Marc Gold: Feedback General Issues

Feedback: General Issues

Feedback means letting the learner know what is wanted of him and if he is achieving it. The term is a little confusing because usually you think of feedback as something that comes afterward; something that's fed back.

In this system, the term refers to anything that provides the learner with information about what he is doing. This information can come before; during, or after a behavior occurs. I use the term in a much more general way for instance than the information theorists who use it to mean perceptual information which comes from a person's own activity.

The information theorists talk about knowledge of results; KR they call it, in much the same way that I use the term feedback. Within knowledge of results they make a distinction between information and incentive.

The terms that I use here are feedback and reinforcement. Let me put reinforcement over here you'll see why in a minute okay? 

There's a distinction in this system between those two. The terms aren't really important it's the concept that counts. In this system, reinforcement is one subcategory of feedback. Reinforcement provides a person with information about what he has done, but in addition it also serves as motivation. That may be positive, it may be negative, but it's simply more than providing information. 

Information theory perspective, by the way, provides a very rich view of issues related to feedback. In classifying kinds of feedback, or knowledge of results as they call it, they make a distinction for instance between Intrinsic Feedback which would be inside the task, and Artificial Feedback which is something that comes from outside, from a trainer or somewhere else.

Intrinsic Feedback is found directly in the activity, or the task. Artificial is something that the trainer adds in etcetera, etcetera. In knowledge of results you can have concurrent or terminal feedback. It can go while a person is acting like a speedometer, or it can happen at the end of the act like throwing a dart and then seeing after you've done it what happened to what you did.

Knowledge of results can also be immediate or delayed. You may find out as soon as you're finished with some act or some movement how you did or you may have to wait and find out sometime later. It can be verbal or non verbal. A fairly important distinction. 

The information can come as physical assistance as pointers or buzzers, which would be non verbal, or it can come as scores that would be letters or numbers or statements which would be verbal. We'll talk more about issues related to language in a minute.

It can also be separate or accumulated.  Separate or accumulated isn't quite as important in this system and with the people we're working with as some of these others would be.

The distinction here though would be you can find out how you do on each part of the task as you go along or you can find out at the end how you did. In a football game the players really know what the score is.

It takes a long time to make a touchdown and what's going and things because there's a lot of time between each score. In a basketball game, I guess, not having ever played basketball, that the points accumulate so fast sometimes that the players really don't know where they stand until there's a moment where things aren't running back and forth where they can look up and see. That would be accumulated knowledge of results.

This is a class of deaf/blind children. It's also an environment loaded with communication. One of the distinctions I just made is very important to this technology, and that's a distinction between verbal and non verbal feedback.
In communicating with people; we as a society have understandably relied very heavily on coding systems. Sign language, finger spelling, various languages are all coding systems.

My definition of verbal feedback is any conveying of information that requires going through a second signal system or a coding system of some kind, where a person in order to understand something has to convert the information coming in, through an established code of some kind before obtaining meaning.

Non verbal feedback would be those situations which require either very little or no processing of symbolic information.  Moving a person's hands through a task doesn't involve converting symbols into meaning.  The meaning is direct. Notice both kinds of communication going on here. 

Let's look at a little different training circumstance. This child is at a point where he is not working so cooperatively with a trainer. A symbol system is being used here but it's in the process of being developed.

Wherever the trainer can, she is providing the opportunity for symbols to be learned. In other cases she's clearly communicating with him in a way that doesn't require him to process any symbols.

There's an interesting side issue that comes up in this particular training situation. I think a superficial inspection would lead one to believe that the teacher is being fairly coercive or strong with the learner.

If you look a little more closely you'll find that she's being very careful to use as little as possible her superior strength over his inferior strength. In the couple of instances where he attempts to move away, and moves around he doesn't do what he's capable of doing which is really pushing away and getting out of there. You don't see that. That's an important point, and is a side issue of our discussion of language here. But never the less something I think needs to be stated. 

The teacher is being very consistent, working as gently as possible with a young man that's still coming around. But in the final analysis you don't see the kinds of things that you see in some places where the trainer is getting angry, or where she's clearly coercing him to do what she wants. The child is pretty much at a state of rest, and cooperative.

There's an old argument in special education. In fact there's a lot of old arguments in special education. Do we teach to a person's weaknesses or do we teach to a person's strengths? You have to do both. 

One of the areas that usually a weakness in people who find it very difficult to learn is language processing. Getting it in; understanding it; getting it out. Being able to take in sounds or symbols and convert them. 

I think we have two responsibilities. We have to do what we can to teach individuals to use coding systems, and we also have to teach people to do things. How to get along; how to do all the things that we do to make it life. 

One of the mistakes that we've made as a profession, and as a society is to try to teach people things using strategies that rely on those people having to work through coding systems.

We've taught them everything through language. If a learner didn't have the language then there was no access to the things. With this strategy it's no wonder that so many people have learned so little. An alternative strategy is to say if you want people to learn language then have lessons on language.

If you want people to learn things, then find ways of teaching those things that don't require the learner to know or use language. If you want language attached to things, then teach what you want the person to know without them having to process language then attach language to it then use the language to bring on the already acquired behavior.

Let's say that you want a person to learn how to sew a button on a shirt. You might teach him without him having to know needle, thread, button, button hole, and then when criterion has been reached; when he's doing the behavior then as he does it start attaching labels to all of those things or to some of the behaviors.

To the things that he already knows and after doing that a number of different times, see what happens when you say needle or when you say, "Pick up the button".

For a person who finds it difficult to learn; learning the behaviors first and then learning the language attached to them after they already exist has an advantage over a person or a situation where a person has to learn not only both of them simultaneously, but in fact has to get the language first in order to even know what it is he's supposed to do.

I'd hasten to point out by the way that some kinds of learning requires the use of language from the start. A lot of things, things that are verbal in and of themselves are things that my comments really don't apply to here.

This is the National Children's Center in Washington, D.C. The men and women here clearly show the capabilities of individuals with severely handicapping conditions to develop sophisticated quality skills in the presence of powerful training procedures, and then to utilize them under minimal normal industrial supervision. In this case to assemble electronic printed circuit boards.

This workshop is an outstanding example of where technology and trainer experience and feelings come together. Learners acquire new skills; competent skills. Trainers develop with practice the ability to integrate the judgment, and their feelings with the bag of tricks. When there is a good interaction between the technology available, and the experience and judgment of the trainer it's obvious what can happen.

"I've been working in the field a couple of years now, and I think one of the most important things I've found is that you really learn by doing. You learn by practicing by trying out different ways. 

As we've worked with the workers here we've found that as trainers we've become much better over time working with them. I think one of the main things we develop is some subtle techniques for knowing when is the right time and for prompting them, for cuing them, for giving them some feedback.

Timing is really important. I also think that one of things you develop is a judgment on how to prompt the person. What kind of feedback? It's very important not to give too much feedback so that you're doing it for the person, or to give too little feedback so that he really doesn't know what to do.

What it really comes down to, to being a good trainer is when you practice those things, and work out your own style and how you're going to work with the trainees."

That's a very important thing she just said. You work out your own style; integrate that with the instructional technology, and look what happens. 

Whether you're teaching someone to brush her teeth or solder electronic components to a circuit board the combination of trainer judgment, and instructional technology gets it.

I talk about the perspective of the information theorists. There's another branch of psychology called behaviorism that has some other ways of looking at some of the distinctions that we have been making.

According to individuals that we might label behaviorists learning occurs in a chain of events. There's a stimulus then there's a response then something that might be called reinforcement including positive and negative things that might happen, or including nothing. That would be punishment, positive reinforcement, negative reinforcement, etcetera or extinction.

Some behaviorists make a distinction between stimulus control and reinforcement control. Stimulus control refers to a circumstance where the major focus of attention is placed on organizing the environment to provide information. Reinforcement control refers to focusing attention on contingencies on cause and effect relationships on what happens after a person does something.

We could get tied up in the semantics of all of this but I think that as a generality the distinction between reinforcement control and stimulus control has some value. When might we focus in on the stimulus control aspects of the situation to provide power for training, and when might we focus our attention on controlling what people learn by payoff? By what we do afterward?

Let me talk about five different situations.

Number one: Increasing the rate of an existing behavior. The person knows how to brush his teeth but he only does it an average of once a week and you wish he did it twice a day. Or; different category, a person produces in a workshop 100 units an hour, and you'd like him to produce 400 units an hour. In one case you're talking about increasing the percentage of time he does something and the other case you're talking about increasing the absolute number that he does in a given period of time.

Number two: Maintaining existing rates of behavior.  He brushes his teeth twice a day and you're happy that he does you just want him to keep doing it. Or the person produces what you want him to produce you want him keep on going like that; that's not so easy to say.

The third category is decreasing the rate of an existing behavior. Usually it's a zero by the way. He's got it all right, what you want is that he didn't have it. Okay?

The fourth category is kind of an interesting one, knowing when to and when not to. Singing is wonderful you wouldn't want to say stop singing. Okay in some places but you don't want him out in a workshop or sitting on a bus singing at the top of his lungs.

In these four situations: Increasing, maintaining, decreasing or knowing when or when not to do something, I think the reinforcement control part of the bag of tricks is likely to contain what's needed to solve the problem.

Now that's a generality there are obviously exceptions to that suggestion but that's the first place that I'd look to solve those four problems or accomplish those things.

The fifth category is teaching a new behavior. That's different. You want the person to learn something that he doesn't know. In this situation, I think it's in the stimulus control part of the bag that you should look first: How to organize the information for him to reduce his alternatives so he can zero in what it is he has to learn. How to provide him with feedback. Not in a sense of changing motivational levels, but in a sense of giving him information.

For teaching new tasks that's where I'd look first. With all these generalities, I might come back later and try something in one of the other sections of the bag. But that's where I'd start first.

Now there are others than just these five. You know you can have transfer of training, problem solving, retention, there are more things to learn. But for these five things that's a distinction that I think is important and is very useful.

Modeling is a strategy that should also be brought up here in our discussion related to General Feedback. Modeling, or imitation as some people would call it, of someone else's behavior provides information that doesn't have very much power. That is, it requires the learner, the person who is doing the modeling, to utilize a great deal of observational ability. Especially if the activity being done by the model is a complicated one.

If you're watching somebody operate a machine, a complicated machine, just watching them isn't going to give you very much. You can't zero in what's going on. But if it's something that's a fairly simple behavior you can watch them and get a lot out of it.

For people who find it very difficult to learn; it's the same thing. People such as Dr. Lou Brown at the University of Wisconsin have used modeling and other procedures very effectively to teach a wide variety of skills to individuals with moderate and in some cases severe mental retardation.

And so this procedure should be looked at very closely for teaching a variety of skills to a variety of people. As the complexity of the information to be learned increases however, it becomes less likely that this technique will provide what's needed in the way of power.

Decisions about general feedback can be made at the same time as initial decisions about method, content and format when you're doing the planning. All of these decisions are made outside of the actual learning moment. This is all part of the planning. When an interaction has occurred between the learner or learners, and the trainer and someone hasn't learned, we come back to planning in stages 5, 6, and 7 where revisions are made.

What we haven't covered yet are those moment by moment decisions that you can't plan where you write up a task analysis and that you have to make as the training moment goes on. These are the specific issues related to feedback and provide the basis for the next film.