Autism, Behaviour-Modification, and PCT: a case brought by a parent

On Autism and Belief


Bill explained to Eileen, a mother of an autistic child, the principle of behavioural modification and made the distinction between behaviourists and control theorists. He outlined ways of observing and understanding what controlled variables an individual control regardless he/she was autistic or not.

Date: Oct 5, 1991

Author: Eileen Prince

Does anyone have any ideas about new ways of “treating” autism. I joined this conference because of applications to ESL (and I’ve been too swamped to really get involved). Hopefully, I’ll be able to become a more active participant. However, in the meantime, I have an almost 13-year-old now basically non-verbal (limited ASL) autistic daughter. Without spending a long time on history, let me just say that although I do in fact appreciate the contributions the B-mod (Behavioural Modification) approach used at the May Institute has made to her ability to learn, I am quite unhappy with their whole approach to communication (purely B-mod, not with a goal of learning any kind of real langauge). Any ideas or comments?

Thank you Eileen Prince

Date: Oct 06, 1991

Subject: CT (Control Theory) and autism

Author: William T. Powers.

Eileen Prince

Welcome to CSGnet, Eileen. There will be many listening who wish they could dip into the resources of control theory (or any theory) and come up with a solution that would help your daughter. I wish I could. At least you can be sure that many people will be racking their brains trying to think of something useful to say.

The basic problem is that no one who understands control theory has studied autistic people to see how the theory can be applied – no one I know about. Control theory isn’t like most other theories: it doesn’t say that if X happens to people, Y will be the resulting effect on their behavior. It’s about the way behavior works; it describes relationships of a very general nature between perception and action. At the same time, it is a theory of individual behavior: in order to apply it to an individual, one must determine what variables that individual is controlling and with respect to what internally specified states, and the quality of that control. The hierarchical model suggests a nested stack of types of controlled variables that people seem to be able to control when all is well – but the particular examples of these types that an individual controls can be discovered only by studying that individual.

Control theory doesn’t use categories such as “autism” to explain behavior. To say that a person is autistic is only to say that certain externally-visible patterns of action have struck people as similar enough (and unusual enough) to be lumped into a “disease entity.” This does not mean that the same defect exists in all autistic people, or that the symptoms arose from some common history, or that the same treatment will succeed with (and not harm) everyone included in this category. The conventional empirical approach to treating problems as “diseases” is simply to try something on people in a given category and see if it helps a statistically-significant number of them. There is no attempt to analyze what has actually gone wrong – what the person can still do normally, and what the person can’t do. There is no attempt to relate deficits to a model of internal functioning. I suppose the idea is that if you accumulate enough experience with treating people in arbitrary categories, you will eventually be able to look up the symptoms in a big book and read off the treatment that has been effective most often in the past. In my view, this approach is an ill-advised attempt to bypass understanding of the human system and find solutions by relying on guesswork and luck. Before the advent of science it was all we had. Sometimes it works. But there has to be a better way.

If behavior modification helped your daughter, this may give some hints about what is wrong and what is right with her. From the standpoint of control theory, behavior modification techniques work by giving a person control over some variable for the state of which that person has a preference. In behavior-mod, the environment (including the modifiers) is arranged so that this variable can be affected by the person only through performance (or termination) of some particular act. If the person is capable of reorganizing at the level required to do or refrain from that sort of action, eventually doing or not doing the act will become the person’s means of controlling the variable. Of course, behavior modifiers think of the variable as a “reinforcer” and believe that it is causing the changes in performance, whereas the control theorist sees the performance as the means by which the person controls the “reinforcer” or controlled variable.

If your daughter, during these treatments, modified her own actions in order to gain control over some reinforcer, this tells us several things of importance. First, it tells us that she can perceive the reinforcer (although what she perceives about it may not have been what the behavior modifiers had in mind). Second, it tells us that she had a preference, a reference level, for the state of the reinforcing thing, situation, or whatever that she perceived. Third, it tells us that she is capable of reorganizing at least to the extent of changing her means of controlling that reinforcer, that perception. And fourth, it tells us that the change in action entailed in achieving control did not result in any serious conflicts with other inner preferences (if it had, she would have continued reorganizing and would not have ended up with the same actions).

Of course, in behavior modification, the reinforcer is not thought of as a controlled variable. As a result, the kind of control learned must be pretty sketchy. Generally, people whose behavior is being modified are given the reinforcer only when they do something closer to what the modifier wants; their own (other) goals are not considered. The person does not get as much reinforcer as desired, and is certainly not taught how to decrease the amount of reinforcer. So the control that is learned is pretty much one-way: one learns how to increase the amount of reinforcer when there is too little of it, relative to the internal reference level for it. Almost nothing is learned about how to control the “reinforcer” both upward and downward as a means of controlling other things.

Behavior modifiers will usually focus on some action that a person is doing that is annoying or dangerous to others, or harmful to the person herself. The output is to be changed. But under control theory, the output is far less important than the consequence it is being used to create and maintain. If we see an autistic child (I have seen some but know little other than that) sits silently rocking back and forth for hours, there are two ways we could interpret what we see. As behaviorists, we could say that this rocking-behavior is what we want to change; we would then find something – candy, praise, pretty lights --that the child is known to like and withhold it until the rocking lessens, then give it as a reinforcer for lessening the rocking. This would “shape” the rocking out of existence. However, if the rocking were an essential means of controlling something else, it would eventually reappear, or some other behavior aimed at the same result would appear --not necessarily one that would be approved.

As control theorists, we would try to guess what perception is being maintained by the rocking actions. We would try supplying that perception ourselves, as a disturbance, until we found something that would cause the rocking to stop. At that point we would be providing what the rocking was providing, so the child’s rocking efforts are no longer necessary to maintain the perceptions. The most simple-minded possibility is that the actions are providing the physical sensations of rhythmic rocking. So we would hold the child and do the rocking ourselves. If the child’s muscles relax, we would know that this is what she was controlling for. The technical definition of a reference level can be stated as “that level of input at which output becomes zero.” When you have what you want you stop trying to get it.

The question then becomes “how much rocking is enough?” For starters, if I were serious about finding out, I would be willing to guess anything up to 24 hours or until I dropped. Then I would start again, and go on until the child decided that doing something else was more important.

And what if the behavior is head-banging? Do we get a hammer and do it for the child? Obviously not – but we might try some bumping with the rubbery side of a hand. We mustn’t think that we’re so smart that we can guess what another person is controlling for without some kind of test of the guess.

I think that the aim of any therapy based on control theory would be to give a person more and more control of things that matter to him or her. With an autistic child, it may be hard to figure out what the child wants and is trying to get. But when you start thinking as a control theorist, it may not prove as hard as it seems: you just have to notice the obvious. I’ve heard (with some astonishment) a parent say of an autistic child doing something bizarre, “Oh, he’s just trying to get attention.” What is wrong with giving the child attention? If doing so causes the bizarre behavior to cease, you know that this is precisely what the child wanted. The next thing is to see what the child does with the attention once he has it. I think you would find out something about the child’s goals, and about what is preventing their attainment. When you understand what a person wants, you understand what that person is doing. And the more you understand of that other person, the more like you that person seems and the better you will be able to help.

I’m not trying to suggest how your daughter should be helped – only to show how a person acquainted with control theory would make a start. Control theory can only help you understand what is going on; it can’t predict what you will find. There may be problems with which you can only cope, not cure. But I should think that if you were to learn the principles of control theory and use them in interacting with your daughter, you would learn more than you know now and might well think of things to do that would not have occurred to you otherwise. At the least you would begin to think of your daughter (I’m sure you do already) as a person trying with great difficulty to accomplish things for herself instead of as an organism shaped by what is done to it – the behavior-mod view. Despite its name and the awful engineering jargon that goes with it, control theory leads us to respect and nurture the individual will, to see others as sharing our own human condition, and to see the attempt to control others as the ultimate insult to the integrity of a living system.

I hope that control theory has something in it that will ease your plight.

Best regards, Bill Powers

Reply from Eileen

Dear Bill

Your message is probably the most refreshing comment on autism (in quotations) and many other aspects of learning, freedom, … that I have read in a long time. I plan to download it and print it out when I get to my office (and printer) tomorrow.

I cannot respond fully until I have a chance to really look at it, (I’ve only read it twice, quickly), but I need to (1) thank you for your time and concern: (2) let you know that I in now way subscribe to Bmod theory, but that I have found that to a limited extent it works in getting Katy to start to focus on learning and then, as I think you were saying, the task itself draws her in and becomes the goal. Also, the reinforcer they use is in fact far less important that setting up the expectation that she can succeed because the task is being given to her.

There is much more that I want to say, but I’m going to control myself now and sign off until I’ve had a chance to digest and think about your message.

Best, Eileen

Reply from David Goldstein, a Clinical Psychologist

Date: Oct 08, 1991

Subject: PCT and autism

Author: David Goldstein

Date: 10/07/91

As a clinical psychologist, I have worked with some autistic people in special education settings, developmental centers and one child in my private practice. As a parent, I have the experience of a child with special needs.

However, I do not present myself as an expert on autism. Based on my experience with autistic people and interest in PCT, I wanted to write an answer to your post.

In my opinion, Behavior Modification is the most useful psychological approach that I know about at the present time for helping autistic people to learn. At some point in the future, PCT may make a contribution to this area. At the present, I would not look to PCT as a source of new assessment or treatment. It may, however, provide a new set of ideas which you could apply when observing your daughter. A new assessment or treatment method could follow from understanding your daughter better. I would like to refer you to the following book for helping autistic people develop language:

The Autistic Child: Language development through behavior modification by

O.Ivar Lovass. New York: Halsted Press, 1977.

There may be an updated version of this book by now. This book contains specific instructions for specific language objectives.

I am sure that at your daughter’s school they must employ the services of a speech and language pathologist. These are the professionals who have the most practical knowledge about teaching language to people who have not acquired it spontaneously or who have lost it through stroke, accident, etc… As in other professions, not all speech and language pathologists are equal. You have to compare what you want from them with what kinds of experience they have actually had.

Also, the use of American Sign Language has been found to be helpful in some cases. I think that you mentioned you are aware of this. One reason is that it is possible for another person to help produce the sign while this cannot be done with the spoken word. This does require you to learn the signs as your daughter learns them. Some parents have a negative attitude towards it. I think that ASL can help the autistic person learn some general things about communication. However, your daughter may be well beyond this point.

I second what Bill Powers said about applying Perceptual Control Theory to autism. If you start with some easily identifiable actions and ask the question of what experience this action is controlling, you may learn something new about your daughter. Don’t forget that some of the experiences may refer to body states, such as fear, as well as states of the environment. Make a guess about what experience is being controlled. Do or say something which should alter the experience which you guess is being controlled. Observe whether your daughter’s action restores the experience which you disturbed to its original state. In PCT, this is procedure is called the Test for the controlled variable.

Here is a slightly different approach. If you find out what kinds of experience your daughter can and cannot control you will be in a better position to understand and possibly help her. The levels of perception are a good starting point. Using the levels of perception, ask yourself to think of examples where she demonstrates the ability to control at each level. If when thinking about a level, you cannot come up with any examples, then perhaps she cannot control at that level at this time. This would be a starting point for focusing reorganization efforts. It takes a while to grasp the different perceptual levels. The people on CSGnet would, I am sure, be glad to help you learn the levels.

Reply from Eileen

Thank you for your message, which I will be downloading, printing and fully responding to soon. Just to say that your assumption that the school (which is not a bad one) employs experts in language is unfortunately not really correct. Yes, they employ a Ph.D. in communication who truly believes that she knows all there is to know about helping autistic people to communicate (note: not to use language of any kind) and, most unfortunately, that she knows what their limitations are. Basically, her approach is to say that if something has not been documented to work with a child at Katy’s “level of functioning” then she will not try or encourage its being tried either. This is, as I have said, very unfortunate. To be continued, and thank you again.


Reply from Richard Marken

David Goldstein writes:

In my opinion, Behavior Modification is the most useful psychological approach that I know about at the present time for helping autistic people to learn. At some point in the future, PCT may make a contribution to this area. At the present, I would not look to PCT as a source of new assessment or treatment.

I hope the behavior mod types are as generous to us in the future.

I would like to refer you to the following book for helping autistic people develop language:

The Autistic Child: Language development through behavior modification by

O.Ivar Lovass. New York: Halsted Press, 1977.

Boy, this is like “ol’ profs week” on CSGNet. I took child development from O. Ivar when I was an undergrad at UCLA. I was very impressed. Finally, a psychologist who could really get something done. And with children yet: children who really needed help.

Of course, I know better now. Actually, I knew something was wrong even back then, but I had to learn control theory before I could articulate it.

Behavior mod is a great example of control in action. The behavior modifier (bm) has a reference for the kind of behavior he/she wants to see. There is a discrepancy between this reference and the actual behavior (the bm wants to see speech but what he/she perceives is silence). So the bm does things which should produce speech – these are the tools of the behavior modification process. One tool is withholding something that the child wants (finding out what this is is usually the toughest part of the process). Then the bm waits for “approximations” to speech and gives the reward only when the desired (reference) approximation accurse. It’s alot like steering a boat into a slip using only the motor. The boat just drafts around, it has no rudder. The direction of the boat depends on local currents and the amount of puch given by the engine. You (the bm) can control the boat by giving it power (reward) when it is pointing in the “right” direction. You withhold the power (reward) when it starts to point away.

There can be no simple rules for doing either task (training the kid or docking the boat) because the effect of your actions depends on other factors that are operating at the same time (disturbances). For example, gunning the boat when you are oriented 20 degrees left of slip may push you even further to the left if the current is also pushing the boat left. The same acceleration might move the boat right into the slip if the current is pushing the other way. Same with the kid. Rewarding a particular sound (like “hel”) may lead to a repetition of “hel” that sounds more like “hello” (the reference word) or to no repetition or to a new sound completely. So there can be no rule in Lovass’ book that says “In order to get the kid to say “hello” always reward him/her after saying “hel” unless he she has previously made a closer approximation to hello”. Control can’t work this way. That is, you cannot generate a particular response to a particular stimulus. Instead, you must continuously adjust your actions (rewarding - withholding or turning the engine on and off) to keep your perception (of what the kid is saying or where the boat is going) approaching your ultimate reference for what you want to perceive.

The problem with the behavior mod approach to dealing with organisms (ALL ORGANISMS,including those that have been catagorized as “autisic”) is that it ignores the fact that the organism being controlled is ALSO a controller.

Behavior modeification is a recipe for almost certain conflict between control systems ( the bm and the controllee). “Almost certain” because most bm’s realize that there is a problem when they start to get into conflict with the controllee. I think the bm is usually willing (if they are even close to being human) to read just their references for what they want to see the controllee doing. Even Lovass said ultimately that his success depended on having the following attitude towards his autistic children … “load 'em up wit luv” (he has a charming norwegian accent).

So I am reluctant to recommend behavior modification as the best current approach to dealing with so called autistic children. The reason is that some people might actually blindly follow its precepts – which could be quite ugly for both parent and child.

Best Regards Rick

Reply from Eileen

“So I am reluctant to recommend behavior mdoification as the best current approach to dealing wtih so called atutisitic children. The reason is that some people might actually blindly follow its precepts – which could be quite ugly for both parent and child.”

I couldn’t agree more. The problem (at least the main one I’ve had with BMod practitioners ) is with blindly following and not thinking or considering alternative approaches. And, although accountability is probably higher than with other approaches (others that I know of), there is still the ultimate fudge, as with their failure to deal adequately with my daughter’s water and bathroom recurring phobia - - the fudge is that it must be something in the environment (the antecedent) that has set the phobia in motion, but unfortunately we don’t know what it is because it must be very subtle…

Eileen Prince