Classic Computer Magazine Archive CREATIVE COMPUTING VOL. 9, NO. 4 / APRIL 1983 / PAGE 180

Learning to learn by learning to play. Marc D. Stowbridge; Peter Kugel.

Learning To Learn By Learning To Play

Computer games have been used for many purposes. They have been used to entertain, to teach students traditional school subjects like arithmetic and spelling, and to help students become comfortable with computers. In a course recently given by one of us (Marc) at Boston College, computer games were used to teach students something they are seldom taught in courses: how to learn. We thought that by teaching students how to learn instead of teaching them what to learn, we might be able to help them get more out of school.

We hoped that by asking them to think about how they were learning (or failing to learn) in the rather simple rulegoverned environment that games provide, we might help them to discover better strategies for learning. In other words, we hoped to help them to learn how to learn.

But, since we were fairly sure that they would think that they had only learned to play games, we then asked them to apply what they had learned by playing to the job of learning something more "serious'--namely computer programming. And we hoped that they would then realize that the ideas that they had gotten from learning to play could be used in their other courses too. We know that some of them got the point.

The course has been given several times to several different types of students. Last summer, for example, it was given to 40 "high-risk' freshmen who had been admitted to Boston College in spite of relatively weak academic records. Although most of these students had not learned what most students have learned before they go to college, they appeared to have the underlying ability to learn those things. During the summer, other efforts were being made to teach them the things that they had missed in traditional "remedial' classes. But we felt that it was at least possible that some of these students simply did not know how to learn in a classroom situation. We felt that if they could learn to learn they might be better able to learn, not only the things that they had missed, but also the new things that would be presented in their college courses.

Computer games have some advantages over traditional games for teaching purposes. One is that they can be played in such a way that each player has complete control over the game. The computer always follows the same program and there are no other people involved who can "take over' the work. Another merit of computer games is that the computer is very strict (but very impersonal and non-threatening) in following the rules. And finally, the computer will play tirelessly, thus freeing the teacher from having to run (or even to play) the games.

Adapting the course to other situations should be easy. The underlying ideas seem to us to be adaptable to other student levels, other computers, other schools, and other teachers whose ideas about what constitutes good learning procedures may be different from ours. Much of the students' work was with computers which allowed quick correction of student errors without requiring a large amount of grading on the part of the teacher. Many of the programs required for the course already exist; good games are available for most computers.

The course was based on four fundamental ideas:

1. Students can improve their ability to learn by thinking about what they are doing when they try to learn and by discussing what they do with others.

2. Such thinking and talking about learning is best done while the students are actually trying to learn some particular thing, rather than through abstract discussions or lectures about general principles.

3. The process of learning to play (and win) computer games provides an excellent sample learning process for this purpose for several reasons:


4. What students learn about learning by learning to play can be transferred to the process of learning other things if one pays attention to the transfer process and does not assume that it will happen automatically.

Not Knowing How To Learn

Some students seem to know what to do in school and some do not. One can argue that the differences between those who do well in school and those who do not are innate and hence unchangeable. To some degree, this is almost certainly true but we feel that it may not be quite as true as some people think it is.

Suppose that you believed (as we do) that learning is something that the learner does, that a learner is not just a passive sponge who sits in one place and absorbs information. Successful learning would then depend, at least in part, on doing the right things.

Virtually every human being seems to be born "knowing' how to learn certain things. Everyone learns to recognize his mother, to nurse properly, to walk, and to talk. Such learning is accomplished with little visible effort on the part of the learner. We seem to be born with built-in "programs' that we follow to learn these things.

The ability to learn in school, however, seems rather less evenly distributed among us. Some children enter school seeming to know what to do. Others don't.

The ability to learn in school might be innate, but suppose it is not. Suppose it must be learned. Imagine, now, that you are arriving on your first day of school without having learned it. If you are one of those people to whom school learning came naturally (and chances are that you are) this may be hard for you to do. But try. Here you are. The teacher does something. You do something in response but it is the wrong thing. (You haven't learned the right thing to do, remember?)

You start out enthusiastic and work hard. Your teacher notices that you still aren't learning. Perhaps you need to be taught more slowly. That doesn't work. Your teacher concludes you lack ability. You get discouraged. Nothing you do works, so you decide you don't like school. That continues for years without anybody realizing that all that is "wrong' with you is that you don't know how to learn in school.

Why doesn't somebody teach you? One reason might be that nobody knows what the problem is. But there are other reasons. One is that many people don't believe that learning can be taught. Another is that even if you believe that learning can be taught, it is not obvious how to teach it. After all, you want to teach it to people who don't know how to learn reading, writing and arithmetic. So how are they going to learn learning?

Here's an idea. Suppose that you try to let people learn to learn the same way they learned to speak their native language. You simply plunk them into a learning situation simple enough for them to handle and let them learn. For this to work, you need to pick something that people have an innate ability to learn. How about playing games? Children (and adults) seem to have a built-in ability to learn to play games.

You could, of course, teach games the way you teach languages in school, and students would probably have the same difficulties. A wag once suggested that if you really wanted to kill baseball in America, all that you would have to do would be to teach baseball in school. But that is not what we did with games in our course. We gave students a few hints and some written instructions and let them figure out how to play on their own.

Learning To Play

In each session, students were told a bit about a game, and sent off to play. They were asked to:

While they were at the computer, they were asked to take notes about what was happening and to keep the printout. Then, when they came back to the next meeting, the class discussed what did and did not work.

Some of the game programs we used were taken from David Ahl's book Basic Computer Games, and some were written especially for this course. The games chosen presented the students with problems of increasing difficulty so that their learning would be cumulative.

Students were guided through the course by a control program that performed various functions:


The Curriculum

On the first day of class, students were given instructions for logging in (we used a time-shared PDP 11/70) and for running the control program called LTL (for Learning to Learn)1

1 The course was run under the auspices of the Learning to Learn program at Boston College, directed by Marcia Heiman and Dan Woods.

The students took this material to the computer and tried (on the whole, successfully) to log on and off without help. If they ran into trouble and asked for help, it was given--sparingly.

They were asked to keep their printout and to keep detailed diaries of what happened. A typical entry in a student diary looked like this:

"The computer typed WHAT IS YOUR L.T.L. NUMBER? I looked at my instruction sheet and typed 501. I waited for a while.

The computer did not do anything. Then I remembered to type RETURN.'

We asked the students to record not only their successes but also their mistakes. Most computer programmers recognize the value of making mistakes and learning from them. But surprisingly few students do. A great deal of elementary education seems aimed at disguising the value of making mistakes and correcting them. We emphasized it.

We felt that, if the students wrote down what they were doing and talked about it, they could not avoid thinking about it. Thinking about learning (like thinking about anything) gives you the opportunity to change the way you do it. That, supposedly, is one of the reasons for teaching philosophy and also the basis for many kinds of psychotherapy. Our feeling was that at least some of the students might have gotten trapped into unproductive learning behavior. Thinking about what they were doing gave them the chance to at least consider changing it when it did not work.

We also hoped to get them to "think about their own thinking.' Our hope was that if they could think about their own thinking, it might occur to them to change it to adapt it to circumstances. A common problem in students is that, once they hit upon a successful way of working, they use it for everything. (As Abe Maslow once said "To the person who has only a hammer, the whole world looks like a nail.') We hoped to make them more flexible.

Guessing Numbers

At the end of the first class--in which we discussed both how they actually did log on and how they learned to log on--they were told to go back to the computer and play their first real game, NGuess, a simple and familiar game in which the computer picks a number between 1 and 100 and the player tries to guess it in 7 guesses or less.

The students were told to learn to play correctly, which is relatively easy but still raised problems for some. And they were told to find a good strategy.

Some students did not even try to play rationally. (They guessed the same number twice.) Others looked for, and found, an optimal strategy.

The program they used was more failsafe than most. Errors (such as O for 0) were trapped, and the student was given a chance to recover. Error comments were as clear as possible. Care here seems important if for no other reason than to raise student ambitions by making it possible for them to solve the problems.

The second class meeting, during which they discussed their play of NGuess, set the pattern for the rest of the course. As problems were raised, they were written on the blackboard and an attempt was made to categorize the problems so that they could be dealt with individually. Problems that arose fell into four categories:


No attempt was made to try to sell one approach to problems over another. It is our feeling that some students resist learning in school because they feel it violates their personal integrity to do what the teacher tells them to do. They feel they are giving in. Such students do better when allowed to use their own strategies. Furthermore, students understand better things they have framed in terms of their own intuitions than what has been framed in the intuitions of others.

One of the great merits of using computer games in this situation is that the computer will reward any approach that works. It need not be the approach that the programmer of the teacher had in mind when the game was presented. And this gives the student a feeling of confidence in himself.

Most students who found the optimal strategy for NGuess did so by thinking of the numbers arranged on a line and then thinking of their guesses as cutting the search area in half each time. This was interpreted to the class as an example of the value of trying to think of problems in terms of spatial images, of trying to look at problems in different ways while looking for solutions, and of using metaphors to better conceptualize (and remember) problems.

Many of the students in this course seemed to lack flexibility in their approach to problems. They would try one way to solve the problem and it would never occur to them to consider a different one. They would consider only two possibilities--either they had not worked hard enough on their one approach or it would not work and thus there was no possible solution to the problem.

The Other Games

After NGuess came a game called Flipflop. In this game, the player tries to turn a row of X's into O's by indicating which symbol in (say) a five-symbol row, is to be "flipped' (from O to X or X to O). The difficulty comes from the fact that, when a particular symbol is flipped, others may be flipped along with it. For example, if the student has reached OOXOO and tells the computer to flip the third symbol, the computer will flip the third symbol, but it might also flip the fifth as well, producing OOOOX. In this round of the game, the computer will always flip the fifth when told to flip the third. But it may do something quite different when asked to flip the fifth and it may do something else in the next round of the game.

There are several different strategies that can be used to win this game, but they all require a systematic recording of observations and testing (and extrapolating) alternative strategies. Other games used in the course (in order of appearance after Flipflop) were:

Blackjack, the familiar card game, also known as 21.

Word, in which players try to guess a five-letter word selected by the computer by guessing its letters one at a time. The computer tells them whether the letter guessed appears in the word and, if it appears, where it appears.

Mastermind, in which players try to break a code.

Star Trek, a popular computer game with relatively complex instructions and strategies.

Each game presented its own problems. Blackjack had instructions that were difficult to figure out--ordinarily an undesirable feature in a computer game--that gave the students a chance to figure out what to do when they could not understand what they read.

Mastermind suggested the idea that theories might be things that one could test by varying parameters one at a time so that, when something turned out not to work out, one knew what it was that was not working.

Star Trek was the most complex game of the group. Its instructions alone require four pages to print out. Good strategies for playing are rather complex, and this is a good game to teach the concept of "divide and conquer'--breaking problems into parts before trying to solve them. It was also a good place to point out that it does not always pay to discard a strategy simply because it does not work the first time. Before totally discarding an approach it sometimes pays to try to first improve (or debug) the approach to see if it can be saved.

A General Recipe

Toward the end of the game playing part of the course, an attempt was made to formulate a general strategy for solving learning problems as they arose. The hope was that, by providing a recipe to follow, one might be able to help the student's transfer strategies, learned from learning to play, to new areas. The purpose of a recipe (or a paradigm) like this one is to help the students' transfer strategies, learned from learning to play, to new areas. The purpose of a recipe (or a paradigm) like this one is to help a student figure out what to do next in new situations. It was suggested that thinking about what you do was appropriate whenever you faced a situation in which you wondered: "What do I do now?' We decided that you might do the following 3 things:


Most of this recipe is simple common sense--except that its use is really neither simple nor common.

We feel that a particularly important feature of this recipe is that it always starts in the same situation: you are wondering what to do next. The time to think about your thinking is always indicated by the same feeling--the feeling that you don't know what to do next. This can happen in a variety of circumstances, but it always feels the same, and hence, is easy to recognize.

When you do recognize it, you have two things to fall back on. One is your recipe and the other is your memories of specific things that worked for you (they may not be right for others) whey you learned to play computer games. We feel that in a course like this, the idea of a "triggering' feeling that tells you when you can use the ideas learned, is very important if you want the ideas to transfer to other courses.

Learning To Do "Useful' Things

Our aim in having students play games was not to make Pac-Man experts. We were trying to make students better at learning things other than games. We wanted them to take what they learned from leaning to play and apply it to their regular courses. The process by which one takes something learned in one area or course and uses it in another, different, area or course, is called "transfer.'

It used to be widely felt that learning something like Latin or logic was good for students because it strengthened their minds and would help them in studying other subjects. But the mind is not exactly like the muscle that this analogy suggests. Learning to lift dumbells may help you learn to lift packing crates but learning to play games need not help you learn to solve calculus problems.

If you want transfer to occur, it helps to practice the actual process of transferring what you have learned by playing to something else. For this purpose, we spent the latter part of the course teaching students how to program computers. Programming is similar enought to game playing (and they both use the computer) that the transfer is fairly natural.

We have noticed that students today are relatively poor at using what they have learned in one class in another. One reason may be that they do not have much practice in such transfer. And one reason why they may not have much practice it that, because of the fragmentation of the modern curriculum, teachers cannot usually assume that their students share any skill that can be transferred.

This was different when Latin and logic were in their heyday. Everyone took them both, so teachers could regularly allude to them, thus giving their students practice with transfer. But this is no longer possible, which may be why Latin and logic no longer seem to work as well as they used to.

By including a bit of learning to program in this course, we could give the students practice with the transfer process because we could assume some things in the students' backgrounds that they shared (the material of the first part) and the transfer of which could be discussed.

The transfer seemed to work--at least when it was thus guided. The students did seem to learn programming more easily and more imaginatively than one would ordinarily expect from similar students.

We do not know whether the ideas learned by playing games also transferred to their other courses. But we have some fragmentary evidence that suggests that, at least in some cases, it did.

We are aware that it is easy to be fooled into believing in the success of one's own teaching methods. We do not think we were fooled but we cannot be sure.

However, the course was short. The effort was relatively small. The possible rewards are considerable. The course can be tailored to fit into virtually any curriculum. We feel that others might want to try this course with their own students. For such people, it may be helpful to summarize some of the main assumptions of the course as we see them and to list some of the things that someone who is trying to adapt this course to other situations might keep in mind:


Many students--perhaps most--do not do as well in school as they might. They are not as open to learning new things as they might be. This fact is discouraging to many people. We know that we have not discovered a cure for this situation that will work for all people in all situations. But for people who see this as a problem and wonder what they can do next, learning to learn by learning to play offers one possible answer.