory” ]

How the credit assignment problems in motor control could be solved after the cerebellum predicts increases in error

SO Verduzco-Flores, RC O’Reilly - Frontiers in Computational Neuroscience, 2015

In the context of Threshold Control Theory and Perceptual Control Theory we show how
to extend our model so it implements anticipative corrections in cascade control systems
that span from muscle contractions to cognitive operations.

[PDF] What are the knowledge needs of teachers to inform the multimodal authoring of their students in years 1–8?

<
SJ Bedford

Page 1. http://researchspace.auckland.ac.nz ResearchSpace@Auckland Copyright
Statement The digital copy of this thesis is protected by the Copyright Act 1994 (New
Zealand). This thesis may be consulted by you, provided


This Google Scholar Alert is brought to you by Google.

Cancel alert
List my alerts

PS I don’t think the TCT group will be happy about their model being labelled as negative feedback control!

···

On 16 Mar 2015, at 11:00, Warren Mansell wmansell@gmail.com wrote:

Hi there, things are moving quickly! See the first reference…

Warren

Begin forwarded message:

From: Google Scholar Alerts scholaralerts-noreply@google.com
Date: 15 March 2015 23:13:57 GMT
To: wmansell@gmail.com
Subject: Scholar Alert - [ “perceptual control theory” ]

Scholar Alert: [ “perceptual control theory” ]

How the credit assignment problems in motor control could be solved after the cerebellum predicts increases in error

SO Verduzco-Flores, RC O’Reilly - Frontiers in Computational Neuroscience, 2015

In the context of Threshold Control Theory and Perceptual Control Theory we show how
to extend our model so it implements anticipative corrections in cascade control systems
that span from muscle contractions to cognitive operations.

[PDF] What are the knowledge needs of teachers to inform the multimodal authoring of their students in years 1–8?

<
SJ Bedford

Page 1. http://researchspace.auckland.ac.nz ResearchSpace@Auckland Copyright
Statement The digital copy of this thesis is protected by the Copyright Act 1994 (New
Zealand). This thesis may be consulted by you, provided


This Google Scholar Alert is brought to you by Google.

Cancel alert
List my alerts