# Control of reinforcement

**URL:** <http://discourse.iapct.org/t/control-of-reinforcement/2993>\
**Category:** CSG1995\
**Created:** [July 30, 1995, 5:58pm UTC](http://discourse.iapct.org/t/control-of-reinforcement/2993 "1995-07-30T17:58:22Z")\
**Posts on this page:** 20\
**Page:** 4

<div class="post-metadata">

**Author:** ![R\_Marken](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/r/8797f3/32.png) [@R\_Marken](http://discourse.iapct.org/u/R_Marken)\
**Post date:** [October 19, 1998, 4:04pm UTC](http://discourse.iapct.org/t/control-of-reinforcement/2993/61 "1998-10-19T16:04:41Z")

</div>

[From Rick Marken (981019.0900)]

Bruce Gregory (981019.1130 EDT)--

> Most of the drivel comes from you

That's what I thought.

> It largely consists of pontifications in the absence of data.

Of course. I guess I should try to imitate you, whose discussions  
of PCT are always based on data;-)

> The drivel consists of lengthy "corrections" to imagined mistakes  
> of understanding.

Sorry, I forgot. Everyone is right in post-modernist academia.

> It also consists of gratuitous insults such as, "psychologists  
> (like Bruce Abbott) believe that environmental events control  
> behavior."

Gratuitous insult? Bruce believes that environmental events  
control behavior. So? What's insulting about that? Bruce even  
agreed that he believes that the environment controls behavior;  
he just doesn't mean "control" in the sense that he thought I  
meant "control". Bruce means "control" in the sense of "influence"  
or "cause". In fact, in the beginning of his textbook there is  
a chapter on "Explaining Behavior" in which he has a section  
on "Exploring the Causes of Behavior". So Bruce would certainly  
agree with colleagues who say that the environment "controls"  
behavior because he would know that they mean "cause" when they  
say "control". One of the things PCT teaches is what control \_is\_.  
Once you know what control is you don't write chapters about  
"the causes of behavior"; you write chapters about the nature  
of control.

> If you want to play your hostile, defensive, and paranoid act,  
> please don't.

Sorry. Gotta play. In my non-post-modernist world it's called  
"teaching".

> One would almost think that the two of you imagine that the  
> Bruces and Marc have somehow devoted their lives to undermining  
> PCT, which only three people in the world understand to begin  
> with. I speak for all of us when I say that nothing could be  
> further from the truth.

I don't care what you say; I care what you do. I don't really  
care whether or not people want to "undermine" PCT; that's like  
trying to undermine f = ma. The basics of PCT are right and  
they will \_eventually\_ (many years from now) become "conventional  
psychology". What annoys me is people who say they love the  
control theory model of behavior and who then go out and publish  
a textbook on behavioral research that rejects virtually every  
aspect of that model.

I guess I'd rather have enemies who are honest about how they  
feel about PCT. What I've got now are enemies who are busy telling  
me how much they love PCT while they are beating me over the head  
every time I discuss some basic implication of PCT.

Best

Rick

> **···**
>
> --  
> Richard S. Marken Phone or Fax: 310 474-0313  
> Life Learning Associates e-mail: rmarken@earthlink.net  
> [http://home.earthlink.net/~rmarken](http://home.earthlink.net/~rmarken)

---

<div class="post-metadata">

**Author:** ![Marc\_Abrams5](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/m/f08c70/32.png) [@Marc\_Abrams5](http://discourse.iapct.org/u/Marc_Abrams5)\
**Post date:** [October 19, 1998, 5:08pm UTC](http://discourse.iapct.org/t/control-of-reinforcement/2993/62 "1998-10-19T17:08:17Z")

</div>

> From [Marc Abrams (981019.1227)]

> [From Bill Powers (981019.0549 MDT)]

Thanks for a fine post but it missed my mark a bit. I have  
long given up on "applications" and "applying" PCT. I do not  
presume we will have all the answers to my questions in my  
life time. I am not trying to get \_all\_ the answers. But I  
think we need to move in a direction that \_tries\_ to unviel  
some of the secrets. Spending bunches of time bashing  
conventional Psychology will not get us what we want.  
Thinking up and doing new research might. Actually, I think  
that is the biggest hurdle we have. But I don't think that  
can be supplied by convetional research psychologists. We  
need new kinds of data we need new methods for getting it. I  
don't think doing the test will get us what we want. I think  
we need to better understand the \_dynamics\_ of the  
interactions that take place between environment, self, and  
PCT agents. \_This\_ I think we can do. Through modeling. It  
is no minor task. It is a \_huge\_ undertaking but I think It  
might be one of the ne research methods that helps us  
understand what kinds of information we need and potential  
ways to get it.  
I mentioned my illness only to point out that there are many  
other things i could be doing ( well not many 🙂 ) the few  
hours I was able to keep my eyes open. Part of that time was  
devoted to PCT, so in a way it did help me stay alive and it  
is helping me get well. It gave me some purpose ( expanding  
my own knowledge about behavior ) and enthusiasim and  
continues to do so. This list has some terrific people on  
it. People who care about PCT. Constant talk of conspiracies  
and enemies do not further our knowledge, nor do i think in  
the realm of things important. There is a lot of work that  
needs to be done. There will always be enemies of PCT for  
any number of reasons. Who cares, Lets try to move on with  
our business.

Bill, I've asked this of Rick and I'll ask it of you. \_WHAT  
THE HELL IS THE DAMN NEXT STEP?\_ Is it the elimination or  
conversion of all of our enemies. If it is, I can't help. Is  
it something we can talk about on the net? Do you have a  
specific game plan?

Again, thanks for a very thoughtful post.

Marc.

---

<div class="post-metadata">

**Author:** ![Abbott\_Bruce](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/a/eb9ed0/32.png) [@Abbott\_Bruce](http://discourse.iapct.org/u/Abbott_Bruce)\
**Post date:** [October 19, 1998, 5:50pm UTC](http://discourse.iapct.org/t/control-of-reinforcement/2993/63 "1998-10-19T17:50:42Z")

</div>

[From Bruce Abbott (981018.1250 EST)]

> Bill Powers (981018.2237 MDT) --

> > Bruce Abbott (981018.1730 EST)

> > Bill, you're making a terrible mistake here; I only wish I could figure out  
> > how to set your thinking strait about this. All I can do is try.

> I understand interval schedules, and even variable-interval schedules. I  
> understand that you are talking about VI schedules. I understand all of  
> your painstaking illustrations of elementary relationships. Let's cut to  
> the chase.

I assumed that you do understand them, but thought it would be a good idea  
to review them anyway, for the benefit of any who might be trying to follow  
our argument, should that number be greater than zero. However, your own  
statements revealed that you were applying what you remembered about the  
ratio studies to draw conclusions about what is going on in the interval  
data. Such confusion is understandable -- you are involved in a number of  
projects and it must be difficult to keep track of what did or did not  
happen in observations collected by someone else a year or more ago -- so I  
took the opportunity to straighten out your recollections and to add some  
additional observations from the VI study that contradict the conclusions of  
your ratio-data based reasoning. So yes, let's cut to the chase.

> Suppose the rat happened to stay at the lever during the whole session,  
> pressing at some average rate and getting food pellets at some other  
> average rate (which would happen on any schedule). Pressing would produce a  
> total number P of presses during the session and a total number of  
> reinforcements R. If the duration of the session is T, we have a mean  
> pressing rate of P/T and a mean reinforcement rate of R/T.

> Now, arbitrarily, let us add a period of duration tau at the end of the  
> session, during which no pressing of the lever takes place, and thus no  
> reinforcements ever occur. We can simply remove the rat from the cage and  
> let it do other things beside press the bar, while we let the clock run for  
> an additional tau minutes or hours or days. Now, at the end of the  
> additional time period, we calculate the mean rate of pressing and the mean  
> rate of reinforcement. They are P/(T+tau) and R/(T+tau). Notice that  
> P/(T+tau) is less than P/T if tau and T are positive numbers. Can you  
> conclude from this that a decrease in the session-average rate of  
> reinforcement has caused a decrease in the session-average rate of pressing?

You are not listening to me, Bill. Over most of the range of VIs  
investigated, pauses were not sufficiently long to produce any such effect.  
If they were, this would have been strongly evident in the differences  
between programmed and obtained rate of reward. It was not. The evidence  
contradicts the assumption on which your conclusion is based.

> Obviously it makes no difference if we add the extra time tau at the end of  
> the session of continuous responding, or break it up into many much shorter  
> periods during which the rat is away from the lever. Also the type of  
> schedule makes absolutely no difference.

Only if you assume that the pauses are a large fraction of tau -- the length  
of the programmed interreinforcement interval.

> Periods away from the bar create a spurious indication of an effect of a  
> changing reinforcement rate on the calculated rate of pressing. Spurious.

How? What difference does it make whether interresponse times are of a  
given length because the rat stood at the bar and pressed at roughly once  
per 10 seconds or left the bar for 9 seconds, then approached it and  
pressed? If the rat spends less and less time at the bar as the probability  
of reward given a press after a given delay declines, how can you say that  
the decline in time spent at the bar is not an effect of the diminishing  
probability of reward?

> And finally, if it is true that the rat spends more time away from the  
> lever when the scheduled intervals are longer, we can express this as tau =  
> f(I) where f is some positive monotonic function and I is the scheduled  
> interval. The result:

> mean pressing rate = P/(T+f(I))

> mean reinforcement rate = R/(T+f(I))

> Note that neither P nor R is a function of the interval, and the decrease  
> in mean pressing rate is not due to the decrease in the mean reinforcement  
> rate.

Would you mind defining your symbols? For some reason you've abandoned mine  
or worse, redefined them, without telling me what they now refer to. I  
think that what I was calling tau you now call I. What is T? What is the  
thing you are now calling tau?

> As to who is blind, I leave that to others to say. You have yet to admit  
> that there is a single thing wrong with the basic methods and analyses of  
> EAB, so you must think me an extraordinarily stupid person for seeing so  
> many problems (quite independently of what PCT has to say).

No, certainly not stupid. In fact, I think that you are an extraordinarily  
bright person. The trouble as I see it is that you are reasoning from a  
conclusion backward to the conditions that will make the conclusion come out  
the way you want it to, and those conditions simply do not exist in the case  
in point. As for EAB, raising that issue is a red herring. My reasoning  
here is not based on EAB but on a solid analysis of the nature of the  
environmental feedback function and real, honest-to-god rock-solid \_data\_.  
The data show that steady-state response rates decline as the programmed  
interreinforcement interval increases. The analysis, together with the  
data, shows that large changes in response rates have very little effect on  
obtained reinforcement rates over most of the range of rates observed. The  
observed reduction in response rates occurs in spite of the fact that  
reinforcement rates remain near their programmed values over most of the  
range of schedule values studied. The question to be answered is why a  
reduced probability of reward leads to a reduced probability of response  
when a reduced probability of response has little effect on the probability  
of reward. Why does the animal spend more and more time away from the bar  
as the programmed reinforcement rate decreases? But before we can get to  
that, you must first recognize that there is indeed a question to be  
anwered; that the relationship to be explained is not the artifact you claim  
it to be. Maybe then we can begin to talk models.

Regards,

Bruce

---

<div class="post-metadata">

**Author:** ![Marc\_Abrams5](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/m/f08c70/32.png) [@Marc\_Abrams5](http://discourse.iapct.org/u/Marc_Abrams5)\
**Post date:** [October 19, 1998, 7:17pm UTC](http://discourse.iapct.org/t/control-of-reinforcement/2993/64 "1998-10-19T19:17:55Z")

</div>

> From [Marc Abrams (981019.1500)]

> [From Rick Marken (981019.0900)]
> 
> Bruce Gregory (981019.1130 EDT)--  
> he just doesn't mean "control" in the sense that he thought

I

> meant "control". Bruce means "control" in the sense of  
> "influence" or "cause". In fact, in the beginning of his  
> textbook there is a chapter on "Explaining Behavior" in

which \>he has a section on "Exploring the Causes of  
Behavior". So \>Bruce would certainly agree with colleagues  
who say that \>the environment "controls" behavior because he  
would know \>that they mean "cause" when they say "control".  
One of the \>things PCT teaches is what control \_is\_.

> Once you know what control is you don't write chapters  
> about "the causes of behavior"; you write chapters about

the \>nature of control.

Sorry Bruce G. Rick is right. I suggested boiling Bruce  
Abbott in oil. I think that's getting him off lightly. Lets  
Burn the heretic at the stake.

> > If you want to play your hostile, defensive, and paranoid  
> > act, please don't.
> 
> Sorry. Gotta play. In my non-post-modernist world it's

called

> "teaching".

HA, HA, HA, HA, HA, HA. and I thought I was in ill health.  
What kind of meds are you taking for your delusions.

Bruce G:

> > One would almost think that the two of you imagine that  
> > the Bruces and Marc have somehow devoted their lives to  
> > undermining PCT, which only three people in the world  
> > understand to begin with. I speak for all of us when I

say \>\>that nothing could be further from the truth.

> I don't care what you say;

That's obvious, Good teaching practice.

> I care what you do. I don't really care whether or not

people \>want to "undermine" PCT; that's like trying to  
undermine f = \>ma. The basics of PCT are right and they will  
\_eventually\_ (many years from now) become "conventional  
psychology".

What is going to make that happen? What you do today effects  
what can be done tomorrow What are \_you\_ doing today ( since  
you are one of a handful of people who truly understand  
PCT ) that will make that tomorrow happen. Getting rid of  
Bruce Abbott won'y do it.

> What annoys me is people who say they love the  
> control theory model of behavior and who then go out and  
> publish a textbook on behavioral research that rejects  
> virtually every aspect of that model.

So? It is done. move on, there is \_nothing\_ to be gained by  
further discussion, Unless of course you believe in  
motivation and reinforcement.

> I guess I'd rather have enemies who are honest about how  
> they feel about PCT. What I've got now are enemies who are  
> busy telling me how much they love PCT while they are  
> beating me over the head every time I discuss some basic  
> implication of PCT.

"Enemies", what a nice way of talking about people who  
disagree with something you have said or done. Does someone  
taste your dinner before you eat it?

Marc

---

<div class="post-metadata">

**Author:** ![Bruce\_Gregory](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/b/4bbf92/32.png) [@Bruce\_Gregory](http://discourse.iapct.org/u/Bruce_Gregory)\
**Post date:** [October 19, 1998, 9:49pm UTC](http://discourse.iapct.org/t/control-of-reinforcement/2993/65 "1998-10-19T21:49:07Z")

</div>

[From Bruce Gregory 9981019.1745 EDT)]

Rick Marken (981016.1600)

> A CV is an observation that \_can\_ be made by one who knows how  
> to make it.

Temperature is a physical variable, but I would be loathe to call it an  
observation. What can be measured by a well defined procedure, is a  
temperature.

> Try my "Test for the Controlled Variable" demo at  
> [http://home.earthlink.net/~rmarken/demos.html](http://home.earthlink.net/~rmarken/demos.html). If you have  
> a friend do this demo then you can observe the variables your  
> friend is controlling.

What you can do is observe a motion and infer that it is relate to a  
controlled perception, but you cannot observe "a controlled variable". The  
temperature might be a controlled variable, but you can't observe this, you  
can only infer it. (The temperature might also be a dependent variable and  
the same time it is a controlled variable. Depends on the framework.)

> When the large square gets darkened, for  
> example, then the movements of the large square are under control;  
> so when you observe these movements you are observing a controlled  
> variable.

To be precise, you are observing the motions of a darkened square and  
inferring that this is what the other person is perceiving and controlling.

> Notice how "real" this controlled variable looks.

Again, all you are observing is the motion of the darkened square. The rest  
is inference.

> I bet  
> it doesn't seem any more theoretical than, say, the distance from  
> your hand above the floor (another observable variable that probably  
> is \_not\_ controlled).

The controlled variable is a theoretical construct. It cannot be observed,  
no how real it seems to you. I take it that you are making the point that  
conventional psychologists rarely if ever interpret their observations in  
terms of a framework that includes the construct "controlled variable." This  
seems quite correct to me.

Bruce Gregory

---

<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:** [October 19, 1998, 9:49pm UTC](http://discourse.iapct.org/t/control-of-reinforcement/2993/66 "1998-10-19T21:49:51Z")

</div>

[From Bill Powers (981019.1549 MDT)]

Bruce Gregory [981019.1130 EDT)--

Do you suppose that all this carping and complaining might qualify as a  
kind of drivel?

Best,

Bill P.

---

<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:** [October 19, 1998, 10:18pm UTC](http://discourse.iapct.org/t/control-of-reinforcement/2993/67 "1998-10-19T22:18:20Z")

</div>

[From Bill Powers (981019.1615 MDT)]

Bruce Abbott (981018.1250 EST)--

> You are not listening to me, Bill. Over most of the range of VIs  
> investigated, pauses were not sufficiently long to produce any such effect.

I was talking specifically about the part of the range where the pauses ARE  
sufficiently long to produce this effect. But I detect no will to see what  
I'm talking about, and I just don't have the energy it takes to keep trying  
to get it across. You are certain you're right. Enjoy.

Best,

Bill P.

---

<div class="post-metadata">

**Author:** ![R\_Marken](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/r/8797f3/32.png) [@R\_Marken](http://discourse.iapct.org/u/R_Marken)\
**Post date:** [October 19, 1998, 10:19pm UTC](http://discourse.iapct.org/t/control-of-reinforcement/2993/68 "1998-10-19T22:19:24Z")

</div>

[From Rick Marken (981018.1520)]

Bruce Gregory (981019.1745 EDT)--

> Temperature is a physical variable, but I would be loathe to  
> call it an observation. What can be measured by a well defined  
> procedure, is a temperature.

> What you can do is observe a motion and infer that it is relate[d]  
> to a controlled perception, but you cannot observe "a controlled  
> variable".

> To be precise, you are observing the motions of a darkened square  
> and inferring that this is what the other person is perceiving  
> and controlling.

> The controlled variable is a theoretical construct. It cannot be  
> observed, no [matter] how real it seems to you. I take it that you  
> are making the point that conventional psychologists rarely if ever  
> interpret their observations in terms of a framework that includes  
> the construct "controlled variable." This seems quite correct to  
> me.

Watch, Mary. I'm not going to say a thing.

Love

Rick

> **···**
>
> --  
> Richard S. Marken Phone or Fax: 310 474-0313  
> Life Learning Associates e-mail: rmarken@earthlink.net  
> [http://home.earthlink.net/~rmarken](http://home.earthlink.net/~rmarken)

---

<div class="post-metadata">

**Author:** ![Bruce\_Gregory](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/b/4bbf92/32.png) [@Bruce\_Gregory](http://discourse.iapct.org/u/Bruce_Gregory)\
**Post date:** [October 20, 1998, 12:24am UTC](http://discourse.iapct.org/t/control-of-reinforcement/2993/69 "1998-10-20T00:24:29Z")

</div>

[From Bruce Gregory (981019.2020 EDT)]

Rick Marken (981018.1520)]

> Bruce Gregory (981019.1745 EDT)--
> 
> \> Temperature is a physical variable, but I would be loathe to  
> \> call it an observation. What can be measured by a well defined  
> \> procedure, is a temperature.
> 
> \> What you can do is observe a motion and infer that it is relate[d]  
> \> to a controlled perception, but you cannot observe "a controlled  
> \> variable".
> 
> \> To be precise, you are observing the motions of a darkened square  
> \> and inferring that this is what the other person is perceiving  
> \> and controlling.
> 
> \> The controlled variable is a theoretical construct. It cannot be  
> \> observed, no [matter] how real it seems to you. I take it that you  
> \> are making the point that conventional psychologists  
> rarely if ever  
> \> interpret their observations in terms of a framework that includes  
> \> the construct "controlled variable." This seems quite correct to  
> \> me.
> 
> Watch, Mary. I'm not going to say a thing.

Thanks Rick. This is the most thoughtful and considerate response I  
have ever gotten from you.  
When you have nothing to say, say nothing. Mary's influence is  
salutary. Keep it up.

Bruce Gregory

---

<div class="post-metadata">

**Author:** ![Abbott\_Bruce](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/a/eb9ed0/32.png) [@Abbott\_Bruce](http://discourse.iapct.org/u/Abbott_Bruce)\
**Post date:** [October 20, 1998, 12:26am UTC](http://discourse.iapct.org/t/control-of-reinforcement/2993/70 "1998-10-20T00:26:07Z")

</div>

[From Bruce Abbott (981019.1925 EST)]

> Bill Powers (981019.1615 MDT) --

> Bruce Abbott (981018.1250 EST)

> > You are not listening to me, Bill. Over most of the range of VIs  
> > investigated, pauses were not sufficiently long to produce any such effect.

> I was talking specifically about the part of the range where the pauses ARE  
> sufficiently long to produce this effect. But I detect no will to see what  
> I'm talking about, and I just don't have the energy it takes to keep trying  
> to get it across. You are certain you're right. Enjoy.

But Bill, I've been telling you that this isn't the range in which most of  
the data lie, so your analysis does not apply over most of the VI values  
investigated. It's not that I'm unwilling to see what you're talking about,  
it's that what you are talking about is mostly irrelevant for the data under  
consideration.

If you are looking for an excuse to drop this discussion, fine. But don't  
blame me for it.

Regards,

Bruce

---

<div class="post-metadata">

**Author:** ![R\_Marken](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/r/8797f3/32.png) [@R\_Marken](http://discourse.iapct.org/u/R_Marken)\
**Post date:** [October 20, 1998, 5:29am UTC](http://discourse.iapct.org/t/control-of-reinforcement/2993/71 "1998-10-20T05:29:54Z")

</div>

[From Rick Marken (981018.2130)]

Bruce Abbott (981018.1250 EST) --

> The data show that steady-state response rates decline as the  
> programmed interreinforcement interval increases... Why does  
> the animal spend more and more time away from the bar as the  
> programmed reinforcement rate decreases? But before we can  
> get to that, you must first recognize that there is indeed  
> a question to be anwered; that the relationship to be explained  
> is not the artifact you claim it to be. Maybe then we can begin  
> to talk models.

Let's assume that the observed relationship between programmed  
reinforcement rate and response rate is not an artifact. What  
does this result have to do with control theory (I assume it is  
a control model that you wanted to talk about)? What evidence is  
there that any variable is under control in this situation?

Best

Rick

> **···**
>
> --
> 
> Richard S. Marken Phone or Fax: 310 474-0313  
> Life Learning Associates e-mail: rmarken@earthlink.net  
> [http://home.earthlink.net/~rmarken/](http://home.earthlink.net/~rmarken/)

---

<div class="post-metadata">

**Author:** ![Chris\_Cherpas](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/c/a3d4f5/32.png) [@Chris\_Cherpas](http://discourse.iapct.org/u/Chris_Cherpas)\
**Post date:** [October 20, 1998, 9:22am UTC](http://discourse.iapct.org/t/control-of-reinforcement/2993/72 "1998-10-20T09:22:39Z")

</div>

[From MC Average (981020.0200 PT)]

Welcome once again ladies and gentlemen to a  
dubious round of "find the controlled variable in  
the reinforcement schedule." But first let's read  
what viewers have written in about last week's show.  
Here's one out of the email bag sent in by Nat  
Schoenfeld from a zen monastary in New York, NY.

> **···**
>
> ==========================================================  
> Dear Abbott,  
> I have paid some artists to paint Xs on the ground in  
> central park that I can see from my window as I meditate.  
> My challenge to you is that before you can find the CV in  
> the CRF, FR, VR, FI, VI, or any other schedule you want  
> to name, that a dog will have covered each X before it  
> is covered by a falling autumn leaf. If I lose, I will  
> eat every leaf that covers each X first. If you lose,  
> you are free to study PCT without the burden of finding  
> the controlled variable in the reinforcement schedule.  
> Well, as I used to say before I retired, Tau tau!  
> ...or was it T Tau?
> 
> # PS: Listening to the sound of one hand clapping is even more boring than those old farts in William James Hall... send Fritos or something to snack on.
> 
> MC: Well, how do you respond to THAT stimulus?...although  
> you know that we know that I know that I don't mean THAT  
> STIMULUS, right?
> 
> Bruce Abbott (981018.1730 EST)--
> 
> > I'm sure that a perfectly good PCT model can be built  
> > that will handle these data.
> 
> MC: That's what I love about this show! And I thought the  
> reign of ad hocracy ended with the fall of futilism, if  
> I'm saying that correctly.
> 
> Bruce Abbott (981018.1250 EST)--
> 
> > Why does the animal spend more and more time away from  
> > the bar as the programmed reinforcement rate decreases?
> 
> MC: I would think any living being would be ready for more  
> and more time \_at\_ the bar, if not the opium den, as the  
> appearance of "reinforcement rate" on CSGnet increases.  
> Jesus, who's that yodeling from the back of the audience?
> 
> Jesus: I am the ghost of PCT past...
> 
> MC: It IS Jesus! Come on dowwwwn. Well, goddam, how'd  
> your crucifixion turn out? But really, what's up with  
> the "ghost of PCT past?"
> 
> Bruce Abbott (981018.1250 EST)--
> 
> > But before we can get to that, you must first  
> > recognize that there is indeed a question to be anwered;
> 
> Jesus: Hey, back off, post-modern whimps. Like I'm from  
> the past, I'm a ghost, I've got name-recognition -- just what  
> the hell do you require in a surprise game-show guest?
> 
> MC: Could you just park the bloody cross and get to the  
> point then?
> 
> Bruce Abbott (981018.1250 EST)--
> 
> > Maybe then we can begin to talk models.
> 
> Jesus: Maybe we can...what?
> 
> Bruce Abbott (981018.1730 EST)--
> 
> > First, we are talking about performance on variable-interval  
> > schedules, not ratio schedules.
> 
> Jesus: OK, THAT'S IT. I'd rather go back to being dead, but  
> this is way beyond turning cheek. Now the Jesus sandals  
> are coming off! I could stomach projectile-vomiting at each  
> other over a volley-ball net easier than hear you arguing  
> about rates of yada yada yada. You can't even find a decent  
> functional relation here, let alone find the controlled  
> variable in the reinforcement schedule, for chrissakes. And  
> maybe that's the point. I'm outta here....hey, no autographs.
> 
> MC: There you have it folks. Is Jesus just dropping more of  
> his body and blood around, or do we have a clue for finding  
> the missing CV? Will those dogs in central park force Bruce  
> into accepting that he is free to study PCT without bringing  
> back Jesus from the dead? And what's so bad about that anyway?
> 
> Bill Powers (981018.0922 MDT)--
> 
> > PCT does not predict what variables will prove to be  
> > controlled.
> 
> Nat Schoenfeld: You hear that, Bruce? You've gotta  
> think for yourself now.
> 
> MC: Whoa! Nat, how did you..?, I mean, Jesus, you look,  
> and smell,...bad.
> 
> Nat: Nope, wrong corpse -- I admit it, I'm dead too, but  
> young Bruce, if I were alive I'd rather piss on a hot  
> shock grid than watch you carry on this knight errantry trip  
> over reinforcment schedules and response rates. Listen, sonny,  
> the children's crusade is over with. We've already been there,  
> done that...(what's that those weirdos in LA are always saying?)
> 
> Rick Marken (981018.2130)--
> 
> > What evidence is there that any variable is under  
> > control in this situation?
> 
> Bruce Abbott (981018.1250 EST)--
> 
> > Why does the animal spend more and more time away from  
> > the bar as the programmed reinforcement rate decreases?
> 
> Nat: Why does the chicken shit in its own food?  
> Whipper snapper, don't try to pull that zen koan stuff  
> on me! Sure, nobody had ever seen anything quite like  
> those old cumulative records, did they? FI scallops,  
> FR steps, VI slopes that almost looked just right...
> 
> Almost. But face it, whatever was there was still too  
> vague to extend. And I'll admit "something" about these  
> schedules of having your food taken away was closer to  
> being quantified more sensitively with concurrents.  
> Closer. Why do you think I got so pushy over "not  
> responding?" You think I cared what the goy from  
> Pennsylvania thought?
> 
> We diced and sliced that stream of bar-presses more  
> ways than a vegematic, and sometimes one or another  
> way of resuffling the data revealed yet another partial  
> ordering. Sometimes, partial. No less, no more.
> 
> MC: Well, speaking of the ultimately temporary nature of  
> things, we've just run out of time for tonight's program.  
> Maybe next week's show could feature a procedure where  
> the rat gets to control some variable we can observe  
> and disturb systematically.
> 
> Until then, Nat, Jesus, thanks for stopping by;  
> Bruce, relax, have a sandwich. And remember  
> home viewers, an autumn leaf falling from a  
> tree may not be the controlled variable you  
> want to bet the park on.
> 
> Average regards,  
> MC

---

<div class="post-metadata">

**Author:** ![Bruce\_Gregory](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/b/4bbf92/32.png) [@Bruce\_Gregory](http://discourse.iapct.org/u/Bruce_Gregory)\
**Post date:** [October 20, 1998, 10:24am UTC](http://discourse.iapct.org/t/control-of-reinforcement/2993/73 "1998-10-20T10:24:17Z")

</div>

[From Bruce Gregory (981020.0620 EDT)]

Rick Marken (981018.1850)]

> Bruce Gregory (981019.2020 EDT)
> 
> \> When you have nothing to say, say nothing. Mary's influence is  
> \> salutary. Keep it up.
> 
> I had plenty to say. I just didn't say it.

I admire your self restraint. On a less ironic note, I had temporarily  
forgotten that PCT has its own way of viewing the world. From the PCT  
perspective, I am clear that controlled variables are as real as mass  
or velocity. They are features of the observable world, and the fact  
that convention psychologists ignore is simply a sign that they do not  
know what is important.

Bruce Gregory

---

<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:** [October 20, 1998, 10:52am UTC](http://discourse.iapct.org/t/control-of-reinforcement/2993/74 "1998-10-20T10:52:28Z")

</div>

[From Bill Powers (981020.0446 MDT)]

Bruce Gregory (981020.0620 EDT)--

> From the PCT  
> perspective, I am clear that controlled variables are as real as mass  
> or velocity.

That's right. No less real, and no more real.

When you're playing with the rubber bands, how real is the relationship of  
the knot to the dot? Are you observing it or imagining it? When you decide  
that a person is keeping the knot \_over\_ the dot, is it possible that the  
person is really maintaining the knot traveling in a circle around the dot?  
How possible?

Best,

Bill P.

Best,

Bill P.

---

<div class="post-metadata">

**Author:** ![Abbott\_Bruce](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/a/eb9ed0/32.png) [@Abbott\_Bruce](http://discourse.iapct.org/u/Abbott_Bruce)\
**Post date:** [October 20, 1998, 3:24pm UTC](http://discourse.iapct.org/t/control-of-reinforcement/2993/75 "1998-10-20T15:24:41Z")

</div>

[From Bruce Abbott (981020.1125 EST)]

> [From MC Average (981020.0200 PT)]

The satire is entertaining, Chris, and I enjoyed it as much as the next guy.  
And, of course, there are those who will find it a satisfying subtitute for  
rational argument. It is so much easier (and much more fun, too) to skewer  
with a rapier wit than to deal with tedious, unwelcome empirical data.

Are you suggesting that the rat's behavior on these schedules has nothing to  
do with the control of perception? The VI study was undertaken to ascertain  
what does happen on VI schedules as the interval size is varied. (I know,  
this has been done before, but I wanted to confirm it for myself and collect  
my own data on it.) The second step is to develop hypotheses as to what  
variable or variables the rat's actions served to control, and then to  
devise Tests of those hypotheses. The data collected rule out one simple  
hypothesis, proposed long ago by Bill Powers and often repeated until I  
provided the nasty bit of data that contradicted it. You (along with Rick  
Marken) apparently see little value in ruling out plausible alternatives.  
If a particular study fails to positively establish the precise nature of  
the CV, then in your views it is worthless, no matter what else the study  
brings to light.

I do not see any conceptual or practical problem in demonstrating that rats  
\_can\_ control certain variables. In fact the whole procedure amounts to a  
tautalogy. Take, for example, the usual human tracking study. If the  
person does as he or she is instructed, that is, keeps the cursor over  
target most of the time despite the disturbances to cursor (or target)  
position, then by definition the person is controlling the position of the  
cursor. If the person is unable to do so, then by definition the person  
isn't controlling the position of the cursor. Tell me that the person is  
controlling the position of the cursor and I will tell you what the person  
\_must\_ do (there's no alternatives) to accomplish that. The much-ballyhooed  
predictive accuracy of such an analysis is predicated on the fact that, if  
the person is doing the task as directed, then there really isn't any other  
way the person could be behaving.

It's like me saying to you, "stand over there." So, you move to the  
designated spot. Now I claim that I can predict with high precision your  
future location in space. I predict that, 30 minutes from now, you will be  
on the designated spot. Thirty minutes later, I check on your location. If  
you are still there where I left you, then I claim to have made an accurate  
prediction. If you are not, then I claim that it doesn't count, because you  
weren't following my directions.

I am not, as it may appear, belittling a control-system analysis or its  
predictive ability. The point is, if the subject is performing the task as  
directed, he or she has essentially \_no\_ degrees of freedom, and this  
accounts for the predictive accuracy of the analysis. If you can show that  
a variable is under control, then while that variable is being controlled,  
certain aspects of the subject's actions are under strong constraint. But  
if your theory can not predict what variables \_will\_ be controlled, under  
what circumstances, then you are left with the job of determining those  
facts empirically, on a case-by-case basis. To the extent that this control  
is idiosyncratic and subject to moment-by-moment changes in CVs and methods  
of CV control, the best you can do is either (a) arrange situations where  
what must be controlled and how in order to perform a given task are  
severely constrained (e.g., tracking tasks) or (b) resign yourself to merely  
explaining observed behavior, after the fact, by identifying what CV must  
have been controlled and how.

The VI schedule situation may not be ideal for demonstrating that rats can  
control \_some\_ variable with good precision (by the way, that's already been  
demonstrated; for example, rats will continually adjust the output of a heat  
lamp to maintain their body temperatures within a certain zone, when they  
are living in a refrigerator), but it raises some interesting questions  
about what the devil the rat \_does\_ control when performing on a VI  
schedule, and what variable or variables it must be perceiving in order to  
do it. It is an interesting case precisely because we cannot currently  
explain the results to our satisfaction, with PCT or any other theory.

I guess that whether you think research like this is valuable or not depends  
on whether your interests lie in demonstrating that PCT can accurately  
describe how some variable is being controlled (\_when\_ it is being  
controlled), or in trying to understand some puzzling bit of behavior for  
which an adequate explanation does not yet exist. I'm doing the latter.

Regards,

Bruce

---

<div class="post-metadata">

**Author:** ![R\_Marken](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/r/8797f3/32.png) [@R\_Marken](http://discourse.iapct.org/u/R_Marken)\
**Post date:** [October 20, 1998, 5:48pm UTC](http://discourse.iapct.org/t/control-of-reinforcement/2993/76 "1998-10-20T17:48:01Z")

</div>

[From Rick Marken (981020.1050)]

Bruce Abbott (981020.1125 EST)--

> The VI study was undertaken to ascertain what does happen on VI  
> schedules as the interval size is varied...The second step is  
> to develop hypotheses as to what variable or variables the rat's  
> actions served to control, and then to devise Tests of those  
> hypotheses.

I will certainly be looking forward to your report on that  
second step!

> The data collected rule out one simple hypothesis, proposed long  
> ago by Bill Powers and often repeated until I provided the nasty  
> bit of data that contradicted it.

I presume you are talking about the hypothesis that rate of  
reinforcement is controlled in operant conditioning experiments.  
It's great that you provided data showing that a particular  
variable is \_not\_ controlled; but that's just the start of the  
Test. Now what you need to do is figure out what variable \_is\_  
being controlled.

> If a particular study fails to positively establish the precise  
> nature of the CV, then in your views it is worthless, no matter  
> what else the study brings to light.

It's not worthless; it has just failed to establish the nature of  
the variable(s) under control. A study is worthwhile if it eliminates  
variables as possible controlled variables. But the aim of research  
on control is to find variables that \_are\_ controlled; and that's  
really not hard to do since organisms are controlling many variables,  
with rather high gain, at the same time.

> I do not see any conceptual or practical problem in demonstrating  
> that rats \_can\_ control certain variables.

I should hope not; that's what research on control is all about.

> In fact the whole procedure amounts to a tautalogy.

Now I see why you are not interested in doing research on  
control theory; it seems like a tautology to you.

> Take, for example, the usual human tracking study. If the person  
> does as he or she is instructed, that is, keeps the cursor over  
> target most of the time despite the disturbances to cursor (or  
> target) position, then by definition the person is controlling  
> the position of the cursor.

It would take to much time to explain why finding that the  
subject is controlling cursor position is 1) not a tautology  
or 2) true by definition. Instead, I will give you another  
example: a person catching a baseball. Now the person is  
instructed to "catch the ball". Can you tell, from these  
instructions, what variable(s) the person will be controlling  
if he follows the instructions? Of course not. You have to  
do Tests (based on hypotheses about the variables the person  
\_might\_ be controlling) to determine what the person is controlling.  
The variables a person is controlling when he follows your  
instructions to "catch the ball" are neither tautological nor  
true by definition. They are a very cool scientific discovery.

> I am not, as it may appear, belittling a control-system analysis  
> or its predictive ability.

No. You are belittling the importance of determining the variables  
organisms control. In other words, you are belittling the  
central insight of PCT; that the behavior you see is organized  
around the control of perceptual variables (controlled variables)  
that are by no means obvious to an observer.

> If you can show that a variable is under control, then while  
> that variable is being controlled, certain aspects of the  
> subject's actions are under strong constraint.

That's correct! And that's why PCT is so powerful. Once you  
know what variable(s) a person is controlling you can predict  
their behavior (actions) with incredible accuracy. The power  
of PCT in terms of predicting behavior (actions) comes from  
knowing the variables the organism controls. Since conventional  
psychology doesn't even know of the existence of controlled  
variables they can only get this level of predictive accuracy  
by chance.

> if your theory can not predict what variables \_will\_ be controlled,  
> under what circumstances, then you are left with the job of  
> determining those facts empirically, on a case-by-case basis.

Predicting what variables will be controlled under what  
circumstances can be done, in principle, once you have a pretty  
good map of an organism's control hierarchy. But right now  
the business of PCT is to determine, empirically, on a case-by-case  
basis, what variables organisms actually \_do\_ control.

> (by the way, that's already been demonstrated; for example, rats  
> will continually adjust the output of a heat lamp to maintain  
> their body temperatures within a certain zone, when they  
> are living in a refrigerator)

That's interesting. Have the researches modeled this? Have they  
thought of using a \_variable\_ disturbance (by having the temperature  
in the refrigerator vary over time)? This is a very interesting  
finding.

> I guess that whether you think research like this is valuable or  
> not depends on whether your interests lie in demonstrating that  
> PCT can accurately describe how some variable is being controlled  
> (\_when\_ it is being controlled), or in trying to understand some  
> puzzling bit of behavior for which an adequate explanation does  
> not yet exist. I'm doing the latter.

My interest is in both; but certainly in the latter. But, as  
a PCTer, when I see a puzzling bit of behavior, my inclination  
is to ask "what variable(s) might this organism be controlling  
in this situation that would lead to me to see this puzzling bit  
of behavior?" Then the hard part begins; testing for the controlled  
variable(s).

Best

Rick

> **···**
>
> --  
> Richard S. Marken Phone or Fax: 310 474-0313  
> Life Learning Associates e-mail: rmarken@earthlink.net  
> [http://home.earthlink.net/~rmarken](http://home.earthlink.net/~rmarken)

---

<div class="post-metadata">

**Author:** ![Marc\_Abrams5](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/m/f08c70/32.png) [@Marc\_Abrams5](http://discourse.iapct.org/u/Marc_Abrams5)\
**Post date:** [October 20, 1998, 7:55pm UTC](http://discourse.iapct.org/t/control-of-reinforcement/2993/77 "1998-10-20T19:55:54Z")

</div>

> From [Marc Abrams (981020.1428)]

Please excuse my incursion into this thread, I am trying to  
understand. I do that best when I can ask questions, please  
don't view them as a challenge to your knowledge.

> [From Rick Marken (981020.1050)]

To Bruce Abbott:

> It's great that you provided data showing that a particular  
> variable is \_not\_ controlled; but that's just the start of

the

> Test. Now what you need to do is figure out what variable  
> \_is\_ being controlled.

Can we agree that it is \_cv's\_ that are being controlled.  
Not \_a\_ cv? I don't think this is a minor point. From a  
research perspective does knowing or finding one cv help us  
find others or help us understand how they are formed and  
how they change? If so how?, Rick & Bruce this is not aimed  
at you in particular, it is aimed at anyone reading this  
post that might have some insight.

Bruce Abbott:

> > If a particular study fails to positively establish the

precise

> > nature of the CV, then in your views it is worthless, no  
> > matter what else the study brings to light.

Rick:

> It's not worthless; it has just failed to establish the

nature of

> the variable(s) under control. A study is worthwhile if it  
> eliminates variables as possible controlled variables. But

the \>aim of research on control is to find variables that  
\_are\_ \>controlled; and that's really not hard to do since  
organisms \>are controlling many variables, with rather high  
gain, at the \>same time.

Rick, you keep on saying it is not difficult to do ( find  
the variables ), If that is in fact the case maybe \_knowing\_  
what is controlled is not the difficult issue in research  
and is as you say, pretty obvious. Yet no one seems to be  
able to come up with experiments that show the dynamics of  
multiple cv's in action ( i.e. changing and being  
reprioritized ). maybe that is the more interesting issue?  
If so, any ideas ( modeling being one ) on how to research  
them. I think Jeff V's model and research is in this  
direction. Rick, how about converting your spreadsheet in  
Mind Readings to Vensim. I can help when your done, with  
developing a front end so people can input data and change  
parameters easily. I think it might also be a good beginning  
to a "PCT agent". I know it's not as much fun as bashing  
conventional Psychologists but I think that could be a major  
contribution for others interested in doin PCT research.

Bruce:

> > I do not see any conceptual or practical problem in  
> > demonstrating that rats \_can\_ control certain variables.

Rick:

> I should hope not; that's what research on control is all  
> about.

Rick, thanks for acknowledging that Bruce A \_cares\_ about  
PCT research. Just remember this for future events 🙂

Bruce

> > In fact the whole procedure amounts to a tautalogy.

Rick:

> Now I see why you are not interested in doing research on  
> control theory; it seems like a tautology to you.

Boy, that was quick :-). Rick try something new and unique.  
\_ASK\_ Bruce if that is the case. Novel idea huh 🙂

Bruce explains:

> > Take, for example, the usual human tracking study. If

the \>\>person does as he or she is instructed, that is, keeps  
the \>\>cursor over target most of the time despite the

> > disturbances to cursor (or target) position, then by  
> > definition the person is controlling the position of the  
> > cursor.

Rick's counter argument:

> It would take to much time to explain why finding that the  
> subject is controlling cursor position is 1) not a

tautology

> or 2) true by definition. Instead, I will give you another  
> example: a person catching a baseball. Now the person is  
> instructed to "catch the ball". Can you tell, from these  
> instructions, what variable(s) the person will be

controlling

> if he follows the instructions? Of course not. You have to  
> do Tests (based on hypotheses about the variables the  
> person \_might\_ be controlling) to determine what the person  
> is controlling.  
> The variables a person is controlling when he follows your  
> instructions to "catch the ball" are neither tautological

nor

> true by definition. They are a very cool scientific

discovery.

This discourse seems reasonable to me, But Rick you said  
Bruce was not interested in doing PCT research. I don't  
think that is an accurate or a fair statement.

Bruce

> > I am not, as it may appear, belittling a control-system  
> > analysis or its predictive ability.

Rick

> No. You are belittling the importance of determining the  
> variables organisms control. In other words, you are

belittling \>the central insight of PCT; that the behavior  
you see is \>organized around the control of perceptual  
variables

> (controlled variables) that are by no means obvious to an  
> observer.

Rick, i am reading Bruce differently then you on this.

You have made certain assumptions and accusations and have  
not confirmed them with Bruce. Several people have said that  
this is the primary reason we get into these nonsense  
threads on CSGnet I will try to avoid being part of it in  
the future.

Rick:

> That's correct! And that's why PCT is so powerful. Once you  
> know what variable(s) a person is controlling you can

predict

> their behavior (actions) with incredible accuracy.The power  
> of PCT in terms of predicting behavior (actions) comes from  
> knowing the variables the organism controls.

Here we disagree a bit. Knowing the cv's are only part of  
the story and not \_neccessarily\_ the most important part.  
we will only know through experimentation and testing )  
It's how they form, change, and are prioritized, that give  
PCT \_part\_ of it's predictive power. Let's not leave out the  
little old reference level which has some say in this and  
the kinds of actions we have learned to deal with  
disturbances, since these are continuous, rather than  
discrete events.

> Since conventional psychology doesn't even know of the  
> existence of controlled variables they can only get this

level \>of predictive accuracy by chance.

And at \_this\_ point what are we capable of?

Bruce:

> > if your theory can not predict what variables \_will\_ be  
> > controlled, under what circumstances, then you are left  
> > with the job of determining those facts empirically, on a  
> > case-by-case basis.

Ok, valid point

Rick:

> Predicting what variables will be controlled under what  
> circumstances can be done,

Other then the test ( which is impractical 1)because the  
test itself will change what people are controlling for 2)  
It is too time consuming) how can this be done? I will  
continue to ask this question as long as you continue to  
insist that it can be done.

But right now the business of PCT is to determine,  
empirically, on a case-by-case basis, what variables  
organisms actually \_do\_ control.

Why? What makes the CV any more important then any other  
component of the process

Marc

---

<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:** [October 20, 1998, 9:19pm UTC](http://discourse.iapct.org/t/control-of-reinforcement/2993/78 "1998-10-20T21:19:41Z")

</div>

[From Bill Powers (981020.1340 MDT)]

Bruce Abbott (981020.1125 EST) --

> Take, for example, the usual human tracking study. If the  
> person does as he or she is instructed, that is, keeps the cursor over  
> target most of the time despite the disturbances to cursor (or target)  
> position, then by definition the person is controlling the position of the  
> cursor. If the person is unable to do so, then by definition the person  
> isn't controlling the position of the cursor.

That really simplifies the modeling problem: the person either controls or  
doesn't control. One degree of freedom, two values, True or False. I guess  
I must have been doing it the long way around, by finding the integration  
constant, the delay, and the decay factor. If I'd known it was that easy I  
could have done this decades ago.

Actually, it's even simpler than that. If you want to get a person to  
juggle three balls and tapdance while whistling Dixie, withhold food for a  
while, then give the person a little bit of it every time he comes closer  
to doing these three things at the same time. How can he do all those  
things? Easy: he was rewarded for doing them. What if he doesn't do them?  
Probably the history of reinforcement made it impossible. Discard that  
subject and find one more susceptible to reinforcement.

> Tell me that the person is  
> controlling the position of the cursor and I will tell you what the person  
> \_must\_ do (there's no alternatives) to accomplish that.

No, you won't. You're forgetting something, aren't you?

> The much-ballyhooed  
> predictive accuracy of such an analysis is predicated on the fact that, if  
> the person is doing the task as directed, then there really isn't any other  
> way the person could be behaving.

Just for the few readers who aren't already aware that this is nonsense:  
What Bruce says is true for a perfect controller in the absence of all  
disturbances. But people are not perfect controllers, and in any proper  
control-system experiment, the controlled variable is subject to continual  
disturbances which neither the experimenter not the controller can predict  
because they're generated randomly during the task. The modeling problem is  
to create a simulation that behaves in the same way the person does, and  
handles unpredictable disturbances in the same way the person does --  
mistakes and all.

Furthermore, a model that does exactly what the instructions say -- keeping  
the cursor exactly over the target, for example -- would not match the real  
person's behavior nearly as well as a model that has its parameters matched  
to the person's actual parameters and thus generates similar tracking  
errors. And unless you could predict the disturbances, it couldn't predict  
the behavior -- the actions -- at all.

> It's like me saying to you, "stand over there." So, you move to the  
> designated spot. Now I claim that I can predict with high precision your  
> future location in space.

How, when all you said about location was "over there"? That's "precision?"

> I predict that, 30 minutes from now, you will be  
> on the designated spot. Thirty minutes later, I check on your location. If  
> you are still there where I left you, then I claim to have made an accurate  
> prediction. If you are not, then I claim that it doesn't count, because you  
> weren't following my directions.

Heck, I can do even better than that. I can predict that no matter what  
happens for the next 30 years (as long as you stay alive), you're going to  
be breathing oxygen. If you die, you violated my instruction to stay alive.

On the other hand, if you show me a model matched to Tom Bourbon's tracking  
performance, and give me a new data set for the same task, I can tell you  
whether Tom was the tracker, or someone else. If tracking were just  
"keeping the cursor over the target," how could I do that? All people would  
be alike.

> I am not, as it may appear, belittling a control-system analysis or its  
> predictive ability.

No, you're just misrepresenting it. Can it really be that after all this  
time you don't understand how the models are matched to the real behavior?

> The point is, if the subject is performing the task as  
> directed, he or she has essentially \_no\_ degrees of freedom, and this  
> accounts for the predictive accuracy of the analysis.

No, it doesn't. What accounts for the accuracy is finding a model with the  
necessary parameters and organization, and then varying the parameters to  
find the best fit of the model to the real behavior.

What the person is told to "do" is not to produce any specific behavior:  
it's to produce a particular perceptual result, by acting in any way that  
(invisible) disturbances make necessary. The "predictive accuracy" goes  
beyond just saying that the person will do the task; it includes predicting  
the exact trajectory of movements for the entire period of the task, and  
(as of the latest try using Vensim's matching routines) doing so within  
1.8% RMS of the range of handle movments. That is, the range of movements  
is 50 times the standard deviation of the mismatch between the model's  
movements and those of the real person. Want to calculate the probability  
that that match could have occurred by chance?

> If you can show that  
> a variable is under control, then while that variable is being controlled,  
> certain aspects of the subject's actions are under strong constraint. But  
> if your theory can not predict what variables \_will\_ be controlled, under  
> what circumstances, then you are left with the job of determining those  
> facts empirically, on a case-by-case basis. To the extent that this control  
> is idiosyncratic and subject to moment-by-moment changes in CVs and methods  
> of CV control, the best you can do is either (a) arrange situations where  
> what must be controlled and how in order to perform a given task are  
> severely constrained (e.g., tracking tasks) or (b) resign yourself to merely  
> explaining observed behavior, after the fact, by identifying what CV must  
> have been controlled and how.

That makes it sound as if the chance of discovering controlled variables  
"in the wild" are just about zero. What this amounts to is saying that we  
can't find any regularities in the effects that people produce by their  
actions. As I sit here before my keyboard, using my computer at a table  
with a flat surface and four legs, in a house with walls and a roof that do  
not leak or fall down, spelling words pretty much the same way every time,  
and being generally embedded in a very large matrix of highly predictable  
consequences of actions (my own and others'), I somehow doubt that  
controlled variables are all that hard to find. And once empirically found  
(you're right about that), it's not absurd to suppose that the same  
controlled variables will be controlled again, or will even continue to be  
controlled at similar values, or that we can identify higher levels of  
controlled variables that will permit us to predict how reference levels  
for lower ones will be set and changed.

> The VI schedule situation may not be ideal for demonstrating that rats can  
> control \_some\_ variable with good precision (by the way, that's already been  
> demonstrated; for example, rats will continually adjust the output of a heat  
> lamp to maintain their body temperatures within a certain zone, when they  
> are living in a refrigerator),

I doubt that they do that on a VI schedule....

> but it raises some interesting questions  
> about what the devil the rat \_does\_ control when performing on a VI  
> schedule, and what variable or variables it must be perceiving in order to  
> do it. It is an interesting case precisely because we cannot currently  
> explain the results to our satisfaction, with PCT or any other theory.

Oh, I think you can explain it with PCT, if you try. I think the rat is  
controlling something affected by food intake -- maybe the food intake  
itself. That much should not be hard to verify. Pinning it down more  
exactly will require adding disturbances to test each dimension of the  
proposed controlled variable, and showing that the rat's actions vary in  
the appropriate way. Of course using an experimental regime in which a huge  
amount of noise is deliberately inserted into the data makes this job much  
harder, and probably limits the precision with which you can identify the  
CV. You may end up saying just "the rat tries to maintain its average food  
intake."

> I guess that whether you think research like this is valuable or not depends  
> on whether your interests lie in demonstrating that PCT can accurately  
> describe how some variable is being controlled (\_when\_ it is being  
> controlled), or in trying to understand some puzzling bit of behavior for  
> which an adequate explanation does not yet exist. I'm doing the latter.

We should really be discussing the data, shouldn't we?

Bst,

Bill P.

---

<div class="post-metadata">

**Author:** ![Abbott\_Bruce](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/a/eb9ed0/32.png) [@Abbott\_Bruce](http://discourse.iapct.org/u/Abbott_Bruce)\
**Post date:** [October 21, 1998, 2:00am UTC](http://discourse.iapct.org/t/control-of-reinforcement/2993/79 "1998-10-21T02:00:33Z")

</div>

[From Bruce Abbott (981020.2200 EST)]

> Bill Powers (981020.1340 MDT)

> > Bruce Abbott (981020.1125 EST)

> > Take, for example, the usual human tracking study. If the  
> > person does as he or she is instructed, that is, keeps the cursor over  
> > target most of the time despite the disturbances to cursor (or target)  
> > position, then by definition the person is controlling the position of the  
> > cursor. If the person is unable to do so, then by definition the person  
> > isn't controlling the position of the cursor.

> That really simplifies the modeling problem: the person either controls or  
> doesn't control. One degree of freedom, two values, True or False. I guess  
> I must have been doing it the long way around, by finding the integration  
> constant, the delay, and the decay factor. If I'd known it was that easy I  
> could have done this decades ago.

When you think a remark is so stupid that it deserves a reply driping in  
sarcasm, it probably isn't. What did you miss?

> Actually, it's even simpler than that. If you want to get a person to  
> juggle three balls and tapdance while whistling Dixie, withhold food for a  
> while, then give the person a little bit of it every time he comes closer  
> to doing these three things at the same time. How can he do all those  
> things? Easy: he was rewarded for doing them. What if he doesn't do them?  
> Probably the history of reinforcement made it impossible. Discard that  
> subject and find one more susceptible to reinforcement.

Ouch, I must have \_really\_ hit a nerve.

> > Tell me that the person is  
> > controlling the position of the cursor and I will tell you what the person  
> > \_must\_ do (there's no alternatives) to accomplish that.

> No, you won't. You're forgetting something, aren't you?

Yes, I will, and no, I'm not. To move the cursor the person must move the  
mouse. As the cursor drifts left of target, the person must move the cursor  
to the right fast enough to offset the drift and return the cursor to  
target. As the cursor drifts right of target, the person must move the  
cursor to the left fast enough to offest the drift and return the cursor to  
target.

> > The much-ballyhooed  
> > predictive accuracy of such an analysis is predicated on the fact that, if  
> > the person is doing the task as directed, then there really isn't any other  
> > way the person could be behaving.

> Just for the few readers who aren't already aware that this is nonsense:  
> What Bruce says is true for a perfect controller in the absence of all  
> disturbances. But people are not perfect controllers, and in any proper  
> control-system experiment, the controlled variable is subject to continual  
> disturbances which neither the experimenter not the controller can predict  
> because they're generated randomly during the task. The modeling problem is  
> to create a simulation that behaves in the same way the person does, and  
> handles unpredictable disturbances in the same way the person does --  
> mistakes and all.

I'm not assuming that people are perfect controllers; in fact I'm well aware  
that they are not. When you set up a task in which you tell a person to  
keep the cursor over the target and show the person how to move the mouse to  
do so, then \_to the extent\_ that the person controls the cursor's position  
accurately, there are simply zero degrees of freedom in how that will be  
accomplished. In fact, for most well-practiced subjects on the tracking  
task, performance will be so close to optimal that the improvement in fit  
resulting from introducing appropriate integrative lag, etc. is likely to be  
small relative to the improvement in predictive ability gained by assuming  
that the person does the task perfectly.

> Furthermore, a model that does exactly what the instructions say -- keeping  
> the cursor exactly over the target, for example -- would not match the real  
> person's behavior nearly as well as a model that has its parameters matched  
> to the person's actual parameters and thus generates similar tracking  
> errors. And unless you could predict the disturbances, it couldn't predict  
> the behavior -- the actions -- at all.

As just noted, a model that does exactly what the instructions say will  
already account for most of the variance in mouse movement. That is the  
point I was making. I did not and do not dispute that a model with proper  
parameters may do slightly better.

As for the disturbances, they are introduced by the experimenter and do not  
need to be predicted in order to predict the behavior.

> > It's like me saying to you, "stand over there." So, you move to the  
> > designated spot. Now I claim that I can predict with high precision your  
> > future location in space.

> How, when all you said about location was "over there"? That's "precision?"

I watch where you go and stand. Now precision is a relative thing. It's  
extremely precise if your position could have been anywhere in the universe.  
Probably a reasonable standard is relative to where you could have gone to  
in the time alloted if you had been free to do so. After 30 minutes you  
could be miles away, so by comparison, locating your position within inches  
is excellent precision.

> Heck, I can do even better than that. I can predict that no matter what  
> happens for the next 30 years (as long as you stay alive), you're going to  
> be breathing oxygen. If you die, you violated my instruction to stay alive.

Now you're getting the idea. Given the conditional, there's not much choice  
in what must be done.

> On the other hand, if you show me a model matched to Tom Bourbon's tracking  
> performance, and give me a new data set for the same task, I can tell you  
> whether Tom was the tracker, or someone else. If tracking were just  
> "keeping the cursor over the target," how could I do that? All people would  
> be alike.

I'm not disputing the improvement gained by fitting proper constants to the  
model. What I'm saying is that the extraordinary precision of prediction  
claimed is due in large part to the fact that any person doing the task well  
couldn't be doing much of anything else (lags, overshoots, etc.); most of  
the variance in action could be accounted for simply by assuming that the  
person is accomplishing the task as directed, and examing the task to see  
what must be done to do so.

> > I am not, as it may appear, belittling a control-system analysis or its  
> > predictive ability.

> No, you're just misrepresenting it. Can it really be that after all this  
> time you don't understand how the models are matched to the real behavior?

No, it can't really be so. Can it really be that after all the interacting  
we've done, all the collaboration on modeling and so on, that you can still  
believe that I don't understand how the models are matched to the real  
behavior? Shouldn't warning bells be going off that maybe you aren't  
understanding what I'm getting at?

> > The point is, if the subject is performing the task as  
> > directed, he or she has essentially \_no\_ degrees of freedom, and this  
> > accounts for the predictive accuracy of the analysis.

> No, it doesn't. What accounts for the accuracy is finding a model with the  
> necessary parameters and organization, and then varying the parameters to  
> find the best fit of the model to the real behavior.

If you know what's involved in performing the task -- what the cv is going  
to be, and its means of control, you can model that without knowing anything  
about the person, and get very close to a person's performance, if the  
person does well on the task. The curve-fitting will improve on this, but  
if performance is already near optimum, it won't add much because there  
isn't much left to improve on.

> What the person is told to "do" is not to produce any specific behavior:  
> it's to produce a particular perceptual result, by acting in any way that  
> (invisible) disturbances make necessary. The "predictive accuracy" goes  
> beyond just saying that the person will do the task; it includes predicting  
> the exact trajectory of movements for the entire period of the task, and  
> (as of the latest try using Vensim's matching routines) doing so within  
> 1.8% RMS of the range of handle movments. That is, the range of movements  
> is 50 times the standard deviation of the mismatch between the model's  
> movements and those of the real person. Want to calculate the probability  
> that that match could have occurred by chance?

Is this before or after you employ curve-fitting to find the optimal  
parameters? The range of movements is close to the same 50 times the  
standard deviation of the mismatch if you assume the the person does as  
\_required\_ by the task -- perfectly. What do you think the probability is  
that you will admit that I have a valid point here? (I'm afraid it's  
vanishingly small . . .)

> > But  
> > if your theory can not predict what variables \_will\_ be controlled, under  
> > what circumstances, then you are left with the job of determining those  
> > facts empirically, on a case-by-case basis. To the extent that this control  
> > is idiosyncratic and subject to moment-by-moment changes in CVs and methods  
> > of CV control, the best you can do is either (a) arrange situations where  
> > what must be controlled and how in order to perform a given task are  
> > severely constrained (e.g., tracking tasks) or (b) resign yourself to merely  
> > explaining observed behavior, after the fact, by identifying what CV must  
> > have been controlled and how.

> That makes it sound as if the chance of discovering controlled variables  
> "in the wild" are just about zero.  
> . . .  
> I somehow doubt that  
> controlled variables are all that hard to find. And once empirically found  
> (you're right about that), it's not absurd to suppose that the same  
> controlled variables will be controlled again, or will even continue to be  
> controlled at similar values, or that we can identify higher levels of  
> controlled variables that will permit us to predict how reference levels  
> for lower ones will be set and changed.

I would hope that this would be the case, and in fact it's what I would  
expect. Yet in discussions on CSGnet it often sounds as if all a PCT-based  
science of behavior can do is predict what a person will do only after we  
know what he or she is trying to do, and that this changes constantly both  
between and within persons. By the time you painstakingly identify what the  
person is controlling here, under this condition, the information is out of  
date, because she's gone on to something else.

> Oh, I think you can explain it with PCT, if you try. I think the rat is  
> controlling something affected by food intake -- maybe the food intake  
> itself. That much should not be hard to verify. Pinning it down more  
> exactly will require adding disturbances to test each dimension of the  
> proposed controlled variable, and showing that the rat's actions vary in  
> the appropriate way. Of course using an experimental regime in which a huge  
> amount of noise is deliberately inserted into the data makes this job much  
> harder, and probably limits the precision with which you can identify the  
> CV. You may end up saying just "the rat tries to maintain its average food  
> intake."

Actually I don't think there's too much doubt about what the rat is  
controlling in the operant chamber by means of lever-pressing, although the  
required model is likely to become extremely complex. It presses the lever  
in order to produce a food pellet, it produces a food pellet in order to  
have food to pick up and consume, and it picks up and consumes food in order  
to -- well, this is where the complexity starts to show up. These are going  
on regardless of the schedule imposed. But I think VI schedules add  
something besides noise; the rat adapts to them and peforms differently than  
it does under, for example, ratio schedules.

> > I guess that whether you think research like this is valuable or not depends  
> > on whether your interests lie in demonstrating that PCT can accurately  
> > describe how some variable is being controlled (\_when\_ it is being  
> > controlled), or in trying to understand some puzzling bit of behavior for  
> > which an adequate explanation does not yet exist. I'm doing the latter.

> We should really be discussing the data, shouldn't we?

Well, yes, but you told me you weren't interested . . .

Regards,

Bruce

---

<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:** [October 21, 1998, 3:43pm UTC](http://discourse.iapct.org/t/control-of-reinforcement/2993/80 "1998-10-21T15:43:47Z")

</div>

[From Bill Powers (981021.0740 MDT)]

Bruce Abbott (981020.2200 EST)--

> When you think a remark is so stupid that it deserves a reply driping in  
> sarcasm, it probably isn't. What did you miss?

Nothing. Your generalization is not 100% accurate.

> > > Tell me that the person is  
> > > controlling the position of the cursor and I will tell you what the person  
> > > \_must\_ do (there's no alternatives) to accomplish that.
> 
> > No, you won't. You're forgetting something, aren't you?
> 
> Yes, I will, and no, I'm not. To move the cursor the person must move the  
> mouse. As the cursor drifts left of target, the person must move the cursor  
> to the right fast enough to offset the drift and return the cursor to  
> target. As the cursor drifts right of target, the person must move the  
> cursor to the left fast enough to offest the drift and return the cursor to  
> target.

Wrong. You're still forgetting something. Shall I tell you what it is, or  
would you rather figure it out for yourself?

> I'm not assuming that people are perfect controllers; in fact I'm well aware  
> that they are not. When you set up a task in which you tell a person to  
> keep the cursor over the target and show the person how to move the mouse to  
> do so, then \_to the extent\_ that the person controls the cursor's position  
> accurately, there are simply zero degrees of freedom in how that will be  
> accomplished.

Oh, well -- that wasn't it. What you're forgetting is that added to the  
effect of the mouse on the cursor is a LARGE disturbance derived from a  
random-number generator. Sometimes moving the cursor to the right requires  
a mouse movement to the right, sometimes a movement to the left, and  
sometimes no movement, depending on how the disturbance is changing. If you  
look at the pattern of mouse movement you will find essentially no  
relationship to the pattern of cursor movement or target movement. So how  
are you going to tell a person "how to move the mouse?"

> In fact, for most well-practiced subjects on the tracking  
> task, performance will be so close to optimal that the improvement in fit  
> resulting from introducing appropriate integrative lag, etc. is likely to be  
> small relative to the improvement in predictive ability gained by assuming  
> that the person does the task perfectly.

Of course -- provided the task is easy. With greater difficulty (larger or  
more rapid disturbances), the RMS tracking errors get larger, but the  
prediction error can actually get smaller, because there is more data about  
the error.

> > Furthermore, a model that does exactly what the instructions say -- keeping  
> > the cursor exactly over the target, for example -- would not match the real  
> > person's behavior nearly as well as a model that has its parameters matched  
> > to the person's actual parameters and thus generates similar tracking  
> > errors. And unless you could predict the disturbances, it couldn't predict  
> > the behavior -- the actions -- at all.
> 
> As just noted, a model that does exactly what the instructions say will  
> already account for most of the variance in mouse movement. That is the  
> point I was making. I did not and do not dispute that a model with proper  
> parameters may do slightly better.

"Slightly better" = RMS prediction error of 10% reduced to 2%.

The instructions say, "Keep the cursor over the target." They don't mention  
that the target will be moving randomly or that a large random disturbance  
will be added to the effect of mouse position. Even if the instructions did  
mention these disturbances, this information would be useless because there  
is no way to predict what the disturbances will be, and thus where the  
mouse will have to move.

> As for the disturbances, they are introduced by the experimenter and do not  
> need to be predicted in order to predict the behavior.

In order to predict the \_outcome\_, cursor over target. There is no way to  
predict where the mouse will move. The experimenter does not determine or  
know in advance what disturbance pattern will be applied to the live  
person; that is left up to a random number generator.

> > Heck, I can do even better than that. I can predict that no matter what  
> > happens for the next 30 years (as long as you stay alive), you're going to  
> > be breathing oxygen. If you die, you violated my instruction to stay alive.
> 
> Now you're getting the idea. Given the conditional, there's not much choice  
> in what must be done.

If you pick the right conditional, you can make your prediction as trivial  
as you please. But why make trivial predictions?

> > On the other hand, if you show me a model matched to Tom Bourbon's tracking  
> > performance, and give me a new data set for the same task, I can tell you  
> > whether Tom was the tracker, or someone else. If tracking were just  
> > "keeping the cursor over the target," how could I do that? All people would  
> > be alike.
> 
> I'm not disputing the improvement gained by fitting proper constants to the  
> model. What I'm saying is that the extraordinary precision of prediction  
> claimed is due in large part to the fact that any person doing the task well  
> couldn't be doing much of anything else (lags, overshoots, etc.); most of  
> the variance in action could be accounted for simply by assuming that the  
> person is accomplishing the task as directed, and examing the task to see  
> what must be done to do so.

The extraordinaty precision is largely due, I will admit, to the mere  
realization that the person is organized as a good tight negative feedback  
control system. You're saying that if a person is organized as a good  
negative feedback control system (the only way this task could possible be  
done well, or at all), then assuming perfect control will predict the  
results almost as well as measuring the actual parameters of control which  
are somewhat less than optimum. That says no more than that this person is  
acting as an almost perfect control system. What the instructions actually  
say is "be a control system." But isn't the whole point to explain how that  
could be done -- HOW the instructions could possibly be followed?

> > > The point is, if the subject is performing the task as  
> > > directed, he or she has essentially \_no\_ degrees of freedom, and this  
> > > accounts for the predictive accuracy of the analysis.

This clearly shows the point you want to make, but your concept of  
accounting for something is pretty weird. What PCT explains is how the  
subject could be performing the task as directed. The directions describe  
only what outcome is to be produced. They don't explain how, given the  
circumstances, it is possible for any organism to produce such outcomes.  
For that you need a control system model; no other model, reinforcement or  
otherwise, can explain it.

> If you know what's involved in performing the task -- what the cv is going  
> to be, and its means of control, you can model that without knowing anything  
> about the person, and get very close to a person's performance, if the  
> person does well on the task.

Tnat is to say, if you know the subject is organized as a good control  
system with an integrating output function, you can predict that the person  
will control well. Does that sound like a great relevation to you?

> > The "predictive accuracy" goes  
> > beyond just saying that the person will do the task; it includes predicting  
> > the exact trajectory of movements for the entire period of the task, and  
> > (as of the latest try using Vensim's matching routines) doing so within  
> > 1.8% RMS of the range of handle movments. That is, the range of movements  
> > is 50 times the standard deviation of the mismatch between the model's  
> > movements and those of the real person. Want to calculate the probability  
> > that that match could have occurred by chance?

> The range of movements is close to the same 50 times the  
> standard deviation of the mismatch if you assume the the person does as  
> \_required\_ by the task -- perfectly. What do you think the probability is  
> that you will admit that I have a valid point here? (I'm afraid it's  
> vanishingly small . . .)

Your statement is correct, and trivial. Of course if the organism is a very  
good control system, it will control in nearly the same way that a perfect  
control system will control. One of the main points of these experiments is  
to verify that the organism is, indeed, acting as a negative feedback  
control system rather than some other kind of system. That can be done by  
comparing performance to that of a perfect control system. But once that  
has been established, a more advanced consideration is to measure the  
parameters of control, to see how one person differs from another and how  
stable the parameters are over time. The same model we get from these  
experiments can be used to predict behavior when control is not so good --  
when the disturbances are larger and faster, so the tracking errors are  
much larger. And they can predict what will happen to control accuracy when  
the situation is changed -- when, for example, the proportionality constant  
in the environmental feedback function is doubled or halved, or when the  
EFF is changed from a proportional relationship to one involving one or two  
integrations (as in controlling the position of a mass on a spring with  
damping, or flying an airplane). One question of great interest is whether  
the parameters of control remain the same when the EFF changes, or if they  
change, and if they change, with what gain, delay, and integration factor  
the changes take place. If all you know is that the instructions were  
followed, there's no way to answer such questions.

> But I think VI schedules add  
> something besides noise; the rat adapts to them and peforms differently than  
> it does under, for example, ratio schedules.

You'll have to prove that. It remains possible that the rats are organized  
in exactly the same way they are under all other schedules, and it is the  
random variations in interval that create apparent differences in the rats'  
behavior.

Best,

Bill P.

[Previous page](http://discourse.iapct.org/t/control-of-reinforcement/2993.md?page=3)

[Next page](http://discourse.iapct.org/t/control-of-reinforcement/2993.md?page=5)