# Failed experiments?

**URL:** <http://discourse.iapct.org/t/failed-experiments/4292>\
**Category:** CSG2004\
**Created:** [December 3, 2004, 3:05am UTC](http://discourse.iapct.org/t/failed-experiments/4292 "2004-12-03T03:05:32Z")\
**Posts on this page:** 9\
**Page:** 1

<div class="post-metadata">

**Author:** ![Bruce\_Abbott1](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/b/838e76/32.png) [@Bruce\_Abbott1](http://discourse.iapct.org/u/Bruce_Abbott1)\
**Post date:** [December 3, 2004, 3:05am UTC](http://discourse.iapct.org/t/failed-experiments/4292/1 "2004-12-03T03:05:32Z")

</div>

[From Bruce Abbott (2004.12.02.2205 EST)]

> ## Rick Marken (2004.12.02.1410)
> 
> > Marc Abrams (2004.12.02.1423)]

> > I count 3 ‘failed’  
> > collaborations that we all know little about because Bill
> 
> > dismisses them out of hand. Work with Martin Taylor, Bruce Abbott,  
> > and most
> 
> > recently John Flach.
> 
> I think it would be _great_ to discuss these failures. I know that two  
> did
> 
> seem to involve failed experiments (Taylor and Abbott) and I think it  
> would
> 
> be useful to make clear what failed and what could be done to improve  
> the
> 
> experiments.

> > What I am interested in seeing  
> > is what varied from expectation and why.
> 
> I think that’s a very good idea. Perhaps we could start by having  
> Martin
> 
> Taylor and Bruce Abbott describe what they did and why they think  
> things
> 
> went wrong.
> 
> > I want to see the data, the facts. Not someone’s  
> > interpretation of them.
> 
> Me too. That’s a great idea!

Well, at the moment I don’t have access to the data (I’m at home and it’s  
at work) and even if I did, it would take more time to resurrect them  
that I have available at present. And I wouldn’t call what Bill Powers  
and I were doing together as “failed collaborations.” (I think  
we collaborated quite nicely!) But what are you counting as a  
“failed” experiment? I conducted two that were designed with  
PCT in mind, one replicating Staddon and Ettinger’s “cyclic  
ratio” study in which rats lever-pressed for food on ratio schedules  
and one examining food consumption and body weight changes in rats that  
could earn a part of their daily food requirement by lever-pressing in an  
operant chamber. Neither produced results that were “dismissed out  
of hand” by Bill Powers.  
In the cyclic ratio study, the ratio size was varied systematically over  
a range of 2 to 64 lever-presses per food-pellet delivery. Staddon and  
Ettinger proposed that rats in this situation are controlling for a  
certain rate of pellet delivery. The conversion of a given rate of  
lever-presses into a given rate of pellet delivery constitutes the  
environmental feedback function of the control system. As the ratio  
of presses to pellets is increased, it takes more presses to produce a  
food pellet, reducing the rate of pellet delivery for a given rate of  
pressing. Staddon and Ettinger predicted that, as the ratio increased,  
rats would increase their press-rate in an attempt to maintain the  
reference rate of pellet delivery, and that’s what they apparently found.  
In my replication of their study I found that the rate of pressing  
actually remained constant across the different ratios, and that the  
apparent increase in response rates that Staddon and Ettinger reported  
was actually an artifact of the way they analyzed their data. The rats  
were _not_ defending against the reduced pellet delivery rate  
resulting from the disturbance to that rate produced by increasing the  
ratio.

Although this result was disappointing (in that I was looking for  
evidence of control of pellet delivery rate and did not find it), the  
findings do not argue against PCT. They merely show that in this  
situation food pellet delivery rate was not a controlled variable.  
The animal may be lever-pressing not to maintain a rate of pellet  
delivery but to make a pellet available for consumption. Consuming the  
pellet disturbs this controlled variable (the pellet is no longer  
available for consumption) and lever-pressing then resumes.

The second project was developed by me with input from Rick and Bill  
during the design phase. The idea was to train a rat to press a lever for  
food pellets in the operant chamber during daily one-hour sessions.  
Generally the animal cannot earn enough food during a single one-hour  
session to maintain its body weight over a 24-hour period, so  
supplemental food is given in the home cage after the experimental  
session. The amount of supplemental food given in the home cage was  
varied over the course of the experiment and constituted a disturbance to  
the rat’s nutrient control system. Reducing the amount of supplemental  
food would require an increased rate of lever-pressing and food  
consumption in the operant chamber if the rat were to maintain a  
reasonably constant rate of food intake over successive days. Increasing  
the amount of supplement would require either that the rat not consume  
all the available supplement in the home cage or that is earn and consume  
less food in the operant chamber, if the overall level of food intake  
were to remain constant. Finally, if the level of nutrients to be  
consumed was specified by the level of stored body fat in a fat-control  
system, then it might be expected that changes in the reference would  
take place relatively slowly (as the animal gradually gained or lost  
weight). This would produce a lag between changing the amount of  
available supplement in the home cage and adjustment of consumption in  
the operant chamber.

That is more-or-less what we found. Unfortunately, the apparent  
level of control was relatively poor. Part of the problem was traced to  
apparent changes in the rats’ metabolic rates. It appears that the  
rats could counter disturbances both by changing their rates of food  
consumption and by changing the rates at which they “burned”  
their fuel. I had no direct way to measure these apparent changes in  
metabolic rates, so an important potential part of the actions used to  
control fat storage could not be included in the model.

But this wasn’t the whole problem. There were also limits on how much the  
rats would eat during the hour-long sessions in the operant chamber. Our  
model and plans had not included the possible role of short-term satiety  
in limiting the amount of food that would be consumed during these  
sessions. It was clear that the rats could hold more food (as they  
did when they had lost weight), but would quit eating before this limit  
was reached. This was most likely due to the food that had been consumed  
early in the session reaching the small intestine and its nutrient  
content starting to enter the bloodstream. The amount consumed was not  
enough to maintain weight when the home-cage supplement was small, and  
there was no way to make up for the loss later, even though the rat’s  
nutrient-control mechanism would normally produced another bout of  
feeding at that time if food had been available. Thus, in my  
interpretation, the restrictions my procedure imposed on feeding meant  
that short-term nutrient-control would lead to problems in maintaining  
longer-term fat-level control. As body weight was lost, the animal would  
increase the amount eaten in the operant chamber, but conflict with  
short-term control over blood-stream nutrient-levels and stomach loading  
meant that the rat would not increase its intake sufficiently to prevent  
further weight loss.

To make a long story short, we learned some things in this study and  
would do it differently now to avoid the problems encountered, but we  
still did see evidence of control at two levels even if the fit of model  
to data was not as good as it might have been if we had been able to  
measure metabolic rate changes and had not allowed short-term factors to  
conflict with long-term body-fat regulation (as measured by body  
weight).

I probably should add one more study to the mix. In this one I  
replicated previous studies showing that rate of lever-pressing for food  
on a variable-interval schedule declines as the average interval size  
increases. Increasing the average interval size reduces the average rate  
of pellet delivery, and Bill had predicted that the animals would  
increase their rates of responding as the interval size increased, in an  
effort to compensate for the reduced rate of pellet delivery imposed by  
the longer schedule. This demonstrates rather conclusively that Bill’s  
model was wrong (rate of pellet delivery was not acting like a controlled  
variable, just as it was not in the cyclic ratio study). However, on VI  
schedules, rates of lever-pressing beyond a certain minimum have little  
if any detectable effect of the rate of pellet delivery. Given that  
increasing the rate of pressing was not “working,” (in the  
sense of opposing the disturbance of lowered pellet-delivery rates), the  
animals might have been undergoing reorganization, which would entail an  
increased level of behavioral variability in an ecoli-type random  
“search” for a more effective way to produce food.

Bruce A.

---

<div class="post-metadata">

**Author:** ![Bozo\_the\_Clown](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/b/f07891/32.png) [@Bozo\_the\_Clown](http://discourse.iapct.org/u/Bozo_the_Clown)\
**Post date:** [December 3, 2004, 7:17am UTC](http://discourse.iapct.org/t/failed-experiments/4292/2 "2004-12-03T07:17:57Z")

</div>

From [Marc Abrams (2004.12.02.2243)]

Thanks for responding Bruce.

I used scare quotes around ‘failed’ for a very specific reason. It has been my understanding that neither party in the collaboration has been satisfied with the results to date.

I did not intend to intimate or suggest that the collaborations were unsuccessful _attempts_.

It is interesting Bruce, that Rick agreed with my assessment of how we (at least Rick and I) perceived the experiments, and Rick would not pass up such an opportunity to tell me all about it if he felt the attempt was satisfactory to all concerned.

You know how much I appreciate what you are doing. I think it’s important that all of us on CSGnet be aware of what is going on. I was unaware that you are currently working on a project with Bill.

Bruce, if I have learned anything recently it is that you never know where an idea might come from and who might see something you just couldn’t, or say something that triggers a new idea.

Bill Powers did more for me by asking two simple questions and then trying to provide an answer then he did and could, in the last 10 years. What he did was help clarify my thinking on a very central issue to me, but not what he intended.The questions; “What is the difference between the color red and blue or red and salty”?

It is in that spirit that I ask to see the data. I am not interested in critiquing your work. I’m interested in seeing whether method, goal, or design have any alternatives. What went or is not going the way either party expected? What was expected? Why do you think you are not seeing what you anticipated? What do you think your alternatives and choices are at this point?

I think behaviorism has many more problems than just thinking actions provide some insight into the mind. I think the large failure of behaviorism _IS NOT_ its failure to be able to predict ‘behavior’ properly. Powers has shown that behaviorism cannot predict _PURPOSE_ or intent and _WITH_ that behavior. I think this failure lies at the heart of the regeneration of interest in Cogsci after a 40 year period of dominance by behaviorism. It is also this fact and reason why most clinicians disdain ‘theorists’. It provides no useful insight into the work these people do. In fact behaviorism has never provided and answer to mind, and _THAT_ is what psychologists are after.

David Goldstein is a great example of this. I use David because he is familiar to most on this list. David has more methods and ‘theories’ then Carter has pills. This is not intended as a slam. This is a _REQUIREMENT_ for David to be an effective therapist. David has been involved with PCT for a very long time and holds a Ph.D. in Psychology. He is _NOT_ a dummy nor is he an ideologue. Yes, he has struggled, in my view, in trying how to ‘apply’ this wonderful concept of an input control system. The method of levels provide his best and his most comfortable use of the theory. Fred Nichols is a world class consultant. he has written any number of wonderful papers on organizational management and he two has struggled to figure out how to apply it and feigns ignorance when the only ignorance he has is his inability to date of figuring out how all this affects what he _SEE’S_. The same holds true for Ed Ford and others who are ‘applying’ PCT.

Anyone ‘applying’ PCT at this point has figured out a way of incorporating what PCT represents; that is, an input control model, and whatever other ‘theories’ they use to account for the things PCT cannot at this point and the purpose of their work and use.

I say all this Bruce before I even read your post because I want you to know up front what my intent was in asking to see your work.

I feel that _ANY_ work that is done with PCT cannot be a failure under any circumstances because if we are honest with ourselves we will learn just as well and maybe even more by ‘failure’ then we will by success. Failure provides, or I believe should provide, the incentive to find out what went wrong and why. Looking for disconfirmation to me is the corner stone to science and I am tickled pink to see Rick feel the same way. I hope he was not bluffing. I do not believe he was. We will see. It is never easy when your basic principles come into question.

Denial is a very powerful and helpful tool we all use to help us maintain any major unpleasant events. I believe denial is a control process, as I believe all cognitive processes are and control systems spare no one. They are absolutely _RUTHLESS_ in their attempts to reach its goal. But they have found some very interesting ‘variants’ in microbiology that prove to be useful. The concept of ‘BI-stability’ comes to mind.

Anyway, it is in this spirit Bruce that I read your response. I have come back here Bruce to tell you I did read the post before I wrote what you just read. I just saw the first paragraph and responded to that first.

After finishing my response I have come back here because this was an exceptional post and this provides the very reason why I think we need to share the work we do and if we feel we are in competition with each other that cannot happen in a profitable way. I am eager to hear your response to my comments. Again, thanks for sharing.

In a message dated 12/2/2004 10:11:30 PM Eastern Standard Time, [babbott@MFIRE.COM](mailto:babbott@MFIRE.COM) writes:

> [From Bruce Abbott (2004.12.02.2205 EST)]
> 
> Well, at the moment I don’t have access to the data (I’m at home and it’s at work) and even if I did, it would take more time to resurrect them that I have available at present.  
> OK, so your answer here is that you are not interested in going over old ground. No one asked for it this evening and no one asked for it within a certain time frame. If you don’t want to go here I have no problem with that. If you don’t feel it would be of any value to look at I fully respect your opinion. I just think it would be interesting for us on CSGnet to see the pro’s and con’s of a psychological experiment as conducted by a control theorist.

I did not intend to put you on the spot.

> And I wouldn’t call what Bill Powers and I were doing together as “failed collaborations.”  
> Neither would I. See above.

> (I think we collaborated quite nicely!)  
> Bruce, I could care less whether you and Bill ‘got along.’ I’m not Cindy Adams Why would you think I would think otherwise?

> But what are you counting as a “failed” experiment?  
> I don’t know. Whatever you did was never really gone over on CSGnet. I just remember certain posts where the ‘results’ were discussed and you and Bill seemed to disagree on what the results meant. No data was ever presented that I either remember or saw. If it was, please point me to the time frame if you can or the subject thread if you can remember, I have the CSGnet archive disk and would like to see the discussion if possible.

If you could, at your leisure and with some free time if you could get together _WHATEVER_ you feel might be worth looking at I would greatly appreciate it.

> Neither produced results that were “dismissed out of hand” by Bill Powers.  
> By this I meant that the results were not discussed on CSGnet extensively. And by extensively I mean that the results were not discussed among others on CSGnet besides you and Bill. Rick doesn’t seem to know much about them as well.

> In the cyclic ratio study, the ratio size was varied systematically over a range of 2 to 64 lever-presses per food-pellet delivery. Staddon and Ettinger proposed that rats in this situation are controlling for a certain rate of pellet delivery.  
> Why? I don’t mean to be ignorant, but what does it matter how fast or slow a rat presses the lever? What is this action supposed to represent?

> The conversion of a given rate of lever-presses into a given rate of pellet delivery constitutes the environmental feedback function of the control system. As the ratio of presses to pellets is increased, it takes more presses to produce a food pellet, reducing the rate of pellet delivery for a given rate of pressing. Staddon and Ettinger predicted that, as the ratio increased, rats would increase their press-rate in an attempt to maintain the reference rate of pellet delivery, and that’s what they apparently found.  
> And the significance of the original study by Staddon and Ettinger was supposed to show what?

Is this meant to show that a rat will maintain whatever behavior they need to in order to maintain or gain a certain amount of food? If not what did I miss?

> In my replication of their study I found that the rate of pressing actually remained constant across the different ratios, and that the apparent increase in response rates that Staddon and Ettinger reported was actually an artifact of the way they analyzed their data. The rats were _not_ defending against the reduced pellet delivery rate resulting from the disturbance to that rate produced by increasing the ratio.  
> Assuming my speculation about what the purpose was intended to show, I might ask these questions. How often did you repeat the experiment with the same rats? At the same and different times of the day. When and how often were the rats fed?

How do you know how a rat ‘analyzes’ data? That is, how do you make the claim that the behavior is due to an ‘analysis’ by the rat?

> Although this result was disappointing (in that I was looking for evidence of control of pellet delivery rate and did not find it),  
> Why would you be looking for the control of delivery rate? What does that represent? Why would a rat want to control the rate? Do rats eat everything available like dogs or do they eat like cats and eat just what they want at any given point in time, coming back as often as needed, but not ‘pigging’ out.

> the findings do not argue against PCT. They merely show that in this situation food pellet delivery rate was not a controlled variable. The animal may be lever-pressing not to maintain a rate of pellet delivery but to make a pellet available for consumption. Consuming the pellet disturbs this controlled variable (the pellet is no longer available for consumption) and lever-pressing then resumes.  
> To bad you can’t do the TesT on the rat. 🙂 But I think this is very useful because it shows one of the major problems with behavioristic experiments. If you can’t always figure the intent of another human by observing their actions How can you hope to do this when you have _ABSOLUTELY_ no idea about how a rat thinks. You might be able to know that a rat like all other living organism’s, have certain intrinsic needs and drives. And when looking at brain fMri’s and bodily chemical processes, we can make some inferences based on what they (other animals) and we (humans) have in common and share and what we don’t.

Trying to interpret what another organism is ‘thinking’ (or cogniting) or the intent is kind of difficult.

Do rats have ‘emotions’? I think animal studies on the mind are kind of meaningless if you are attempting to understand the higher levels. They are much more useful in looking at the biological processes and correlates.

Bruce, can you see off of this why clinicians might be a bit cynical about ‘theory’ when the basis is on that of a rat?

How about experiments that deal with introspection? I know its difficult, but if you have the ability to develop viable scales of feelings and emotions over time. That is, time series data dealing with cognitive correlates in humans.

The purpose of the measurements is not a way of getting ‘absolute’ values for these things, but to be able to _DERIVE_ ratio scale data for individuals that can be used as a means of measuring the ‘magnitude’ of these higher level concepts like ‘anger’ and ‘happiness’ and be able to compare people not based on the words people use but on the derived scales. I used this technique 20 years ago to great effect in my consulting work as a way to get people to see that the words people use are often misleading in terms of how they may actually feel and in trying to figure out how ‘angry’ someone needs to be before problems begin can be important. Looking back this would have been a very useful way of ‘experimenting’ with groups of people I was doing SD modeling work for. I used this in my pre-PCT days and of course this tool was not intended for this purpose but I think it would be wonderful.

> The second project was developed by me with input from Rick and Bill during the design phase. The idea was to train a rat to press a lever for food pellets in the operant chamber during daily one-hour sessions.  
> Why?

> Generally the animal cannot earn enough food during a single one-hour session to maintain its body weight over a 24-hour period, so supplemental food is given in the home cage after the experimental session.  
> Interesting. Are you saying the rat knows how to get food yet would starve itself rather then press the lever? Are you sure the rat can make the connection between lever press and food?

> The amount of supplemental food given in the home cage was varied over the course of the experiment and constituted a disturbance to the rat’s nutrient control system.  
> How do you know this? Do you always feel the same way after eating the same amount of food. There are days I can eat a house and others barely a nibble. With no changes in anything I do.

> Reducing the amount of supplemental food would require an increased rate of lever-pressing and food consumption in the operant chamber if the rat were to maintain a reasonably constant rate of food intake over successive days. Increasing the amount of supplement would require either that the rat not consume all the available supplement in the home cage or that is earn and consume less food in the operant chamber, if the overall level of food intake were to remain constant. Finally, if the level of nutrients to be consumed was specified by the level of stored body fat in a fat-control system, then it might be expected that changes in the reference would take place relatively slowly (as the animal gradually gained or lost weight). This would produce a lag between changing the amount of available supplement in the home cage and adjustment of consumption in the operant chamber.  
> And this finding would show what? Control? _IF_ this were the case I would have the same questions for you I currently have for Bill but unlike Bill you have no idea how a rat thinks.

So I think this shows _very nicely_ that rats _DO NOT_ control their actions. Which is in perfect harmony with PCT. What is controlled in PCT is _INTENT_ or purpose and _NOT_ actions, and you have no way of ever actually finding out why the rats did what they did?

Do you see my assessment differently?

> That is more-or-less what we found. Unfortunately, the apparent level of control was relatively poor.  
> Predictable though. Why did you expect the results to be otherwise?

> Part of the problem was traced to apparent changes in the rats’ metabolic rates. It appears that the rats could counter disturbances both by changing their rates of food consumption and by changing the rates at which they “burned” their fuel. I had no direct way to measure these apparent changes in metabolic rates, so an important potential part of the actions used to control fat storage could not be included in the model.  
> This is not, in my view, a ‘problem.’ These experiments, to me, seem go right to the core and root with what might be one of the limitations of behavioristic experiments. You have no know way of knowing how ‘hungry’ a rat feels and what produces that feeling. Even if you knew in general what chemical reactions and processes are involved, how, without ‘asking’ each rat how hungry they might be at any point in time? _AND_ if they are in the ‘mood’ to eat?

On the other hand this does _not_ make all of behaviorism ‘bad.’ I think Bill Powers and his use of introspection is both a very useful and necessary thing in order to study the ‘Mind.’ After all, all we ever know exists within the confines of our heads. So any experiments that are done with the intent of studying the ‘Mind’ must figure out a way of doing it by using individuals with introspection and I believe there are existing tools available for this purpose.

No, the ‘measuring’ tools will not allow for the reproduction of a human cognitive system, but I think we can get a pretty good idea about how all this happens even though we may not be able to ‘predict’ it for any one individual.

I think it could be like Darwin’s theory of evolution and that is a descriptive rather than a predictive theory

> But this wasn’t the whole problem. There were also limits on how much the rats would eat \>during the hour-long sessions in the operant chamber. Our model and plans had not \>included the possible role of short-term satiety in limiting the amount of food that would be \>consumed during these sessions. It was clear that the rats could hold more food (as they \>did when they had lost weight), but would quit eating before this limit was reached.

> This was most likely due to the food that had been consumed early in the session \>reaching the small intestine and its nutrient content starting to enter the bloodstream. The \>amount consumed was not enough to maintain weight when the home-cage supplement \>was small, and there was no way to make up for the loss later, even though the rat’s \>nutrient-control mechanism would normally produced another bout of feeding at that time if \>food had been available.

Well, this answers one of my questions.

> Thus, in my interpretation, the restrictions my procedure imposed on feeding meant that short-term nutrient-control would lead to problems in maintaining longer-term fat-level control. As body weight was lost, the animal would increase the amount eaten in the operant chamber, but conflict with short-term control over blood-stream nutrient-levels and stomach loading meant that the rat would not increase its intake sufficiently to prevent further weight loss.  
> Lets assume you’re correct in your assessment here. What is the significance of the finding? What does this mean in terms of human behavior?

Do you really think you could not conduct a test of this type on humans? That is, why use rats in an experiment of this type? I am not challenging you here Bruce, I’m ignorant about design and I’m eager to learn. I have asked you these questions because logically this is how I think. Not because I think I have some unique ability to assess psychological experimental designs 🙂 I want to know if I can why you do what you do.

> To make a long story short, we learned some things in this study and would do it differently now to avoid the problems encountered, but we still did see evidence of control at two levels even if the fit of model to data was not as good as it might have been if we had been able to measure metabolic rate changes and had not allowed short-term factors to conflict with long-term body-fat regulation (as measured by body weight).  
> So, the real reason for the experiment was to show the existence of control in a rat. Yes? Or was it to show a specific intent in the rat? Help me out here. 🙂

> I probably should add one more study to the mix. In this one I replicated previous studies \>showing that rate of lever-pressing for food on a variable-interval schedule declines as the \>average interval size increases. Increasing the average interval size reduces the average \>rate of pellet delivery, and Bill had predicted that the animals would increase their rates of \>responding as the interval size increased, in an effort to compensate for the reduced rate of \>pellet delivery imposed by the longer schedule. This demonstrates rather conclusively that \>Bill’s model was wrong (rate of pellet delivery was not acting like a controlled variable, just \>as it was not in the cyclic ratio study).

_VERY_ interesting. So Bill could not predict behavior either could he? Hmmmm. Maybe what you helped show in this experiment is that behavior, as PCT predicts is _NEVER_ controlled. It is _ALWAYS_ the _INTENT_ that is controlled, or the purpose if you will. That is, the _MIND_ needs to be the focus of attention _NOT_ the ACTIONS of an organism. What gets controlled is our interpretation (perceptions) of how well or poorly our perceptions match our intent or purpose.

Every action we produce is in response to some specific error. We learn that certain actions will generally produce certain results, so when we are hungry (error) we learn that putting something in our mouth and ingesting it will satisfy that need. That is reduce the error and by doing this we will ‘feel’ ‘better.’ The ‘reason’ we eat is to feel better. The 'reason we eat is also because of control. Some errors are cognitive and involve consciousness and others don’t.

> However, on VI schedules, rates of lever-pressing beyond a certain minimum have little if \>any detectable effect of the rate of pellet delivery.

Please explain the significance of this?

> Given that increasing the rate of pressing was not “working,” (in the sense of opposing \>the disturbance of lowered pellet-delivery rates),

An assumption and speculation in the sense that you really don’t know why this was happening.

> the animals might have been undergoing reorganization, which would entail an increased \>level of behavioral variability in an ecoli-\>type random “search” for a more effective way to \>produce food.

OK, so here you are saying you don’t have a real idea as to what is going on and you are attributing it to some unknown process or processes.

Please tell me what you agree with me on and what you don’t and most importantly why you feel the way you do. I have the greatest deal of respect for your efforts and if your experiments are like this, there is no doubt we can all learn a lot by them.

Maybe not what you intended , but learn nonetheless.

I think exposing experimental ideas might prove useful _IF_ folks were sincerely interested in helping _YOU_ and willing to discuss the issues. I for one am. I have learned a great deal in this one post and I hope with your response and subsequent clarifications and with hopefully the input of others, maybe we can all contribute to some future designs that are profitable for all.

Thanks a bunch Bruce.

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:** [December 3, 2004, 2:18pm UTC](http://discourse.iapct.org/t/failed-experiments/4292/3 "2004-12-03T14:18:05Z")

</div>

[From Bill Powers (2004.12.03.0640 MST)]

Bruce Abbott (2004.12.02.2205 EST)]

> > \> Marc Abrams (2004.12.02.1423)]
> 
> > I count 3 'failed' collaborations that we all know little about because  
> > Bill dismisses them out of hand. Work with Martin Taylor, Bruce Abbott,  
> > and most recently John Flach.

The very first time I met Marc Abrams was in Durango at a CSG meeting in  
the early 90s. He took me aside and tried to persuade me that Rick Marken  
was being "disloyal to PCT," was working against me, and should be got rid  
of. In the quoted paragraph above we see that this policy of  
destructiveness continues. Martin and Bruce can speak for themselves; my  
interactions with John Flach over a period of something like a year are in  
the public record. I spent hundreds of hours working on programs and  
writeups and learning from John how to do frequency-domain calculations,  
and toward the end John finally got the point of my main demonstration. He  
was very excited about it, but it was his decision not to abandon his old  
model (the McReuer and Jex model, which I had refuted, purporting to show  
adaptation in a human control system) and to go his own way. "Refute", by  
the way, does not mean just "deny," but "to disprove and overthrow by  
argument, evidence, or proof." I showed conclusively that no adaptation was  
required to explain the McReuer and Jex data; John agreed that I had done  
so, but did not want to give up the model. I don't know what he has done  
with this in the last year or so.

I have never dismissed anyone's ideas "out of hand," meaning immediately  
upon hearing them or for superficial reasons hastily arrived at. In fact, I  
have been told by many people that they don't understand why I have spent  
so much of my time working with certain people whose comments show mainly  
how little effort they have put into understanding PCT, or how completely  
they have distorted PCT to support their own previous ideas. I am not  
referring to Marc.

Best,

Bill P.

---

<div class="post-metadata">

**Author:** ![Bozo\_the\_Clown](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/b/f07891/32.png) [@Bozo\_the\_Clown](http://discourse.iapct.org/u/Bozo_the_Clown)\
**Post date:** [December 3, 2004, 5:19pm UTC](http://discourse.iapct.org/t/failed-experiments/4292/4 "2004-12-03T17:19:06Z")

</div>

From [Marc Abrams (2004.12.03.1121)]

In a message dated 12/3/2004 9:21:33 AM Eastern Standard Time, [powers\_w@FRONTIER.NET](mailto:powers_w@FRONTIER.NET) writes:

> [From Bill Powers (2004.12.03.0640 MST)]
> 
> The very first time I met Marc Abrams was in Durango at a CSG meeting in  
> the early 90s. He took me aside and tried to persuade me that Rick Marken  
> was being “disloyal to PCT,” was working against me, and should be got rid  
> of.  
> You twist both my words and intent, but I do understand your pain.

What I said to you was that Rick was being and having a negative influence on CSGnet because of his abrasive posts. These are words I told to Rick directly not once but many times over the years. I am not quite sure what you mean by ‘getting rid’ of him means? How do you ‘get rid’ of person? I suggested that you find a more appealing Tonto if you felt you needed one.

The fact that you felt compelled to reveal this now on CSGnet reveals more about your desperation than it does about anything else.

I am sorry you have viewed my posts as both harmful and hurtful. My intent was to get you to think about possibly ‘opening’ up your wonderful mind to an additional new thought or two.

Like many other things that have come into very clear focus for me the last several weeks, I realize who and what you are and what CSGnet is.

I am truly sorry for the pain I have caused you and the damage I might have inflicted. It was never my intent. I love your work and believe it or not feel extremely saddened to see you in such pain. This was never my intent. I wish I understood sooner what was involved.

You on the other hand need to understand _why_ what I have said was in fact ‘painful’ because you have to live with yourself.

> In the quoted paragraph above we see that this policy of  
> destructiveness continues. Martin and Bruce can speak for themselves; my  
> interactions with John Flach over a period of something like a year are in  
> the public record.  
> How do you think John Flach learned of Mary’s passing? I spoke to him and told him. I told him you needed some support and you were having a difficult time.

I spoke to John because I learned about his book on Control from the SD list. _Control for Humans_. I asked him what happened between the two of you and like George Richardson, John said he loved your work but you two had some philosophical differences. John, like George, _AND_ myself has the highest regard for your work, but like _many_ others Bill, we disagree with certain aspects of your model and or philosophy.

I wish you had the capacity to view this kind of thing as a way of learning rather than as a way of ‘slapping you down.’ But that is something you need to work through. I can only tell you Bill, if you are willing it is _NEVER_, _EVER_ to late. People really want to help Bill. But they can only do that from their own perspective. _NOT YOURS_. As long as you demand that others adhere to _YOUR_ way of thinking or the highway, you will have what you have and no more. If that is satisfactory to you and it seems to be, than so be it.

> I spent hundreds of hours working on programs and  
> writeups and learning from John how to do frequency-domain calculations,  
> and toward the end John finally got the point of my main demonstration. He  
> was very excited about it, but it was his decision not to abandon his old  
> model (the McReuer and Jex model, which I had refuted, purporting to show  
> adaptation in a human control system) and to go his own way.

Why did he refuse to give up his model if you ‘refuted’ it? Could it be like me here on CSGnet that John simply has a different perspective and different purposes then you do?

Why would John refuse your model if it was superior to his and provided a better way for him to get what he wanted?

> “Refute”, by  
> the way, does not mean just “deny,” but “to disprove and overthrow by  
> argument, evidence, or proof.” I showed conclusively that no adaptation was  
> required to explain the McReuer and Jex data; John agreed that I had done  
> so, but did not want to give up the model. I don’t know what he has done  
> with this in the last year or so.

Just something to think about Bill. If John thought your model provided no new insights to him, why aren’t you interested in understanding _WHY_ this is the case.

That is, he stuck with what he knew because he was comfortable with it. Apparently he saw no benefit in making the switch to something he did not fully understand.

I would want to know why that was the case.

Very perplexing, I will speak to John about this. He obviously believes you did not ‘refute’ his claim. There seems to be a bit of a contradiction here.

> I have never dismissed anyone’s ideas “out of hand,” meaning immediately  
> upon hearing them or for superficial reasons hastily arrived at. I

Bill, this is a _perfect_ example of what _I_ mean by ‘out of hand.’

By ‘out of hand’ I mean not reviewing the reasons why there was a disagreement and what the points of contention were, and to see if they were either intractable or not.

Apparently it is very painful for you to deal with ‘failure.’ That is _rejection_. Or in a slightly different view, people who have different opinions then you do. You seem to have some twisted sense of people looking to dominate you and dismissing _YOU_ and your ideas out of hand. That is, unwilling to hear your ideas. That unfortunately is what you learned to control for and you do one heckuva job doing it. You give as well as you get. You learned to control this _VERY_ well.

> n fact, I  
> have been told by many people that they don’t understand why I have spent  
> so much of my time working with certain people whose comments show mainly  
> how little effort they have put into understanding PCT, or how completely  
> they have distorted PCT to support their own previous ideas. I am not  
> referring to Marc.

Bill, what you have to realize is that _IF_ PCT _DOES NOT_ support their ideas they will not want to learn it.

Asking people to understand human behavior with the current PCT model is like asking a microbiologist to build a model of the worlds ecosystem. It might be theoretically possible but who the hell is going to do it. In 35 years you have gotten your answer. Maybe, just maybe its time to look at a different approach. That is of course, if in fact you are uncomfortable with the one you are currently in and I don’t think that is the case. I think you are _EXACTLY_ where you want to be. And that, to end this post with, has been the mistake for which I am truly sorry.

If I thought this is where you wanted to be I would _NOT_ have bothered nor will I in the future. I know my place.

Marc

Best,

Bill P.

---

<div class="post-metadata">

**Author:** ![Richard\_Marken2](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/r/edb3f5/32.png) [@Richard\_Marken2](http://discourse.iapct.org/u/Richard_Marken2)\
**Post date:** [December 3, 2004, 9:50pm UTC](http://discourse.iapct.org/t/failed-experiments/4292/5 "2004-12-03T21:50:22Z")

</div>

[From Rick Marken (2004.12.03.1350)]

> Marc Abrams (2004.12.03.1121)--

> I spoke to John because I learned about his book on Control from the SD list.  
> \_Control for Humans\_.

There is a pretty good review of that book available on the web at:

[http://www.mindreadings.com/BookReview.htm](http://www.mindreadings.com/BookReview.htm)

RSM

> **···**
>
> --  
> Richard S. Marken  
> MindReadings.com  
> Home: 310 474 0313  
> Cell: 310 729 1400
> 
> --------------------
> 
> This email message is for the sole use of the intended recipient(s) and  
> may contain privileged information. Any unauthorized review, use,  
> disclosure or distribution is prohibited. If you are not the intended  
> recipient, please contact the sender by reply email and destroy all copies  
> of the original message.

---

<div class="post-metadata">

**Author:** ![Bozo\_the\_Clown](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/b/f07891/32.png) [@Bozo\_the\_Clown](http://discourse.iapct.org/u/Bozo_the_Clown)\
**Post date:** [December 4, 2004, 1:13am UTC](http://discourse.iapct.org/t/failed-experiments/4292/6 "2004-12-04T01:13:34Z")

</div>

From [Marc Abrams (2004.12.03.2003)]

In a message dated 12/3/2004 4:54:00 PM Eastern Standard Time, [marken@MINDREADINGS.COM](mailto:marken@MINDREADINGS.COM) writes:

> [From Rick Marken (2004.12.03.1350)]
> 
> There is a pretty good review of that book available on the web at:
> 
> [BookReview2002](http://www.mindreadings.com/BookReview.htm)  
> It’s not an accurate review in my opinion. You missed the entire purpose and thrust of the book. It’s actually a very good intro to the field of control. Your recommendation of it being one part of a two part book with B:CP being the second book is in my opinion misplaced.

You are so tied into PCT you can’t see the forest for the tree’s.

I will not respond to any reply to this, just like I will not reply to the other two threads any longer. If you doubt my word, watch.

I don’t seek, want or need your approval.

Marc

---

<div class="post-metadata">

**Author:** ![Richard\_Marken2](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/r/edb3f5/32.png) [@Richard\_Marken2](http://discourse.iapct.org/u/Richard_Marken2)\
**Post date:** [December 4, 2004, 4:29pm UTC](http://discourse.iapct.org/t/failed-experiments/4292/7 "2004-12-04T16:29:46Z")

</div>

[From Rick Marken (2004.12.04.0830)]

Marc Abrams (2004.12.03.2003)--

Rick Marken (2004.12.03.1350)--

There is a pretty good review of that book available on the web at:

[http://www.mindreadings.com/BookReview.htm](http://www.mindreadings.com/BookReview.htm)

It's not an accurate review in my opinion. You missed the entire purpose and thrust of the book. It's actually a very good intro to the field of control.

Actually, I think I said that. I said that the book was a good introduction control theory, just not to control theory as applied to understanding living systems. The book presents control theory from the engineering perspective -- that of one who wants to build artifactual control systems -- rather than from the behavioral perspective -- that of one who wants to understand natural, living control systems.

Your recommendation of it being one part of a two part book with B:CP being the second book is in my opinion misplaced.

Why? It seemed well placed to me. It gives credit to Jagacinski and Flach for doing a competent job of presenting control theory while gently pointing out that what they fail to do is show how the theory can be used to understand the controlling done by living systems. Most notably, they don't explain how one determines the variables that a control system is controlling. That is, there is nothing in the book about testing for controlled variables.

You are so tied into PCT you can't see the forest for the tree's.

What forest am I missing?

RSM

> **···**
>
> ---  
> Richard S. Marken  
> [marken@mindreadings.com](mailto:marken@mindreadings.com)  
> Home 310 474-0313  
> Cell 310 729-1400

---

<div class="post-metadata">

**Author:** ![Bruce\_Abbott1](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/b/838e76/32.png) [@Bruce\_Abbott1](http://discourse.iapct.org/u/Bruce_Abbott1)\
**Post date:** [December 5, 2004, 6:59pm UTC](http://discourse.iapct.org/t/failed-experiments/4292/8 "2004-12-05T18:59:18Z")

</div>

[From Bruce Abbott (2004.12.05.1400 EST)]

> Marc Abrams (2004.12.02.2243) –
> 
> Bruce Abbott (2004.12.02.2205 EST)
> 
> In the cyclic ratio study, the ratio size was varied systematically  
> over a range of 2 to 64 lever-presses per food-pellet delivery. Staddon  
> and Ettinger proposed that rats in this situation are controlling for a  
> certain rate of pellet delivery.
> 
> Why? I don’t mean to be ignorant, but what does it matter how fast  
> or slow a rat presses the lever? What is this action supposed to  
> represent?

It’s not supposed to represent anything. It’s a sample of learned  
behavior that can be studied in the laboratory in an attempt to learn  
more about the factors that influence behavior. Because rats and  
humans have somewhat similarly organized nervous systems, then the  
principles identified by studying rats in such situations should tell us  
something about the principles underlying some forms of human behavior as  
well. More specifically, if the central thesis of PCT is correct, then  
rats should behave as living control systems, and behavior observed in  
the operant chamber and previously accounted for in terms of Skinnerian  
principles (i.e., selection by consequences) should be capable of being  
explained and predicted in terms of a PCT-consistent generative  
model.

Staddon and Ettinger had claimed to show that the rat in this  
experimental situation varies its rate of responding across ratio  
requirements so as to control the rate of food pellet delivery. This is  
consistent with a PCT model and seemed to demonstrate (in essence, via  
their version of the Test) that rate of food pellet delivery is a  
controlled variable for rats under these conditions. However, my analysis  
of their data indicated that their conclusion (that the rats acted to  
oppose the effect of increased ratio requirement by raising their  
response rates) was based on a faulty analysis of their data. To confirm  
my own analysis, I ran a replication of their study, which showed that  
indeed, the Staddon and Ettinger analysis was incorrect and that, in  
fact, rate of pellet delivery was not a controlled variable in this  
setting.

> The conversion of a given rate of lever-presses  
> into a given rate of pellet delivery constitutes the environmental  
> feedback function of the control system. As the ratio of presses to  
> pellets is increased, it takes more presses to produce a food pellet,  
> reducing the rate of pellet delivery for a given rate of pressing.  
> Staddon and Ettinger predicted that, as the ratio increased, rats would  
> increase their press-rate in an attempt to maintain the reference rate of  
> pellet delivery, and that’s what they apparently found.
> 
> And the significance of the original study by Staddon and Ettinger  
> was supposed to show what?

Rats learn to control and do control rate of food delivery. (At least in  
this respect, rats behave as living control systems.)

> Is this meant to show that a  
> rat will maintain whatever behavior they need to in order to maintain or  
> gain a certain amount of food? If not what did I miss?

Not amount of food but the rate at which it is produced and  
consumed.

> In my replication of their study I found that the  
> rate of pressing actually remained constant across the different ratios,  
> and that the apparent increase in response rates that Staddon and  
> Ettinger reported was actually an artifact of the way they analyzed their  
> data. The rats were not defending against the reduced pellet delivery  
> rate resulting from the disturbance to that rate produced by increasing  
> the ratio.
> 
> Assuming my speculation about what the purpose was intended to show,  
> I might ask these questions. How often did you repeat the experiment with  
> the same rats? At the same and different times of the day. When and how  
> often were the rats fed?

The animals were tested on six different ratios ranging from 2 to 64. In  
one phase of the study they were assessed six times on each ratio during  
each experimental session and tested daily for something like a month.  
Different rats were tested at different times of day. They were weighed  
after each experimental session and then given enough food in the home  
cage to keep their weights from increasing or decreasing across  
days.

> How do you know how a rat  
> ‘analyzes’ data? That is, how do you make the claim that the behavior is  
> due to an ‘analysis’ by the rat?

I should have written more clearly. Staddon and Ettinger analyzed  
the data, not the rats.

> Although this result was disappointing (in that I  
> was looking for evidence of control of pellet delivery rate and did not  
> find it),
> 
> Why would you be looking for the control of delivery rate? What does  
> that represent? Why would a rat want to control the rate? Do rats eat  
> everything available like dogs or do they eat like cats and eat just what  
> they want at any given point in time, coming back as often as needed, but  
> not ‘pigging’ out.

Because Staddon and Ettinger had concluded that rats do control the  
delivery rate. I didn’t think it was likely that they would. Rats eat in  
bouts or “meals,” about six or so each day.

> the findings do not argue against PCT. They  
> merely show that in this situation food pellet delivery rate was not a  
> controlled variable. The animal may be lever-pressing not to  
> maintain a rate of pellet delivery but to make a pellet available for  
> consumption. Consuming the pellet disturbs this controlled variable (the  
> pellet is no longer available for consumption) and lever-pressing then  
> resumes.
> 
> To bad you can’t do the TesT on the rat. 🙂 But I think this is  
> very useful because it shows one of the major problems with behavioristic  
> experiments. If you can’t always figure the intent of another human by  
> observing their actions How can you hope to do this when you have  
> _ABSOLUTELY_ no idea about how a rat thinks. You might be able to know  
> that a rat like all other living organism’s, have certain intrinsic needs  
> and drives. And when looking at brain fMri’s and bodily chemical  
> processes, we can make some inferences based on what they (other animals)  
> and we (humans) have in common and share and what we don’t.

There is no need to know how a rat “thinks” (if indeed it can  
be said to think at all). And you certainly can do the Test on a rat,  
just as with a human being. All you need to do is disturb some variable  
you think the rat may be controlling and observe whether the rat’s  
actions change so as to resist the effect of the disturbance on the  
variable.

> Trying to interpret what another organism is ‘thinking’ (or cogniting) or  
> the intent is kind of difficult.
> 
> Do rats have ‘emotions’? I think animal studies on the mind are kind of  
> meaningless if you are attempting to understand the higher levels. They  
> are much more useful in looking at the biological processes and  
> correlates.

They do appear to have emotions, although probably not the rich variety  
humans are capable of experiencing. The juveniles even engage in  
rough-and-tumble “play.” However, I wouldn’t choose rats as  
subjects for studying, say, pride. There are limits to their cognitive  
capacities.

> Bruce, can you see off of this why clinicians might be a bit cynical  
> about ‘theory’ when the basis is on that of a rat?

No. Animals such as laboratory rats are studied in the laboratory because  
in some respects they are similar enough to human beings that what we  
learn about them will help us to understand certain aspects of the human  
perception, behavior, emotion, and so on. Moreover, we can study those  
systems under conditions in which we have much greater control over  
variables than would be the case with a human being, and also can perform  
manipulations (e.g., drugs, brain stimulation) that we would not expose a  
human being to except under very special and limited circumstances (e.g.,  
the only hope the person has for improvement and the person has given  
permission to try it). With rats we can also rather quickly isolate  
genetic factors through selective breeding studies.

Such findings must always be tested to determine whether they extend to  
humans, but the basic research using rats (and other animals) helps to  
identify more clearly and efficiently those basic principles or  
influences that may be important in the human case and thus helps to  
guide human research and theoretical development.

> How about experiments that deal with introspection? I know its difficult,  
> but if you have the ability to develop viable scales of feelings and  
> emotions over time. That is, time series data dealing with cognitive  
> correlates in humans.

Those studies have their place and can provide useful hints that can  
guide the development of theory. HPCT provides a nice example of theory  
development guided partly by introspection.

> The purpose of the measurements is not a way of getting ‘absolute’ values  
> for these things, but to be able to _DERIVE_ ratio scale data for  
> individuals that can be used as a means of measuring the ‘magnitude’ of  
> these higher level concepts like ‘anger’ and ‘happiness’ and be able to  
> compare people not based on the words people use but on the derived  
> scales. I used this technique 20 years ago to great effect in my  
> consulting work as a way to get people to see that the words people use  
> are often misleading in terms of how they may actually feel and in trying  
> to figure out how ‘angry’ someone needs to be before problems begin can  
> be important. Looking back this would have been a very useful way of  
> ‘experimenting’ with groups of people I was doing SD modeling work for. I  
> used this in my pre-PCT days and of course this tool was not intended for  
> this purpose but I think it would be wonderful.
> 
> The second project was developed by me with input from Rick and Bill  
> during the design phase. The idea was to train a rat to press a lever for  
> food pellets in the operant chamber during daily one-hour sessions.
> 
> Why?

That was explained in succeeding statements.

> Generally the animal cannot earn enough food  
> during a single one-hour session to maintain its body weight over a  
> 24-hour period, so supplemental food is given in the home cage after the  
> experimental session.
> 
> Interesting. Are you saying the rat knows how to get food yet would  
> starve itself rather then press the lever? Are you sure the rat can make  
> the connection between lever press and food?

No, I’m saying that there wasn’t enough time for the rat to earn enough  
food to maintain body weight purely by lever-pressing for food in the  
operant chamber.

> The amount of supplemental food given in the home  
> cage was varied over the course of the experiment and constituted a  
> disturbance to the rat’s nutrient control system.
> 
> How do you know this? Do you always feel the same way after eating  
> the same amount of food. There are days I can eat a house and others  
> barely a nibble. With no changes in anything I do.

Rats in nature are usually able to maintain satisfactory levels of  
nutrition as evidenced by their ability to grow, stay healthy, and  
reproduce. Unless you assume that this is an accident, there must be some  
mechanism that is a work to keep food entering the system at a rate  
sufficient to prevent malnutrition. Of course, other factors affect the  
desire to eat besides having the level of food intake fall below the  
level required to maintain good nutrition. However, these could not  
“override” nutrient control for very long without producing  
nutritional deficiencies, and a rats in the lab normally don’t show  
these, it could be safely assumed that over the course of the study  
(which lasted over 300 days), the rat would be at least attempting to  
maintain a satisfactory state of nutrition, and under the circumstances  
this would involve maintaining an adequate rate of food intake. Thus, if  
we reduced the amount of food available in the home cage, at some point  
the rat would have to increase its consumption of food in the operant  
chamber in an effort to maintain the same rate of nutritional  
intake.

> Reducing the amount of supplemental food would  
> require an increased rate of lever-pressing and food consumption in the  
> operant chamber if the rat were to maintain a reasonably constant rate of  
> food intake over successive days. Increasing the amount of supplement  
> would require either that the rat not consume all the available  
> supplement in the home cage or that is earn and consume less food in the  
> operant chamber, if the overall level of food intake were to remain  
> constant. Finally, if the level of nutrients to be consumed was  
> specified by the level of stored body fat in a fat-control system, then  
> it might be expected that changes in the reference would take place  
> relatively slowly (as the animal gradually gained or lost weight). This  
> would produce a lag between changing the amount of available supplement  
> in the home cage and adjustment of consumption in the operant  
> chamber.
> 
> And this finding would show what? Control? _IF_ this were the case I  
> would have the same questions for you I currently have for Bill but  
> unlike Bill you have no idea how a rat thinks.

Yes, control, at at least two levels. (I have no idea what you are  
talking about in the last part of your second sentence.)

> So I think this shows _very nicely_ that rats _DO NOT_ control their  
> actions. Which is in perfect harmony with PCT. What is controlled in PCT  
> is _INTENT_ or purpose and _NOT_ actions, and you have no way of ever  
> actually finding out why the rats did what they did?

Not so fast! What is controlled in PCT are certain perceptions. Intent or  
purpose are words that describe what the actions are organized to  
accomplish, which is to bring the controlled perception near some  
reference value and keep it there by opposing disturbances to that  
perception. They are not what is controlled. As for finding out why the  
rats did what they did, that is what the research is all about –  
narrowing down the possible reasons until you arrive at one that  
satisfactorily accounts for the data.

> Do you see my assessment differently?

See above.

> That is more-or-less what we found.  
> Unfortunately, the apparent level of control was relatively poor.
> 
> Predictable though. Why did you expect the results to be  
> otherwise?

Predictable based on what? Our predictions were based on the reasonable  
hypothesis that these rats would behave so as to keep nutrient intake and  
body weight under control. However, as noted, the findings did not come  
out quite as expected, for reasons some of which we could only speculate  
about. (These hypotheses could be tested in further research.)

> Part of the problem was traced to apparent changes  
> in the rats’ metabolic rates. It appears that the rats could  
> counter disturbances both by changing their rates of food consumption and  
> by changing the rates at which they “burned” their fuel. I had  
> no direct way to measure these apparent changes in metabolic rates, so an  
> important potential part of the actions used to control fat storage could  
> not be included in the model.
> 
> This is not, in my view, a ‘problem.’ These experiments, to me, seem  
> go right to the core and root with what might be one of the limitations  
> of behavioristic experiments. You have no know way of knowing how  
> ‘hungry’ a rat feels and what produces that feeling. Even if you knew in  
> general what chemical reactions and processes are involved, how, without  
> ‘asking’ each rat how hungry they might be at any point in time? _AND_ if  
> they are in the ‘mood’ to eat?

It is not necessary to know how hungry a rat feels. What is necessary is  
the ability to identify and measure those variables that determine how  
much the rat will eat when given the chance. Assuming that the rat  
experiences hunger, a scientific analysis must begin by assuming that  
this hunger is a function of system variables within the rat, not by some  
magical nonphysical entity called “the mind” that for no  
apparent reason just “gets in the mood” (or not) to  
eat.

> On the other hand this does _not_ make all of behaviorism ‘bad.’ I think  
> Bill Powers and his use of introspection is both a very useful and  
> necessary thing in order to study the ‘Mind.’ After all, all we ever know  
> exists within the confines of our heads. So any experiments that are done  
> with the intent of studying the ‘Mind’ must figure out a way of doing it  
> by using individuals with introspection and I believe there are existing  
> tools available for this purpose.

Probably most of what goes on in our brains does not even register in  
consciousness, and is therefore out of reach of introspection. And what  
does register in consciousness can be misleading, because we can take it  
to be a cause of behavior when what we are observing is itself just  
further behavior (if of a private sort) that itself needs to be  
explained.

> To make a long story short, we learned some things  
> in this study and would do it differently now to avoid the problems  
> encountered, but we still did see evidence of control at two levels even  
> if the fit of model to data was not as good as it might have been if we  
> had been able to measure metabolic rate changes and had not allowed  
> short-term factors to conflict with long-term body-fat regulation (as  
> measured by body weight).
> 
> So, the real reason for the experiment was to  
> show the existence of control in a rat. Yes? Or was it to show a specific  
> intent in the rat? Help me out here. 🙂

The former.

> > I probably  
> > should add one more study to the mix. In this one I replicated  
> > previous studies \>showing that rate of lever-pressing for food on a  
> > variable-interval schedule declines as the \>average interval size  
> > increases. Increasing the average interval size reduces the average  
> > rate of pellet delivery, and Bill had predicted that the animals  
> > would increase their rates of \>responding as the interval size  
> > increased, in an effort to compensate for the reduced rate of \>pellet  
> > delivery imposed by the longer schedule. This demonstrates rather  
> > conclusively that \>Bill’s model was wrong (rate of pellet delivery was  
> > not acting like a controlled variable, just \>as it was not in the  
> > cyclic ratio study).
> 
> _VERY_ interesting. So Bill could not predict behavior either could he?  
> Hmmmm. Maybe what you helped show in this experiment is that behavior, as  
> PCT predicts is _NEVER_ controlled. It is _ALWAYS_ the _INTENT_ that is  
> controlled, or the purpose if you will. That is, the _MIND_ needs to be  
> the focus of attention _NOT_ the ACTIONS of an organism. What gets  
> controlled is our interpretation (perceptions) of how well or poorly our  
> perceptions match our intent or purpose.

I don’t follow you here. PCT does not specify in advance what perceptions  
will be under control under given circumstances, nor does it specify what  
means will be used to exert that control. It provides a theoretical  
framework for constructing a generative model (a hypothesized mechanism)  
that can be tested against data to determine its veracity. Bill’s  
original prediction for how the rat’s behavior would change across  
schedule parameters was based on a particular PCT-consistent model, and  
that model clearly failed the test (the changes in behavior were opposite  
to those predicted by the model). This fact does NOT show that  
“behavior is never controlled.” It shows that under these  
conditions the rats were not using rate of lever-pressing as an attempted  
means of controlling the rate of food delivery.

> Every action we produce is in response to some specific error. We learn  
> that certain actions will generally produce certain results, so when we  
> are hungry (error) we learn that putting something in our mouth and  
> ingesting it will satisfy that need. That is reduce the error and by  
> doing this we will ‘feel’ ‘better.’ The ‘reason’ we eat is to feel  
> better. The 'reason we eat is also because of control. Some errors are  
> cognitive and involve consciousness and others don’t.
> 
> > However, on VI schedules, rates of lever-pressing  
> > beyond a certain minimum have little if \>any detectable effect of the  
> > rate of pellet delivery.
> 
> Please explain the significance of this?

During acquisition of lever-pressing, the rats learn that, up to a point,  
increasing their rate of lever-pressing increases the rate at which they  
receive and can ingest food. However, because of the nature of VI  
schedules, above a certain rather minimal rate of responding, further  
increases in rate have almost ability to increase the rate of food  
delivery further. When the average interval size of the schedule is  
increased, the rate of food delivery falls even if the rate of  
lever-pressing remains the same. If the rats are trying to keep the  
rate up where it was before, they will increase their rate of lever  
pressing. However, because of the nature of the VI schedule, this  
change in response rate will actually have very little effect in opposing  
the decrease in pellet delivery rate.

> > Given that increasing the rate of pressing was  
> > not “working,” (in the sense of opposing \>the disturbance of  
> > lowered pellet-delivery rates),
> 
> An assumption and speculation in the sense that you really don’t know why  
> this was happening.

It is neither an assumption nor speculation that increasing the rate of  
responding doesn’t “work” to restore the rate of pellet  
delivery after an increase in the average interval size. It’s a  
demonstrable fact. What is speculation is that this failure of an  
increase in response rate to help (i.e., that the  
change in output action is not effectively countering the disturbance to  
pellet delivery rate) sets the stage for a search for alternative means  
of control. Another possibility is that rate of pellet delivery was never  
a controlled variable.

> > the animals might have been undergoing  
> > reorganization, which would entail an increased \>level of behavioral  
> > variability in an ecoli-\>type random “search” for a more  
> > effective way to \>produce food.
> 
> OK, so here you are saying you don’t have a real idea  
> as to what is going on and you are attributing it to some unknown process  
> or processes.

It’s speculation, yes. Research is like that. Findings may fail to  
support a model, which leads to the generation of alternative  
explanations (models) that can then be submitted to test. It’s how  
science progresses.

I hope you find this reply satisfactory, Marc. I’m going to be quite busy  
with term papers and finals over the next two weeks and don’t expect to  
be contributing further to this discussion at least for the time  
being.

Bruce

---

<div class="post-metadata">

**Author:** ![Bozo\_the\_Clown](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/b/f07891/32.png) [@Bozo\_the\_Clown](http://discourse.iapct.org/u/Bozo_the_Clown)\
**Post date:** [December 5, 2004, 10:49pm UTC](http://discourse.iapct.org/t/failed-experiments/4292/9 "2004-12-05T22:49:03Z")

</div>

Bruce,

_THANK YOU_, I _REALLY_ mean it.

I spoke to you about ECAC’s and I have decided to start my own Yahoo Group; Cognitive Control Theory.

Right now it is myself, Bruce Gregory and Ralph Levine. I am inclosing in this post the description anyone would find if they went onto yahoo and found the group.

I would be honored if you would think about joining our effort, and in the pursuit of that effort, I would like to use the thread you started on CSGnet with your post on ‘failed’ experiments and include this one as well. I would like to take this to the Yahoo list. What do you say? Please consider

I am _NOT_ trying to put you on the spot here Bruce. It is _EXTREMELY_ important that we understand what the Bruce Abbott’s of this world both like and dislike about what we are attempting to do.

Here is the description of the group

**Title:**  
Cognitive Control Theory  
**Description:**  
In taking up Dr. Rodolfo Llinas argument that human cognition or ‘mindness’ as he refers to it, is an empirical problem and not a philosophical one. Cognitive Control Theory (CCT) is an attempt to model through computer simulation ‘mindness’ as described by Llinas in his book _i of the vortex_.

Mindness or cognition has been an elusive topic of study for a long period of time. Human behavior as we know it is a function of cognition and Psychology as a clinical science depends on it.

Through discussion, research, and modeling, we hope to develop a theory of human cognition that is fully compatible with clinical psychology and of interest to all those interested in the fields of social science including economics, political science, and education.

It’s very much of a long shot and we are in it more for the ‘ride’ than actually thinking we can accomplish this completely, but at the heart of the quest lies a profound interest in what Llinas calls the _I of the vortex_.

This group is open to all adults interested in the topic and all professionals who might be able to contribute. We only ask that you have an open mind and a willingness to look at new ideas and accept the ideas of others. A reading of Llinas Book _i of the vortex_ would be useful but not a requirement. An understanding of System Dynamics and Perceptual Control Theory (PCT) might also help your ability to enjoy, partake and contribute to our purpose and _your_ ultimate benefit.

If you are interested please drop me a line @ [msabraams50@yahoo.com](mailto:msabraams50@yahoo.com) and just give we your real name, what your purpose and interest is in the group. He have no requirements. All are welcome, we just like knowing who we are talking with and what their interest in the group might be.

In a message dated 12/5/2004 2:03:25 PM Eastern Standard Time, [babbott@MFIRE.COM](mailto:babbott@MFIRE.COM) writes:

> **···**
>
> > [From Bruce Abbott (2004.12.05.1400 EST)]
> > 
> > > Marc Abrams (2004.12.02.2243) –