# Changing the foundations of PCT (was Testing for Control In Experiments)

**URL:** <http://discourse.iapct.org/t/changing-the-foundations-of-pct-was-testing-for-control-in-experiments/2452>\
**Category:** CSG2010\
**Created:** [August 11, 2010, 6:32pm UTC](http://discourse.iapct.org/t/changing-the-foundations-of-pct-was-testing-for-control-in-experiments/2452 "2010-08-11T18:32:52Z")\
**Posts on this page:** 12\
**Page:** 1

<div class="post-metadata">

**Author:** ![MartinT](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/m/ee7513/32.png) [@MartinT](http://discourse.iapct.org/u/MartinT)\
**Post date:** [August 11, 2010, 6:32pm UTC](http://discourse.iapct.org/t/changing-the-foundations-of-pct-was-testing-for-control-in-experiments/2452/1 "2010-08-11T18:32:52Z")

</div>

[Martin Taylor 2010.08.11.14.06]

A few days ago, Bill sent a series of very strange messages to which my reaction has varied from day to day as I contemplated them. Apart from one obviously correct technical point that I will deal with in my experimental model on another thread, I was initially flabbergasted, and since then I have been by turns annoyed, puzzled, worried and generally at a loss as to how to respond. I will not respond to the main themes here. This message is only about Bill's radical redefinition of PCT, which once meant "PERCEPTUAL Control Theory", and now seems to mean something quite different, at least to Bill.

> [From Bill Powers (2010.08.07.0210 MDT)]RM: I also don't agree with calling p[p[S]-A] a controlled variable.
> 
> > MMT: Of course it is a controlled variable. Just as with any control system, the output signal influences the perceptual signal, and changes its value to approach its reference value. What's not a controlled variable about that?
> > 
> > > RM: Bill addressed this in his comments earlier. A controlled variable (in PCT) has an environmental correlate, the controlled quantity, which can be detected by an experimenter.
> > 
> > MMT: The controlled quantity is ALWAYS a perception, never an environmental variable. That's PCT 101, first lesson.
> 
> BP: Here we go again. Rick is defining a controlled variable as something the external observer can see. If you would read his words that would be obvious.

Yes, it was blindingly obvious.

It was equally obvious that the idea violates the foundations of PCT, which is why I wrote my correction: "The controlled quantity is ALWAYS a perception, never an environmental variable. That's PCT 101, first lesson."

If for some reason you no longer think that perceptions are the only variables that can be controlled, and that environmental variables ("something the external observer can see") also can be controlled, I'm afraid I cannot agree. Your PCT has fallen into the supernatural void. Failing some clairvoyant abilities in the subject that would allow the true state of an externally observable variable to serve as the input to the comparator of a control system, the perceptual signal is the only way that the control unit can get information about the state of ANYTHING the external observer can see. I know you have considerable disdain for "conventional" science and scientists, but I think this change to the foundation of PCT goes too far beyond the bounds of normal (or quantum) physics, even for you.

If, as has seemed the case recently, you are controlling strongly for perceiving me to be wrong in just about everything I write, nevertheless it seems a bit extreme to use your authority as originator to alter the very foundations of Perceptual Control Theory as a means of controlling that perception. I don't accept it, and if this leads to a conflict (something I usually try to avoid), so be it.

For my part, I will continue to write about PCT as though only perceptions can be controlled.

Martin

---

<div class="post-metadata">

**Author:** ![Bill\_Powers1](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/b/8e7dd6/32.png) [@Bill\_Powers1](http://discourse.iapct.org/u/Bill_Powers1)\
**Post date:** [August 11, 2010, 8:18pm UTC](http://discourse.iapct.org/t/changing-the-foundations-of-pct-was-testing-for-control-in-experiments/2452/2 "2010-08-11T20:18:19Z")

</div>

[From Bill Powers (2010.08.11.1245 MDT)]

Martin Taylor 2010.08.11.14.06 –

> A few days ago, Bill sent a  
> series of very strange messages to which my reaction has varied from day  
> to day as I contemplated them. Apart from one obviously correct technical  
> point that I will deal with in my experimental model on another thread, I  
> was initially flabbergasted, and since then I have been by turns annoyed,  
> puzzled, worried and generally at a loss as to how to respond. I will not  
> respond to the main themes here. This message is only about Bill’s  
> radical redefinition of PCT, which once meant “PERCEPTUAL Control  
> Theory”, and now seems to mean something quite different, at least  
> to Bill.

That was not the intended import of my message at all. I was pointing out  
that Rick was defining a controlled variable as an externally-observed or  
deduced variable, whereas you were defining it as a perceptual signal  
representing the value of a controlled variable. This accounted for most  
of the disagreements between you. The only profitable course, it would  
seem, would be for you and Rick to discuss this definitional problem and  
resolve it to your mutual satisfaction. I wasn’t saying what my opinion  
would have been.

I am not the PCT policeman. I have my own way of sorting out meanings and  
I try to be consistent, but that is not binding on anyone but me. If  
asked, I will explain why I choose my way, and I will be willing to  
listen to counterproposals. But I don’t automatically assume I am  
right.

The following interchange shows that you, apparently, are not willing to  
have this discussion with Rick:

> > [RM: I also don’t agree with  
> > calling p[p[S]-A] a controlled variable.
> > 
> > > MMT: Of course it is a controlled variable. Just as with any control  
> > > system, the output signal influences the perceptual signal, and changes  
> > > its value to approach its reference value. What’s not a controlled  
> > > variable about that?

Your intent appears to be to bully Rick into agreeing that you are right  
with no explanations or discussion and no respect for Rick’s point of  
view, as well as no understanding of it. Rick’s definition also fits your  
definition of a controlled variable: the controlled input quantity as the  
observer sees it, is affected by the system’s output and its value is  
thereby changed to approach a reference level (remembering that reference  
_level_ is defined in terms of the observer’s understanding of the  
controlled quantity and its intended or preferred value).

> > BP: Here we go again. Rick is  
> > defining a controlled variable as something the external observer can  
> > see. If you would read his words that would be obvious.
> 
> MMT: Yes, it was blindingly obvious.
> 
> It was equally obvious that the idea violates the foundations of PCT,  
> which is why I wrote my correction: “The controlled quantity is  
> ALWAYS a perception, never an environmental variable. That’s PCT 101,  
> first lesson.”

BP: So in your opinion, it violates the foundations of PCT. In fact, your  
way of saying it violates my attempt to put some order in the way we talk  
about controlled variables by reserving the term “quantity” to  
mean observable physical variables outside the system, and  
“signal” to mean variables inside the system.  
“Variable” applies equally inside and outside, so can’t be used  
to distinguish the observer’s view from the organism’s view.  
In fact neither the observer nor the control system can directly perceive  
the nature of the controlled variable itself. Suppose the input function  
is k1_x1 + k2_x2. The perceptual signal inside the system can’t represent  
this function, because the signal is just a scalar variable with a value  
like 3.6 or 135 (impulses per second as the neurologist sees it). The  
observer can’t see this function, either, because all that is actually  
visible from outside are x1 and x2. An infinity of different combinations  
of x1 and x2 can produce any given magnitude of perceptual signal. Only  
if the observer’s input device applies the same function to x1 and x2  
that the observed system applies will the observer’s perceptual signal  
covary with the system’s. Then both observer and system will perceive the  
same _magnitude_ of the controlled variable, but neither will know  
what is actually being controlled.

To know what is being controlled, one must know the internal structure of  
the perceptual input function. Evidently we can perceive, at some level,  
that sort of structure. That may happen at the relationship level, if we  
consider “relationship” to include the concept of  
“function.” But this perception is part of a model, not a  
direct perception of what goes on in the input function.

> MMT: If for some reason you no  
> longer think that perceptions are the only variables that can be  
> controlled, and that environmental variables (“something the  
> external observer can see”) also can be controlled, I’m afraid I  
> cannot agree. Your PCT has fallen into the supernatural void.

The external observer can, given a set of lower-level variables and some  
methodology, deduce what function of them is being controlled. Given a  
target position T and a cursor position C, the observer can see that a  
controlled variable can be defined as C - T. But the observer  
_perceives_ only the distance between the target and cursor, a  
single scalar value, just as the person doing the tracking does.

Of course the observer can be wrong, especially if the correlations in  
the data are low. It’s common for several equally defensible guesses  
about the nature of a controlled variable to exist.

> MMT: Failing some clairvoyant  
> abilities in the subject that would allow the true state of an externally  
> observable variable to serve as the input to the comparator of a control  
> system, the perceptual signal is the only way that the control unit can  
> get information about the state of ANYTHING the external observer can  
> see.

BP: Yes, and this is also the only way the observer can get information  
about whatever is being controlled.

> MMT: I know you have  
> considerable disdain for “conventional” science and scientists,  
> but I think this change to the foundation of PCT goes too far beyond the  
> bounds of normal (or quantum) physics, even for you.

BP: I have disdain for sloppy thinking that is presented as the word of  
God. Sloppy thinking alone I can usually forgive or ignore. Actually,  
some of my best friends are scientists, and they don’t do that very  
often, as far as I know. I also have disdain for conventional science,  
but “disdain” isn’t quite the right word. I’m willing to admit  
that we are all doing our best and are honest about it, most of the time.  
But human beings always add a lot of puffery to their own  
accomplishments, so we have to discount a lot of the claims that are made  
about what scientists say they know or understand. My greatest portion of  
disdain is reserved for the puffery.

> MMT: If, as has seemed the case  
> recently, you are controlling strongly for perceiving me to be wrong in  
> just about everything I write, nevertheless it seems a bit extreme to use  
> your authority as originator to alter the very foundations of Perceptual  
> Control Theory as a means of controlling that perception. I don’t accept  
> it, and if this leads to a conflict (something I usually try to avoid),  
> so be it.

BP: My objectives are strictly a matter of your interpretation in this  
case. I was not favoring Rick’s view, only pointing out that it exists  
and deserves to be dealt with. I didn’t say what the  
“authoritative” view would be.

> MMT: For my part, I will  
> continue to write about PCT as though only perceptions can be  
> controlled.

I sympathize with your frustration, even if it’s mostly your own doing.  
Reorganizing is uncomfortable. Most of your misinterpretations, of which  
there have been many lately, have brought out hidden assumptions that  
have now become the subject of discussion. That’s where the heat has been  
coming from, not from any outside source.

Best,

Bill P.

---

<div class="post-metadata">

**Author:** ![MartinT](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/m/ee7513/32.png) [@MartinT](http://discourse.iapct.org/u/MartinT)\
**Post date:** [August 18, 2010, 8:02pm UTC](http://discourse.iapct.org/t/changing-the-foundations-of-pct-was-testing-for-control-in-experiments/2452/3 "2010-08-18T20:02:46Z")

</div>

[Martin Taylor 2010.08.18. 14.48]

```
I'm still out of town, but I have just read the following:

```

> [From Bill Powers (2010.08.11.1245 MDT)]
> 
> ```
> Martin Taylor 2010.08.11.14.06 --
> 
> ```
> 
> > ```
> > A few days ago, Bill
> > 
> > ```
> > 
> > sent a  
> > series of very strange messages to which my reaction has varied  
> > from day  
> > to day as I contemplated them. Apart from one obviously correct  
> > technical  
> > point that I will deal with in my experimental model on another  
> > thread, I  
> > was initially flabbergasted, and since then I have been by turns  
> > annoyed,  
> > puzzled, worried and generally at a loss as to how to respond. I  
> > will not  
> > respond to the main themes here. This message is only about  
> > Bill’s  
> > radical redefinition of PCT, which once meant “PERCEPTUAL  
> > Control  
> > Theory”, and now seems to mean something quite different, at  
> > least  
> > to Bill.
> 
> ```
> That was not the intended import of my message at all. I was
> 
> ```
> 
> pointing out  
> that Rick was defining a controlled variable as an  
> externally-observed or  
> deduced variable, whereas you were defining it as a perceptual  
> signal  
> representing the value of a controlled variable.

```
When you say "A perceptual signal representing the value of a

```

controlled variable", it sounds very much to me as though you are  
saying that “the controlled variable” is NOT the value of the  
perceptual signal.

```
For nearly two decades, I have been under a misapprehension, then,

```

that the ONLY controlled variable in PCT was the value of the  
perceptual signal. My understanding of basic physics, however, leads  
me to believe that whatever the Powers-Marken version of PCT may  
from time to time redetermine, the science of control argues that my  
twenty-year understanding is the correct one. I see no possibility  
for controlling anything other than the value of a perceptual  
signal, an implication carried even by the very title: “Behaviour,  
the control of perception”.

```
I had expected that rather than jumping down my throat for

```

(correctly) calling Rick out when he “defined” a controlled variable  
as something detectable by an external observer, you would remind  
him that “Perceptual Control Theory” demands that the controlled  
variable is and is only a perception.

> ```
> This accounted for most
> 
> ```
> 
> of the disagreements between you.

```
Indeed. You aren't the only person who noted that fact.

```

> ```
> The following interchange shows that you, apparently, are not
> 
> ```
> 
> willing to  
> have this discussion with Rick:
> 
> > > ```
> > > [RM: I also don't
> > > 
> > > ```
> > > 
> > > agree with  
> > > calling p[p[S]-A] a controlled variable.
> > > 
> > > > ```
> > > > MMT: Of course it is a controlled variable. Just as with any
> > > > 
> > > > ```
> > > > 
> > > > control  
> > > > system, the output signal influences the perceptual signal,  
> > > > and changes  
> > > > its value to approach its reference value. What’s not a  
> > > > controlled  
> > > > variable about that?
> 
> ```
> Your intent appears to be to bully Rick into agreeing that you are
> 
> ```
> 
> right  
> with no explanations or discussion and no respect for Rick’s point  
> of  
> view, as well as no understanding of it.

```
I think I included a complete explanation, didn't I. You even quoted

```

it: “the output signal influences the perceptual signal, and changes  
its value to approach its reference value.” What more explanation is  
needed?

```
I understand Rick's point of view very well (so far as I am able to

```

determine). His point of view is that there is no control if the  
input to the perceptual function of a control system is not  
observable by an external observer. I “bullied” Rick by referring to  
a foundational principle of PCT, one you expound at some length  
later in the message to which I am responding, that the perceptual  
function has no way of knowing where its signals come from, and  
still less does the controlled variable (the perceptual signal)  
convey any such information.

```
I rather suspect that if the roles were reversed, and it was Rick

```

reminding me rather than me reminding Rick, that all behaviour is  
the control of perception and not of environmental variables, it  
would not be Rick you would be correcting.

> ```
> Rick's definition also fits your
> 
> ```
> 
> definition of a controlled variable: the controlled input quantity  
> as the  
> observer sees it, is affected by the system’s output and its value  
> is  
> thereby changed to approach a reference level (remembering that  
> reference  
> _level_ is defined in terms of the observer’s understanding  
> of the  
> controlled quantity and its intended or preferred value).

```
All you say here is that perceptual variables are influenced by

```

feedback pathways from the control system’s output, without  
reference to how those signals get from the output to the perceptual  
input function. By implication, it is possible that on some  
occasions those feedback pathways may go through the external  
environment and create effects observable by an external observer.  
Nobody has disputed any of that, except possibly Rick. (Or, maybe  
that’s not what you meant, given your later distinction between  
“quantity” and “variable”.) Rick’s “definition” says that ALL  
control is of the environmental variables (or, giving an extremely  
charitable interpretation of preceding statements by Rick, of  
perceptual variables whose values are determined by externally  
observable variables).

```
I'm puzzled by "reference

```

_level_ is defined in terms of the observer’s understanding of  
the  
controlled quantity and its intended or preferred value". What does  
an observer have to do with the reference level of a control unit?  
To bring an observer into the definition of “reference level” seems  
very odd. Hitherto, it has just been a value input to the comparator  
of a control unit, for comparison with the perceptual signal value.  
No observer needed or implied. Where did this “observer” in the  
definition spring from?

> > > ```
> > > BP: Here we go
> > > 
> > > ```
> > > 
> > > again. Rick is  
> > > defining a controlled variable as something the external  
> > > observer can  
> > > see. If you would read his words that would be obvious.
> > 
> > ```
> > MMT: Yes, it was blindingly obvious.
> > 
> > It was equally obvious that the idea violates the foundations of
> > 
> > ```
> > 
> > PCT,  
> > which is why I wrote my correction: “The controlled quantity is  
> > ALWAYS a perception, never an environmental variable. That’s PCT  
> > 101,  
> > first lesson.”
> 
> ```
> BP: So in your opinion, it violates the foundations of PCT.
> 
> ```

```
But it doesn't violate yours? My understanding that this was the

```

case was the reason I changed the subject line to “Changing the  
foundations of PCT”.

```
Here's how I bullied Rick in my continuation of "The controlled

```

quantity is ALWAYS a perception, never an environmental variable.  
That’s PCT 101, first lesson." The bullying commences thus:“as Bill  
corrected me in the partner thread, the output of a control unit  
knows or cares nothing about where its signal goes. Control exists  
if the output influences the perception. If the influence of the  
output reaches the perception through the external environment, so  
be it. If the influence of the output reaches the perception by way  
of an internal connection we call imagination, so be it. If the  
influence of the output reaches the perception both through the  
external environment and through internal pathways, so be it. Any  
which way, it’s irrelevant to the control unit.”

```
But of course, that bullying comment contradicts Rick's definition,

```

for which I suppose I should apologise.

> ```
> In fact, your
> 
> ```
> 
> way of saying it violates my attempt to put some order in the way  
> we talk  
> about controlled variables by reserving the term “quantity” to  
> mean observable physical variables outside the system, and  
> “signal” to mean variables inside the system.  
> “Variable” applies equally inside and outside, so can’t be used  
> to distinguish the observer’s view from the organism’s view.

```
??? Who cares about the observer's view unless the observer is part

```

of the object of study? When you are dealing with the  
observer-observed system, of course the observer matters. But why  
bring the observer into a discussion of which variable is controlled  
in a single control system?

```
Is using the technically correct "variable" in place of your

```

idiosyncratically chosen “quantity” justification for your disputing  
that “defining a controlled variable as something the observer can  
see is a violation of the foundations of PCT?” Is that definition a  
violation of the foundations of PCT, or is it not?

```
In what follows you begin to support my assertion that it is...

```

> ```
> In fact neither the observer nor the control system can directly
> 
> ```
> 
> perceive  
> the nature of the controlled variable itself. …

> ```
> To know what is being controlled, one must know the internal
> 
> ```
> 
> structure of  
> the perceptual input function.

```
You are now taking the "analyst's viewpoint", which you derided in a

```

message on which I still intend to comment.

> ```
> Evidently we can perceive, at some level,
> 
> ```
> 
> that sort of structure. That may happen at the relationship level,  
> if we  
> consider “relationship” to include the concept of  
> “function.” But this perception is part of a model, not a  
> direct perception of what goes on in the input function.

```
Yes, indeed! That's the analyst's viewpoint all right!

```

> > ```
> > MMT: If for some
> > 
> > ```
> > 
> > reason you no  
> > longer think that perceptions are the only variables that can be  
> > controlled, and that environmental variables (“something the  
> > external observer can see”) also can be controlled, I’m afraid I  
> > cannot agree. Your PCT has fallen into the supernatural void.
> 
> ```
> The external observer can, given a set of lower-level variables
> 
> ```
> 
> and some  
> methodology, deduce what function of them is being controlled.  
> Given a  
> target position T and a cursor position C, the observer can see  
> that a  
> controlled variable can be defined as C - T. But the observer  
> _perceives_ only the distance between the target and cursor,  
> a  
> single scalar value, just as the person doing the tracking does.
> 
> ```
> Of course the observer can be wrong, especially if the
> 
> ```
> 
> correlations in  
> the data are low. It’s common for several equally defensible  
> guesses  
> about the nature of a controlled variable to exist.

```
Now we are going into the process whereby an observer becomes an

```

analyst. What you say is true, so far as I can see, but isn’t it a  
bit off the track? The point is, as you seem to concede, that the  
perceptual variable is what is controlled.

> > ```
> > MMT: Failing some
> > 
> > ```
> > 
> > clairvoyant  
> > abilities in the subject that would allow the true state of an  
> > externally  
> > observable variable to serve as the input to the comparator of a  
> > control  
> > system, the perceptual signal is the only way that the control  
> > unit can  
> > get information about the state of ANYTHING the external  
> > observer can  
> > see.
> 
> ```
> BP: Yes, and this is also the only way the observer can get
> 
> ```
> 
> information  
> about whatever is being controlled.

```
Yes, indeed. But why concern yourself with a hypothetical observer?

```

Is it a phiosophical question like “If a tree falls in a forest with  
nobody near, does it make any sound?” If a person acts without an  
observer to see, is the person controlling?

> > ```
> > MMT: For my part, I
> > 
> > ```
> > 
> > will  
> > continue to write about PCT as though only perceptions can be  
> > controlled.
> 
> ```
> I sympathize with your frustration, even if it's mostly your own
> 
> ```
> 
> doing.  
> Reorganizing is uncomfortable. Most of your misinterpretations, of  
> which  
> there have been many lately, have brought out hidden assumptions  
> that  
> have now become the subject of discussion.

```
Would you mind describing a few of these "many" misinterpretations?

Is it a misinterpretation that Rick has said (whether he believes it

```

or not) and that you have agreed (whether you believe it or not)  
that for control to exist, some variable observable by an external  
observer must be influenced by the control system’s output, and if  
no such variable is influenced by the output, control does not  
exist?

```
Here's what Rick said [From Rick Marken (2010.08.06.1630)] that

```

leads me to the foregoing misinterpretation: " A controlled variable  
(in PCT) has an environmental correlate, the controlled quantity,  
which can be detected by an experimenter. There is no environmental  
correlate of p[pS]-A] (because A exists only in the subject’s  
imagination) so it’s really not a controlled variable."

```
My misinterpretation unfortunately led me to believe that Rick was

```

claiming that if the inputs to a perceptual function did not come  
from some particular class of sources (variables whose variation was  
perceptible to an outside observer), the corresponding perception  
was not controlled, regardless of the fact that it was brought to a  
reference value by variation of the output of the control unit. I  
hope you can explain that Rick did not mean that, because I am  
unable to figure out an alternative interpretation.

```
In [From Bill Powers (2010.08.07.0210 MDT)] you seemed to use my

```

interpretation of Rick’s words, as you said: “Rick is defining a  
controlled variable as something  
the external observer can see. If you would read his words that  
would be  
obvious.” So am I misinterpreting you as agreeing that a controlled  
variable CAN be something the external observer can see, and that I  
am therefore wrong to say: “For my part, I will  
continue to write about PCT as though only perceptions can be  
controlled.”? In your view, is it the case that PCT does now allow  
for control of external variables (sorry “quantities”)?

```
However, this is only one of my many recent misinterpretations,

```

which you say are all my own doing. I’d like to be made aware of the  
most important of the others.

```
Martin

```

---

<div class="post-metadata">

**Author:** ![BUSINESS\_GAVIN\_HERMA](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/b/2acd7d/32.png) [@BUSINESS\_GAVIN\_HERMA](http://discourse.iapct.org/u/BUSINESS_GAVIN_HERMA)\
**Post date:** [August 19, 2010, 3:54am UTC](http://discourse.iapct.org/t/changing-the-foundations-of-pct-was-testing-for-control-in-experiments/2452/4 "2010-08-19T03:54:49Z")

</div>

(Gavin Ritz 2010.08.19.15.41NZT)

Hi there Martin

I understand your concern  
about the perceptual signals. This baffled me to until I realised the actual relationship  
with physics, chem and PCT.

perceptual signal is a transduced signal (it has to be), of all incoming energy  
forms and processes. There is no “matter” external variable that isn’t  
transduced. Pressure, sounds, light, temperature, chemical. This relates to all  
the human (and animal) modalities.

And this has repercussions  
for physics and chemistry too. As then all our science is the reflection and  
abstraction of the transduced signals only and that’s TRUE of science. (Perceptual signals).

The big question in my  
mind is what about those signals that cannot be transduced. (Dark matter, dark  
energy-most of the universe) We have no answers for non-matter concepts. No  
wonder we run into immediate problems at the high levels of HPCT.

Regards

Gavin

> **···**
>
> From my understanding the

---

<div class="post-metadata">

**Author:** ![Bill\_Powers1](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/b/8e7dd6/32.png) [@Bill\_Powers1](http://discourse.iapct.org/u/Bill_Powers1)\
**Post date:** [August 19, 2010, 4:28pm UTC](http://discourse.iapct.org/t/changing-the-foundations-of-pct-was-testing-for-control-in-experiments/2452/5 "2010-08-19T16:28:25Z")

</div>

[From Bill Powers (2010.08.19.0748 MDT)]

Martin Taylor 2010.08.18. 14.48 –

> MMT: When you say “A  
> perceptual signal representing the value of a controlled variable”,  
> it sounds very much to me as though you are saying that “the  
> controlled variable” is NOT the value of the perceptual signal.
> 
> For nearly two decades, I have been under a misapprehension, then, that  
> the ONLY controlled variable in PCT was the value of the perceptual  
> signal. My understanding of basic physics, however, leads me to believe  
> that whatever the Powers-Marken version of PCT may from time to time  
> redetermine, the science of control argues that my twenty-year  
> understanding is the correct one. I see no possibility for controlling  
> anything other than the value of a perceptual signal, an implication  
> carried even by the very title: “Behaviour, the control of  
> perception”.

> I had expected that rather than  
> jumping down my throat for (correctly) calling Rick out when he  
> “defined” a controlled variable as something detectable by an  
> external observer, you would remind him that “Perceptual Control  
> Theory” demands that the controlled variable is and is only a  
> perception.

BP: Sorry, but that’s not how I do things. PCT is a THEORY. It’s not a  
fact. Not yet. Every demonstration is a test of the THEORY. The way we  
test a theory, in my view, is to see what experimental results it  
predicts, then do the experiments and compare the results with the  
predictions. This means that the final criterion is observation, not  
theory. Observation is the reality; theory is a guess as to what might  
exist behind the observations. If (competent) observation disagrees with  
theory, it is theory that is wrong.  
Rick defines a controlled variable as something he can observe. Theory  
says it is a perception of his own that he is observing, and that if he  
sees it being controlled (resisting disturbances, etc.), that is because  
the behaving system being observed also is controlling the same  
perception. That is a tremendously important concept, but it still rests  
on a theory that remains to be proven. Not believed in, not taken as a  
matter of settled and agreed faith, not accepted because it is logical  
and makes sense. There will never be a time when we can stop challenging  
the theory; it is and will always remain a picture of an _imagined_  
reality. That is true of all theories with which we try to make sense of  
our experiences. It is true of the theory of electricity, of  
thermodynamics, of God, of quantum mechanics, of basic physics.

> > BP earlier:This accounted for  
> > most of the disagreements between you.
> 
> MMT: Indeed. You aren’t the only person who noted that  
> fact.

BP: Perhaps, but for the moment I seem to be the only one here who  
doesn’t automatically think that his opinion counts for more than Rick’s.  
If I disagree with Rick, I feel called upon to explain why, and from my  
own understanding, not by citing Scripture.

…

> > BP earlier: Your intent appears  
> > to be to bully Rick into agreeing that you are right with no explanations  
> > or discussion and no respect for Rick’s point of view, as well as no  
> > understanding of it.
> 
> MMT: I think I included a complete explanation, didn’t I. You even quoted  
> it: “the output signal influences the perceptual signal, and changes  
> its value to approach its reference value.” What more explanation is  
> needed?
> 
> I understand Rick’s point of view very well (so far as I am able to  
> determine). His point of view is that there is no control if the input to  
> the perceptual function of a control system is not observable by an  
> external observer.

BP: That is nothing like what I think his view is. I think he believes  
that IF the behaving system is organized as PCT claims it is, then the  
control that an external observer can see is explained by imagining a  
system hidden inside the other person, acting to control its own  
perceptions. If the external observer (i.e., any observer but the  
behaving system itself) does not see any control going on, there is no  
way to find out if covert control is happening. It could be happening,  
but it can only be imagined, not observed. It’s not a testable part of  
the theory.

> MMT: I “bullied” Rick  
> by referring to a foundational principle of PCT, one you expound at some  
> length later in the message to which I am responding, that the perceptual  
> function has no way of knowing where its signals come from, and still  
> less does the controlled variable (the perceptual signal) convey any such  
> information.

BP: Yes, that is what I call bullying. A bully knows what the  
foundational principles are, and berates anyone who fails to treat them  
with the awe and respect they deserve, implying that the transgressor is  
ignorant, stupid, or perverse for not seeing the obvious truth. But I say  
that all such “foundational principles” can be challenged by  
anyone at any time, and the only defense is to demonstrate that they  
correctly predict what we observe. To do nothing more than point to the  
principles we are all supposed to believe in is to turn science into  
faith-based guessing.

So now I am bullying you. How does it feel?

> MMT: I’m puzzled by  
> “reference _level_ is defined in terms of the observer’s  
> understanding of the controlled quantity and its intended or preferred  
> value”. What does an observer have to do with the reference level of  
> a control unit? To bring an observer into the definition of  
> “reference level” seems very odd.

BP: “Reference level” is the state of the controlled quantity  
that an observer sees as the goal-state being produced and maintained by  
the behavior. “Reference signal” or “reference  
perception” is part of the theory that explains how an observable  
reference level can exist. In the standard PCT diagram, everything below  
the system-environment boundary line is observable by an external  
observer, though the observer must learn how to combine the physical  
variables in the right way to see with some degree of approximation what  
the controlled quantity is.

> MMT: Hitherto, it has just been  
> a value input to the comparator of a control unit, for comparison with  
> the perceptual signal value. No observer needed or implied. Where did  
> this “observer” in the definition spring  
> from?

BP: You and I are the observers. There may be others. The reference  
_signal_ is an entity of the model referring to an invisible,  
imagined variable which we (meaning you and I) infer from finding the  
reference _level_ of the observable controlled _quantity._  
Observing means perception by a system other than the behaving  
system.

> > BP: So in your opinion, it  
> > violates the foundations of PCT.
> 
> MMT: But it doesn’t violate yours? My understanding that this was the  
> case was the reason I changed the subject line to “Changing the  
> foundations of PCT”.

BP: It doesn’t violate mine at all. I agree that an observed controlled  
variable or quantity probably corresponds to a controlled perception  
inside the behaving system (as well as the observing system, if the  
observer guesses right). I agree that a hypothetical controlled quantity  
inside the behaving system can’t be observed from outside, and thus has  
to be imagined by the observer, which means by you and me.

> MMT: Here’s how I bullied Rick  
> in my continuation of “The controlled quantity is ALWAYS a  
> perception, never an environmental variable. That’s PCT 101, first  
> lesson.” The bullying commences thus:“as Bill corrected me in  
> the partner thread, the output of a control unit knows or cares nothing  
> about where its signal goes. Control exists if the output influences the  
> perception. If the influence of the output reaches the perception through  
> the external environment, so be it. If the influence of the output  
> reaches the perception by way of an internal connection we call  
> imagination, so be it. If the influence of the output reaches the  
> perception both through the external environment and through internal  
> pathways, so be it. Any which way, it’s irrelevant to the control  
> unit.”

BP: You’re speaking as if the theory is gospel truth and brooks no  
contradiction. You’re hammering on the pulpit and preaching down to the  
sinners, showing them the right way to think without any attempt at  
persuasion, without questions like “what do you mean by …?”  
You’re threatening to expel them into the outer darkness, away from those  
who have been saved and think correctly. I do call this bullying, though  
the threat of force is not directly from you – it’s from the community  
of Right Thinkers.

> MMT: But of course, that  
> bullying comment contradicts Rick’s definition, for which I suppose I  
> should apologise.

BP: Contradiction is always permissible, even with respect to fundamental  
principles. But by my rules, you have to explain why you are rejecting  
what is said, not just substitute something you believe is better and say  
it at more length, and louder.

> > BP earlier: In fact, your way of  
> > saying it violates my attempt to put some order in the way we talk about  
> > controlled variables by reserving the term “quantity” to mean  
> > observable physical variables outside the system, and “signal”  
> > to mean variables inside the system. “Variable” applies equally  
> > inside and outside, so can’t be used to distinguish the observer’s view  
> > from the organism’s view.
> 
> MMT: ??? Who cares about the observer’s view unless the observer is part  
> of the object of study?

BP: How can there be an object of study without an observer?

> MMT: When you are dealing with  
> the observer-observed system, of course the observer matters. But why  
> bring the observer into a discussion of which variable is controlled in a  
> single control system?

BP: Because there is still an observer, or several: the parties to the  
discussion.

> MMT: Is using the technically  
> correct “variable” in place of your idiosyncratically chosen  
> “quantity” justification for your disputing that “defining  
> a controlled variable as something the observer can see is a violation of  
> the foundations of PCT?” Is that definition a violation of the  
> foundations of PCT, or is it not?

BP: It’s a claim, not a definition. My attempt was to separate an entity  
in a brain model (signal) from an entity in the physics model (quantity),  
to distinguish the general term (variable) according to context.

> > BP earlier: Evidently we can  
> > perceive, at some level, that sort of structure. That may happen at the  
> > relationship level, if we consider “relationship” to include  
> > the concept of “function.” But this perception is part of a  
> > model, not a direct perception of what goes on in the input function.
> 
> MMT: Yes, indeed! That’s the analyst’s viewpoint all  
> right!

BP: Then the analyst’s point of view is the one from which we all must  
begin. The analyst is constructing a model that might explain the rest of  
experienced reality. The model, clearly, is part of experienced reality,  
but it is not the rest of reality that is to be explained, and it comes  
after, not before, that which requires explanation.

…

> MMT: Now we are going into the  
> process whereby an observer becomes an analyst. What you say is true, so  
> far as I can see, but isn’t it a bit off the track? The point is, as you  
> seem to concede, that the perceptual variable is what is  
> controlled.

BP: Sorry for sounding like President Clinton, but the answer depends on  
what you mean by “is.” Do you mean the perceptual variable  
actually is, in objective reality and independent of human  
interpretation, what is controlled, or do you mean that according  
to PCT (in which T stands for Theory) the perceptual variable is proposed  
to be what is controlled?

…

> > BP earlier: Yes, and this is  
> > also the only way the observer can get information about whatever is  
> > being controlled.
> 
> MMT: Yes, indeed. But why concern yourself with a hypothetical observer?  
> Is it a philosophical question like “If a tree falls in a forest  
> with nobody near, does it make any sound?” If a person acts without  
> an observer to see, is the person controlling?

BP: I am a quite real observer and I trust you see yourself that way,  
too. I’m talking about how you and I get information, not some vague  
hypothetical observer. Admittedly, I am generalizing from myself to  
others, but so far all the evidence supports the idea that they, too, can  
observe.

> MMT: For my part, I will  
> continue to write about PCT as though only perceptions can be  
> controlled.

BP: Fine, but I hope you include disclaimers such as  
“theoretically,” or “according to PCT” or “as  
far as we know today,” or sometimes “as has been demonstrated  
in many cases,” and so on.

> MMT: Would you mind describing a  
> few of these “many” misinterpretations?

BP: I have been doing so here.

> MMT: Is it a misinterpretation  
> that Rick has said (whether he believes it or not) and that you have  
> agreed (whether you believe it or not) that for control to exist, some  
> variable observable by an external observer must be influenced by the  
> control system’s output, and if no such variable is influenced by the  
> output, control does not exist?

BP: Yes, that is a misinterpretation.

> MMT: Here’s what Rick said [From  
> Rick Marken (2010.08.06.1630)] that leads me to the foregoing  
> misinterpretation: " A controlled variable (in PCT) has an  
> environmental correlate, the controlled quantity, which can be detected  
> by an experimenter. There is no environmental correlate of p[pS]-A]  
> (because A exists only in the subject’s imagination) so it’s really not a  
> controlled variable."

“A controlled variable has an environmental correlate” means,  
to me, that according to PCT there is an internal perception being  
controlled, and that there is some corresponding external state of a set  
of variables that an experimenter can detect. I might point out some  
exceptions to that (the taste of lemonade) to see if he would modify that  
statement, but I wouldn’t just say he’s wrong. From the second quoted  
sentence I would say he’s defining a controlled variable as something  
observable, so I would try to see if he would agree that in PCT an  
imagined perception is a variable, too, though an observer can’t know  
about it. He might modify that statement a bit, too, or adopt the usage  
that an observable controlled variable should be called a controlled  
quantity, thus making it clear that the observer (Rick, you, I) is seeing  
it as something under control.

> MMT: My misinterpretation  
> unfortunately led me to believe that Rick was claiming that if the inputs  
> to a perceptual function did not come from some particular class of  
> sources (variables whose variation was perceptible to an outside  
> observer), the corresponding perception was not controlled, regardless of  
> the fact that it was brought to a reference value by variation of the  
> output of the control unit. I hope you can explain that Rick did not mean  
> that, because I am unable to figure out an alternative  
> interpretation.

BP: I have given it to you. To understand Rick’s point of view, and I  
assume he will correct this if it is wrong, you have to begin with what  
the observer can observe: a controlled quantity in the common (-seeming)  
environment. If the observer succeeds in seeing it as under control, the  
observer will perceive something that is maintained near some some state  
we call its reference level or reference condition. According to PCT, the  
controlling system will also be perceiving that variable or something  
close to it, with a reference signal specifying the state of the  
perception that corresponds to what we observe as its reference level or  
condition. The theory provides for imagined perceptions and their  
control, but those are observable only from within the controlling  
system. The observer can imagine them, but there is no way to test the  
correctness of that hypothesis, and for that reason, one guess as to what  
is imagined is as good as any other (except when the behaving system is  
doing the guessing).

> MMT: In [From Bill Powers  
> (2010.08.07.0210 MDT)] you seemed to use my interpretation of Rick’s  
> words, as you said: “Rick is defining a controlled variable as  
> something the external observer can see. If you would read his words that  
> would be obvious.” So am I misinterpreting you as agreeing that a  
> controlled variable CAN be something the external observer can see, and  
> that I am therefore wrong to say: “For my part, I will continue to  
> write about PCT as though only perceptions can be controlled.”? In  
> your view, is it the case that PCT does now allow for control of external  
> variables (sorry “quantities”)?
> 
> However, this is only one of my many recent misinterpretations, which you  
> say are all my own doing. I’d like to be made aware of the most important  
> of the others.

BP: The basic misinterpretation that I see is taking theory as the  
reality and the observer as an unnecessary part of the theory. My  
position is that the observed world is the reality, and that theories are  
attempts by observers to make sense of it. They go astray when they start  
thinking that the figments of their imaginations, the models they are  
constructing as explanations, are more real than the experiences they are  
trying to explain.

In your comments about Rick, you appear to be assuming that PCT is  
primary and real observations are secondary, so that if there is a clash,  
PCT wins. I think Rick takes the oppositive view. I should leave the rest  
of this to Rick.

Best,

Bill P.

---

<div class="post-metadata">

**Author:** ![rsmarken](http://discourse.iapct.org/user_avatar/discourse.iapct.org/rsmarken/32/3553_2.png) [@rsmarken](http://discourse.iapct.org/u/rsmarken)\
**Post date:** [August 19, 2010, 5:18pm UTC](http://discourse.iapct.org/t/changing-the-foundations-of-pct-was-testing-for-control-in-experiments/2452/6 "2010-08-19T17:18:39Z")

</div>

[From Rick Marken (2010.08.19.1020)]

> Bill Powers (2010.08.19.0748 MDT) to Martin Taylor 2010.08.18. 14.48 --

> In your comments about Rick, you appear to be assuming that PCT is primary  
> and real observations are secondary, so that if there is a clash, PCT wins.  
> I think Rick takes the oppositive view. I should leave the rest of this to  
> Rick.

Thanks, Bill. You got it all right. My reply to Martin would have been  
much shorter than yours. I was just going to say that, from my  
perspective, PCT is a theoretical account of the \_fact\_ of control.  
So, as you say, I see observation (of control phenomena) as primary;  
theory as secondary, in the sense that the theory must constantly be  
tested against observation, and revised if the observations demand it.

You are right that I was unclear about the "environmental correlate"  
of a controlled variable. I guess it would be more correct to say  
that, per PCT, there is a \_function\_ of environmental variables that  
correlates with the controlled perceptual variable. So the perception  
of the taste of lemonade is some function of what physics and  
chemistry tells us are the physical variables (chemicals) that make up  
lemonade.

But I was actually reluctant to reply to Martin at all because I find  
these verbal disagreements pretty uninteresting. I would rather talk  
about actual working models and how they relate to relevant data. I've  
now got a working version of my model of the behavior in a yes/not  
detection task; it's not very complicated and still needs work but it  
does produce results like those seen in yes/no detection tasks. Once  
I've produced a version of Martin's model I will distribute the code  
for both and we can go from there.

I assume that both Martin's and my model will behave in exactly the  
same way in the yes/no detection task. So once we get acceptable  
versions of both models we can start figuring how to develop  
experiments that distinguish the models and then start collecting data  
(rather than sitting around arguing about what the theory "really"  
says). I think a discussion of how well the theory accounts for the  
facts will be much more satisfying than a discussion how well I  
account for the theory;-)

Best

Rick

> **···**
>
> --  
> Richard S. Marken PhD  
> rsmarken@gmail.com  
> www.mindreadings.com

---

<div class="post-metadata">

**Author:** ![Bill\_Powers1](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/b/8e7dd6/32.png) [@Bill\_Powers1](http://discourse.iapct.org/u/Bill_Powers1)\
**Post date:** [August 19, 2010, 7:25pm UTC](http://discourse.iapct.org/t/changing-the-foundations-of-pct-was-testing-for-control-in-experiments/2452/7 "2010-08-19T19:25:29Z")

</div>

[From Bill Powers (2010.08.19.1325 MDT)]

> I would rather talk  
> about actual working models and how they relate to relevant data. I've  
> now got a working version of my model of the behavior in a yes/not  
> detection task; it's not very complicated and still needs work but it  
> does produce results like those seen in yes/no detection tasks. Once  
> I've produced a version of Martin's model I will distribute the code  
> for both and we can go from there.

Great. That's what we need to do. I'll be interested to see how you handle the digital control system: As I see it, when the error goes to zero, the output should go to the zero-output state, which should make the error nonzero on the next iteration, which should make the output go to maximum, and so on forever.

Best,

Bill P.

---

<div class="post-metadata">

**Author:** ![MartinT](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/m/ee7513/32.png) [@MartinT](http://discourse.iapct.org/u/MartinT)\
**Post date:** [August 22, 2010, 4:58am UTC](http://discourse.iapct.org/t/changing-the-foundations-of-pct-was-testing-for-control-in-experiments/2452/8 "2010-08-22T04:58:44Z")

</div>

[Martin Taylor 2010.08.22.00.58

OK. I give up.

Martin

---

<div class="post-metadata">

**Author:** ![MartinT](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/m/ee7513/32.png) [@MartinT](http://discourse.iapct.org/u/MartinT)\
**Post date:** [August 22, 2010, 10:27pm UTC](http://discourse.iapct.org/t/changing-the-foundations-of-pct-was-testing-for-control-in-experiments/2452/9 "2010-08-22T22:27:32Z")

</div>

[Martin Taylor 2010.08.22.17.35

```
 [Martin Taylor 2010.08.22.00.58

```

> ```
> OK. I give up.
> 
> Martin
> 
> ```

```
I suppose I owe you an explanation as to why I give up.

```

Fundamentally, it is because I am unpracticed in conducting a  
discussion in which the premises can be arbitrarily changed by one  
participant without explanation or notice that they are being  
changed.

```
Examples:

1. We are discussing implications of a theory known as Perceptual

```

Control Theory. Hitherto, this theory has been that the only  
variable that can be controlled is the output of a function known as  
the perceptual input function, that this value is compared to a  
reference value supplied from outside, and that the difference  
affects an output function whose output influences an input to the  
perceptual input function to complete a feedback loop. Nothing is  
stated about how the output influences the input, since there are  
many possibilities. What has always mattered has been that ONLY the  
perceptual signal value, labelled “P” in the figure, can be  
controlled, and that the control occurs because in some undefined  
way O influences S.

```
![ctrl_unit2.logo.png|120x80](upload://rQ6aDpDdn73lmbT4kFPFmyaSNu4.png)

In the course of the discussion, I have been told, first by Rick and

```

then by Bill, that there are now new conditions imposed on what  
happens between “O” and “S” in the figure, namely that “O” must  
influence something observable by an independent entity. When I  
question where this new condition comes from, and that my  
understanding of PCT has been only that O influences S, I am told I  
am “bullying”.

```
2. Over the many years in which I have been aware of PCT, the effect

```

of perceptual control by and within an organism has been considered  
(in the theory) independent of whether anyone else observed the  
control happening. Although it seems that the potential existence of  
an observer does matter in quantum physics (a finding questioned by  
Bill P.), until now it has never been a part of Perceptual Control  
Theory, but suddenly it has become an essential factor. As Rick said  
[From Rick Marken (2010.08.06.1630)]: “There is no environmental  
correlate of p[pS]-A] (because A exists only in the subject’s  
imagination) so it’s really not a controlled variable.” Nothing has  
been argued as to how the potential existence of an observer  
modifies the process of control, but it is suddenly asserted that  
the potential presence of an observer is essential for a variable to  
be controlled, and Bill P. has supported this assertion without  
providing any supporting reasons for such a fundamental change in  
PCT.

```
3. In answer to Rick's claim that a perceptual value cannot be

```

controlled if it is the output of a function that happens to have  
one of its inputs an imagined variable value, I said [Martin Taylor  
2010.08.11.14.06]  
: “Of course it is a controlled variable. Just as with any control  
system, the output signal influences the perceptual signal, and  
changes its value to approach its reference value.” If this is not  
the definition of a controlled variable, then the definition has  
been changed under my nose, without warning. Nevertheless, my  
stating it thus has been called “bullying” by Bill, in not one but  
two messages. I have never understood that an appeal to what used to  
be a basic premise of a theory could be called “bullying” when  
discussing an implication of the theory, but I suppose that since  
the argument is incontrovertible Rick would have no way to counter  
it, so an ad hominem argument had to be produced in place of a  
scientific one that doesn’'t exist.

```
4. I said: [MT]

```

> ```
> I had expected that
> 
> ```
> 
> rather than  
> jumping down my throat for (correctly) calling Rick out when he  
> “defined” a controlled variable as something detectable by an  
> external observer, you would remind him that “Perceptual Control  
> Theory” demands that the controlled variable is and is only a  
> perception.

```
BP: Sorry, but that's not how I do things.

[MT now] It is exactly how you do things. I said ""Perceptual

```

Control Theory" demands …", and you said, as though contradicting  
me, “PCT is a THEORY. It’s not a  
fact.” It is, and you don’t contradict me in saying so. The fact  
that the alternative is to accept clairvoyance, or to deny the  
accepted laws of thermodynamics, is really irrelevant (other than to  
say that PERCEPTUAL control is a theory with very strong support in  
normal physics).

```
When we are dealing with the implications of a theory, it is normal

```

to take the basic assumptions of the theory as tentative facts and  
see where they lead. It is completely improper to disclaim the basic  
tenets of the theory on the grounds that it IS a theory, in order to  
avoid addressing the issue. [Parenthetically: how many hundred times  
have you NOT taken Rick to task when he has said that “conventional  
psychologists” fail to note the FACT of control?]

```
The reason I give up is that I have no skill in changing the basic

```

premises of an argument to suit a conclusion I would like to reach;  
I am accustomed to working on the basis that when the basic tenets  
of a theory have been defined, and when a derivation from them has  
been argued and accepted, that agreement persists until other  
evidence (e.g a failure of a model to perform as expected) is  
adduced or new arguments developed to change the conclusion.  
Likewise, I have no skill in conducting arguments in which the other  
party changes the premises at will to suit their desired conclusion.  
Since it is clear your objective is to win, rather than to reach  
supportable scientific results, I really have no choice but to  
withdraw.

```
Martin

```

---

<div class="post-metadata">

**Author:** ![Bill\_Powers1](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/b/8e7dd6/32.png) [@Bill\_Powers1](http://discourse.iapct.org/u/Bill_Powers1)\
**Post date:** [August 22, 2010, 11:54pm UTC](http://discourse.iapct.org/t/changing-the-foundations-of-pct-was-testing-for-control-in-experiments/2452/10 "2010-08-22T23:54:26Z")

</div>

[From Bill Powers (2010.08.22.1640 MDT)]

Martin Taylor 2010.08.22.17.35 –

I appreciate the expansion on your post.

> MMT: 1. We are discussing  
> implications of a theory known as Perceptual Control Theory. Hitherto,  
> this theory has been that the only variable that can be controlled is the  
> output of a function known as the perceptual input function, that this  
> value is compared to a reference value supplied from outside, and that  
> the difference affects an output function whose output influences an  
> input to the perceptual input function to complete a feedback loop.  
> Nothing is stated about how the output influences the input, since there  
> are many possibilities. What has always mattered has been that ONLY the  
> perceptual signal value, labelled “P” in the figure, can be  
> controlled, and that the control occurs because in some undefined way O  
> influences S.
> 
> ![1f7324c.jpg](http://discourse.iapct.org/uploads/default/original/1X/e740063cd59e945958dee712ee45f5c0117143ec.jpeg)
> 
> In the course of the discussion, I have been told, first by Rick and then  
> by Bill, that there are now new conditions imposed on what happens  
> between “O” and “S” in the figure, namely that  
> “O” must influence something observable by an independent  
> entity. When I question where this new condition comes from, and that my  
> understanding of PCT has been only that O influences S, I am told I am  
> “bullying”.

BP: The theory leads to exactly what you are saying. If PCT is correct,  
all that can be controlled are perceptions. Therefore when we see other  
people controlling aspects of what we see as their environments, the  
theory tells us that they are controlling perceptual signals representing  
those aspects.

However, we do not observe the perceptions that other people are  
controlling. they are hypothetical. Neither do we observe that we  
ourselves are controlling perceptions; it seems to us that we are  
controlling real aspects of a real external world. PCT is a step toward  
explaining how this can happen.

> MMT: 2. Over the many years in  
> which I have been aware of PCT, the effect of perceptual control by and  
> within an organism has been considered (in the theory) independent of  
> whether anyone else observed the control happening. Although it seems  
> that the potential existence of an observer does matter in quantum  
> physics (a finding questioned by Bill P.), until now it has never been a  
> part of Perceptual Control Theory, but suddenly it has become an  
> essential factor.

BP: That is a misunderstanding of what Rick and I are saying. We do not  
say that an observer is required by the theory to be present. If the  
theory is correct, a controlling person is controlling a perception by  
acting on a largely unknown external world, although that is not how it  
seems to the person. Unlike quantum mechanics, PCT does not require the  
presence of an observer for the theory to work properly.

> MMT: As Rick said [From Rick  
> Marken (2010.08.06.1630)]: “There is no environmental  
> correlate of p[pS]-A] (because A exists only in the subject’s  
> imagination) so it’s really not a controlled variable.”

BP: He should have said “It’s really not an observable controlled  
variable.” I asked him about that, and that is what he said he  
meant. If you had checked to find out what he meant in this critical  
regard, you could have discovered this just as easily.

> MMT: Nothing has been argued as  
> to how the potential existence of an observer modifies the process of  
> control, but it is suddenly asserted that the potential presence of an  
> observer is essential for a variable to be controlled, and Bill P. has  
> supported this assertion without providing any supporting reasons for  
> such a fundamental change in PCT.

BP: You are running with the ball in a direction completely off the  
playing field. In PCT, the observer has no effect on the process of  
control just by observing it (the observer can do things to affect it,  
but nothing mysterious).

> MMT; 3. In answer to Rick’s  
> claim that a perceptual value cannot be controlled if it is the output of  
> a function that happens to have one of its inputs an imagined variable  
> value,

BP: That is not what Rick intended to say, as a simple question to him  
about his meaning quickly verified. If you would simply change your  
statement a little, it would match what Rick means: “A perceptual  
value cannot be known by an external observer to be controlled if the  
perceptual value depends on unobservable variables inside the  
subject.” We are all external observers when it comes to explaining  
the behavior of a subject other than ourselves in an experiment. We can’t  
use the theory as if we could directly verify it by observing the  
entities in it. We have to use indirect experimental means.

> MMT: le. Just as with any  
> control system, the output signal influences the perceptual signal, and  
> changes its value to approach its reference value." If this is not  
> the definition of a controlled variable, then the definition has been  
> changed under my nose, without warning. Nevertheless, my stating it thus  
> has been called “bullying” by Bill, in not one but two  
> messages.

BP: No, I called your critical and superior attitude, the unquestioning  
assumption that you are right and Rick is wrong, bullying, not what you  
said.

> MMT: I have never understood  
> that an appeal to what used to be a basic premise of a theory could be  
> called “bullying” when discussing an implication of the theory,  
> but I suppose that since the argument is incontrovertible Rick would have  
> no way to counter it, so an ad hominem argument had to be produced in  
> place of a scientific one that doesn’'t exist.

BP: An appeal to a basic premise of a theory is only that, an appeal to a  
statement on which a theory rests. It’s not a proof that they theory or  
the premise, is right. But I wouldn’t call that bullying – it simply  
misses the point, but we all do that on occasion. What is bullying is to  
speak as if from superior logic and understanding and as if the target of  
your criticisms is too stupid to realise his error. Perhaps you simply  
don’t grasp how your words are heard at the receiving end.

> 1. I said: [MT]
> 
> > I had expected that rather than  
> > jumping down my throat for (correctly) calling Rick out when he  
> > “defined” a controlled variable as something detectable by an  
> > external observer, you would remind him that “Perceptual Control  
> > Theory” demands that the controlled variable is and is only a  
> > perception.
> 
> BP: Sorry, but that’s not how I do things.
> 
> [MT now] It is exactly how you do things. I said ““Perceptual  
> Control Theory” demands …”, and you said, as though  
> contradicting me, “PCT is a THEORY. It’s not a fact.” It is,  
> and you don’t contradict me in saying so.

BP: That’s right. But citing PCT as if it’s a fact, which is what you  
have been doing, is not how I do things or how I think they should be  
done. You are complaining about non-adherence to the claims of a theory,  
whereas Rick and I have been saying that the theory can be used only to  
the extent that the data in a given circumstance support it. If the data  
are not there to support the idea that a variable is being controlled in  
the imagination mode, then claiming that you know it is being controlled  
that way is a mistake.

> MMT: The fact that the  
> alternative is to accept clairvoyance, or to deny the accepted laws of  
> thermodynamics, is really irrelevant (other than to say that PERCEPTUAL  
> control is a theory with very strong support in normal  
> physics).

BP: The alternative is not to accept clairvoyance, it’s simply to say  
“I don’t know.” There could be many alternative explanations  
that might prove to be correct and the fact that you don’t know what they  
are is irrelevant. Anyway, there is nothing in what Rick or I have said  
that demands clairvoyance; on the contrary, your insistence that the  
subject is imagining one variable in a relationship between two variables  
can be known only by clairvoyance to everyone else but that subject. That  
is the basic objection Rick and I both have to your model: there is no  
way to test it.

> MMT: When we are dealing with  
> the implications of a theory, it is normal to take the basic assumptions  
> of the theory as tentative facts and see where they lead.

BP: Tentative, yes. That means disclaimers, it means saying IF the  
subject was imagining this or that, we should expect certain  
consequences. It also means retaining a sufficient degree of awareness  
thsat the assumption may not be a fact, and limiting the deductions from  
it accordingly.

> MMT: It is completely improper  
> to disclaim the basic tenets of the theory on the grounds that it IS a  
> theory, in order to avoid addressing the issue.

BP: What about disclaiming them on the grounds that they are untestable  
and that any deductions from them might have to be retracted in  
toto?

> MMT:[Parenthetically: how many  
> hundred times have you NOT taken Rick to task when he has said that  
> “conventional psychologists” fail to note the FACT of  
> control?]

BP:I would never criticise him for saying such things. I have critized  
him for being belligerant and using terms like “a\*\*hole” to  
embellish his remarks.

> MMT: The reason I give up is  
> that I have no skill in changing the basic premises of an argument to  
> suit a conclusion I would like to reach; I am accustomed to working on  
> the basis that when the basic tenets of a theory have been defined, and  
> when a derivation from them has been argued and accepted, that agreement  
> persists until other evidence (e.g a failure of a model to perform as  
> expected) is adduced or new arguments developed to change the  
> conclusion.

BP: I think of theories as ideas that have to be tested and retested  
throughout their entire lives. There is never a time when you can just  
take it for granted that they are correct, which seems to be what you are  
recommending. Every time PCT is used to explain an experimental finding,  
I want to see it challenged and to find evidence in the experimental  
results to meet the challenge. Imagining that someone else is imagining  
is not a way to do that. It’s not an unreasonable guess, but it’s not the  
only possible guess and no important conclusions should be drawn from it  
until we find a way to test it.

As to your accusation of changing premises to suit a conclusion I would  
like to reach, that is a figment of your imagination. You have shown no  
evidence, even after this long interchange, of understanding my point,  
which is the same as it has always been. I have never changed the  
structure of the PCT model in the ways you describe. But some aspects of  
the model, such as the imagination mode, are at present unobservable in  
nature, and we can make up any details we wish, which greatly weakens any  
conclusions we draw, making them unfalsifiable.

Control is a fact that we can observe, given the operational definitions  
we have worked out over the years. It can be explained by PCT, but not  
every aspect of PCT can be verified experimentally, which is a weakness  
of the theory we have to keep trying to remedy. In any case, theory does  
not come first: observation does. The theory exists only to explain  
observations and can never be more believable than the observations.

An external observer – which includes you – can’t test to see whether  
imagination is involved in an entirely hidden control process. That is  
the point you have misinterpreted as a claim that control can’t happen  
without an external observer. If you have misinterpreted that, then  
nothing you have said that is based on the mistaken interpretation has  
any relevance to the discussion.

Imagination is of importance subjectively – it’s a phenomenon that  
individuals can experience, and requires explanation, but the explanation  
doesn’t apply to our observations of other people simply because we can’t  
observed imagination happening or even infer it from observable  
evidence.

Bill P.

---

<div class="post-metadata">

**Author:** ![BUSINESS\_GAVIN\_HERMA](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/b/2acd7d/32.png) [@BUSINESS\_GAVIN\_HERMA](http://discourse.iapct.org/u/BUSINESS_GAVIN_HERMA)\
**Post date:** [August 23, 2010, 12:06am UTC](http://discourse.iapct.org/t/changing-the-foundations-of-pct-was-testing-for-control-in-experiments/2452/11 "2010-08-23T00:06:11Z")

</div>

( Gavin  
Ritz 2010.08.23.11.53NZT)

[Martin Taylor  
2010.08.22.17.35  
[Martin Taylor 2010.08.22.00.58

Martin

I had expected that  
rather than jumping down my throat for (correctly) calling Rick  
out when he “defined” a controlled variable as something detectable by  
an external observer, you would remind him that “Perceptual Control  
Theory” demands that the controlled variable is and is only a perception.

Welcome to the world of Illusion’s

I’m not sure that you read my other  
threads, just about everything is a controlled variable. Law, science, social  
conditions, knowledge, etc

You have now reached the inevitable conclusion  
that something very strange is going on. Only PCT could have brought you here. Bill is battling with it  
because he thinks that PCT is a science when science falls into the same nest  
of illusions.

You have reached the highest human  
conundrum. Don’t be hard on Bill his theory is actually very robust until we climb the levels. PCT  
runs into the same problem that all social theories do, read Holism, or  
Stafford Beers’ Viable System, or Peter Checkland’s Systems Thinking. You require requisitely to mirror  
yourself (your model). Is this not a requirement over and above science?

Personal Honesty is a very good reward. I’m  
afraid it’s outside the realms of PCT.

Kind regards

Gavin

> **···**
>

---

<div class="post-metadata">

**Author:** ![BUSINESS\_GAVIN\_HERMA](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/b/2acd7d/32.png) [@BUSINESS\_GAVIN\_HERMA](http://discourse.iapct.org/u/BUSINESS_GAVIN_HERMA)\
**Post date:** [August 23, 2010, 2:34am UTC](http://discourse.iapct.org/t/changing-the-foundations-of-pct-was-testing-for-control-in-experiments/2452/12 "2010-08-23T02:34:33Z")

</div>

(Gavin Ritz 2010.08.23.14.27NZT)

[From Bill Powers  
(2010.08.22.1640 MDT)]

Martin Taylor 2010.08.22.17.35 –

This is one of the most appropriate responses  
I have seen on this list. Honest and too the point.

![image00126.jpg](http://discourse.iapct.org/uploads/default/original/1X/e580a4b61d9bfe5d8462b36ce6161cd56b2de91b.jpeg)

> **···**
>
> - 
> 
> I appreciate the expansion on your post.
> 
> MMT: 1. We are discussing  
> implications of a theory known as Perceptual Control Theory. Hitherto, this  
> theory has been that the only variable that can be controlled is the output of  
> a function known as the perceptual input function, that this value is compared  
> to a reference value supplied from outside, and that the difference affects an  
> output function whose output influences an input to the perceptual input  
> function to complete a feedback loop. Nothing is stated about how the output  
> influences the input, since there are many possibilities. What has always  
> mattered has been that ONLY the perceptual signal value, labelled “P”  
> in the figure, can be controlled, and that the control occurs because in some  
> undefined way O influences S.
> 
> ![[]]()
> 
> In the course of the discussion, I have been told, first by Rick and then by  
> Bill, that there are now new conditions imposed on what happens between  
> “O” and “S” in the figure, namely that “O” must  
> influence something observable by an independent entity. When I question where  
> this new condition comes from, and that my understanding of PCT has been only  
> that O influences S, I am told I am “bullying”.
> 
> BP: The theory leads to exactly what you are saying. If PCT is correct, all  
> that can be controlled are perceptions. Therefore when we see other people  
> controlling aspects of what we see as their environments, the theory tells us  
> that they are controlling perceptual signals representing those aspects.
> 
> However, we do not observe the perceptions that other people are controlling.  
> they are hypothetical. Neither do we observe that we ourselves are controlling  
> perceptions; it seems to us that we are controlling real aspects of a real  
> external world. PCT is a step toward explaining how this can happen.
> 
> MMT: 2. Over the many  
> years in which I have been aware of PCT, the effect of perceptual control by  
> and within an organism has been considered (in the theory) independent of  
> whether anyone else observed the control happening. Although it seems that the  
> potential existence of an observer does matter in quantum physics (a finding  
> questioned by Bill P.), until now it has never been a part of Perceptual  
> Control Theory, but suddenly it has become an essential factor.
> 
> BP: That is a misunderstanding of what Rick and I are saying. We do not say  
> that an observer is required by the theory to be present. If the theory is  
> correct, a controlling person is controlling a perception by acting on a  
> largely unknown external world, although that is not how it seems to the  
> person. Unlike quantum mechanics, PCT does not require the presence of an  
> observer for the theory to work properly.
> 
> MMT: As Rick said [From  
> Rick Marken (2010.08.06.1630)]: “There is no environmental correlate  
> of p[pS]-A] (because A exists only in the subject’s imagination) so it’s really  
> not a controlled variable.”
> 
> BP: He should have said “It’s really not an observable controlled  
> variable.” I asked him about that, and that is what he said he meant. If  
> you had checked to find out what he meant in this critical regard, you could  
> have discovered this just as easily.
> 
> MMT: Nothing has been  
> argued as to how the potential existence of an observer modifies the process of  
> control, but it is suddenly asserted that the potential presence of an observer  
> is essential for a variable to be controlled, and Bill P. has supported this  
> assertion without providing any supporting reasons for such a fundamental  
> change in PCT.
> 
> BP: You are running with the ball in a direction completely off the playing  
> field. In PCT, the observer has no effect on the process of control just by  
> observing it (the observer can do things to affect it, but nothing mysterious).
> 
> MMT; 3. In answer to  
> Rick’s claim that a perceptual value cannot be controlled if it is the output  
> of a function that happens to have one of its inputs an imagined variable  
> value,
> 
> BP: That is not what Rick intended to say, as a simple question to him about  
> his meaning quickly verified. If you would simply change your statement a  
> little, it would match what Rick means: “A perceptual value cannot be  
> known by an external observer to be controlled if the perceptual value depends  
> on unobservable variables inside the subject.” We are all external  
> observers when it comes to explaining the behavior of a subject other than  
> ourselves in an experiment. We can’t use the theory as if we could directly  
> verify it by observing the entities in it. We have to use indirect experimental  
> means.
> 
> MMT: le. Just as with any  
> control system, the output signal influences the perceptual signal, and changes  
> its value to approach its reference value." If this is not the definition  
> of a controlled variable, then the definition has been changed under my nose,  
> without warning. Nevertheless, my stating it thus has been called  
> “bullying” by Bill, in not one but two messages.
> 
> BP: No, I called your critical and superior attitude, the unquestioning  
> assumption that you are right and Rick is wrong, bullying, not what you said.
> 
> MMT: I have never  
> understood that an appeal to what used to be a basic premise of a theory could  
> be called “bullying” when discussing an implication of the theory,  
> but I suppose that since the argument is incontrovertible Rick would have no  
> way to counter it, so an ad hominem argument had to be produced in place of a  
> scientific one that doesn’'t exist.
> 
> BP: An appeal to a basic premise of a theory is only that, an appeal to a  
> statement on which a theory rests. It’s not a proof that they theory or the  
> premise, is right. But I wouldn’t call that bullying – it simply misses the  
> point, but we all do that on occasion. What is bullying is to speak as if from  
> superior logic and understanding and as if the target of your criticisms is too  
> stupid to realise his error. Perhaps you simply don’t grasp how your words are  
> heard at the receiving end.
> 
> 1. I said: [MT]
> 
> I had expected that  
> rather than jumping down my throat for (correctly) calling Rick out when he  
> “defined” a controlled variable as something detectable by an  
> external observer, you would remind him that “Perceptual Control  
> Theory” demands that the controlled variable is and is only a perception.
> 
> BP: Sorry, but that’s not how I do things.
> 
> [MT now] It is exactly how you do things. I said ““Perceptual Control  
> Theory” demands …”, and you said, as though contradicting me,  
> “PCT is a THEORY. It’s not a fact.” It is, and you don’t contradict  
> me in saying so.
> 
> BP: That’s right. But citing PCT as if it’s a fact, which is what you have been  
> doing, is not how I do things or how I think they should be done. You are  
> complaining about non-adherence to the claims of a theory, whereas Rick and I  
> have been saying that the theory can be used only to the extent that the data  
> in a given circumstance support it. If the data are not there to support the  
> idea that a variable is being controlled in the imagination mode, then claiming  
> that you know it is being controlled that way is a mistake.
> 
> MMT: The fact that the  
> alternative is to accept clairvoyance, or to deny the accepted laws of  
> thermodynamics, is really irrelevant (other than to say that PERCEPTUAL control  
> is a theory with very strong support in normal physics).
> 
> BP: The alternative is not to accept clairvoyance, it’s simply to say “I  
> don’t know.” There could be many alternative explanations that might prove  
> to be correct and the fact that you don’t know what they are is irrelevant.  
> Anyway, there is nothing in what Rick or I have said that demands clairvoyance;  
> on the contrary, your insistence that the subject is imagining one variable in  
> a relationship between two variables can be known only by clairvoyance to  
> everyone else but that subject. That is the basic objection Rick and I both  
> have to your model: there is no way to test it.
> 
> MMT: When we are dealing  
> with the implications of a theory, it is normal to take the basic assumptions  
> of the theory as tentative facts and see where they lead.
> 
> BP: Tentative, yes. That means disclaimers, it means saying IF the subject was  
> imagining this or that, we should expect certain consequences. It also means  
> retaining a sufficient degree of awareness thsat the assumption may not be a fact,  
> and limiting the deductions from it accordingly.
> 
> MMT: It is completely  
> improper to disclaim the basic tenets of the theory on the grounds that it IS a  
> theory, in order to avoid addressing the issue.
> 
> BP: What about disclaiming them on the grounds that they are untestable and  
> that any deductions from them might have to be retracted in toto?
> 
> MMT:[Parenthetically: how  
> many hundred times have you NOT taken Rick to task when he has said that  
> “conventional psychologists” fail to note the FACT of control?]
> 
> BP:I would never criticise him for saying such things. I have critized him for  
> being belligerant and using terms like “a\*\*hole” to embellish his  
> remarks.
> 
> MMT: The reason I give up  
> is that I have no skill in changing the basic premises of an argument to suit a  
> conclusion I would like to reach; I am accustomed to working on the basis that  
> when the basic tenets of a theory have been defined, and when a derivation from  
> them has been argued and accepted, that agreement persists until other evidence  
> (e.g a failure of a model to perform as expected) is adduced or new arguments  
> developed to change the conclusion.
> 
> BP: I think of theories as ideas that have to be tested and retested throughout  
> their entire lives. There is never a time when you can just take it for granted  
> that they are correct, which seems to be what you are recommending. Every time  
> PCT is used to explain an experimental finding, I want to see it challenged and  
> to find evidence in the experimental results to meet the challenge. Imagining  
> that someone else is imagining is not a way to do that. It’s not an  
> unreasonable guess, but it’s not the only possible guess and no important  
> conclusions should be drawn from it until we find a way to test it.
> 
> As to your accusation of changing premises to suit a conclusion I would like to  
> reach, that is a figment of your imagination. You have shown no evidence, even  
> after this long interchange, of understanding my point, which is the same as it  
> has always been. I have never changed the structure of the PCT model in the  
> ways you describe. But some aspects of the model, such as the imagination mode,  
> are at present unobservable in nature, and we can make up any details we wish,  
> which greatly weakens any conclusions we draw, making them unfalsifiable.
> 
> Control is a fact that we can observe, given the operational definitions we  
> have worked out over the years. It can be explained by PCT, but not every  
> aspect of PCT can be verified experimentally, which is a weakness of the theory  
> we have to keep trying to remedy. In any case, theory does not come first:  
> observation does. The theory exists only to explain observations and can never  
> be more believable than the observations.
> 
> An external observer – which includes you – can’t test to see whether  
> imagination is involved in an entirely hidden control process. That is the  
> point you have misinterpreted as a claim that control can’t happen without an  
> external observer. If you have misinterpreted that, then nothing you have said  
> that is based on the mistaken interpretation has any relevance to the  
> discussion.
> 
> Imagination is of importance subjectively – it’s a phenomenon that individuals  
> can experience, and requires explanation, but the explanation doesn’t apply to  
> our observations of other people simply because we can’t observed imagination  
> happening or even infer it from observable evidence.
> 
> Bill P.
