Suppose I tell you that William James was a functionalist. That information enters your memory. If that memory is strong enough, you can recall the information at a later time, perhaps for a test. Or you might recognize the correct answer (again, perhaps on a test). Similarly, once my stomach was hurting and I didn't know what to do. I recalled from my Boy Scout days that I shouldn't take an antacid, because I might have appendicitis. I didn't know an antacid from a laxative, but this information suggested that I might have appendicitis. (I went to the hospital, and I did have appendicitis. They took out my appendix before it burst.)

Most people call this type of knowledge memory. (I will sometimes call it conscious knowledge.) Is memory the only type of knowledge? Memory is conscious, so everyone knows about memory and agrees that it exists. The other types of knowledge are not conscious, so they are often unacknowledged. So the common assumption is that memory is the only type of knowledge. This is just an assumption. There is no evidence to support it, and good evidence for the existence of other types of knowledge. I count four types of knowledge (Frick, R. W., & Lee, Y. S.,1995).

Our educational system is essentially built on this assumption that memory is the only type of knowledge. It is perfectly designed for teaching memory. As we shall see, it is poorly designed for teaching mental models, and mental models are very important. So our educational system fails. But read on.


Memory is the only type of knowledge that is conscious, and memory is very closely intertwined with consciousness. First, conscious awareness is needed to form memories. Suppose a word is spoken to you but you aren't paying attention, so the word doesn't become conscious. That word still has a variety of effects, because it enters your Inferential System. However, it does not enter memory -- you will not be able to recall the word at a later time or even recognize that the word was presented.

Conscious awareness is also needed to use memories -- a memory has no effect on behavior until it is consciously recalled. Once I went to my friend Fred's office to play a game of bridge. Everyone was there but Fred. We stood around for a few minutes, then someone asked where Fred was. I easily recalled Fred saying he would be in New York City. So I had this information in memory, but it did me no good until I consciously recalled it -- I trudged up the stairs to Fred's office and stood around aimlessly just like everyone else.

Or suppose your spouse tells you to buy milk on your way home. If you consciously recall this instruction, you can stop at the store and buy milk. If you don't consciously recall this instruction, you will not find yourself unexpectedly stopping at the store. Instead, you will drive right by the store, even though the key information is in your memory and easily recalled once you get home if someone asks or reminds you.

Because of this close association, I will sometimes call memory conscious knowledge.


A second type of knowledge is habits. Habits were discussed in the chapter on automaticity. Psychologists have shown that you can learn habits without conscious awareness, and of course one of the defining features of a habit is that you can use it without conscious attention.. For example, you can tie your shoes without consciously recalling any information. In fact, you have probably forgotten your instructions for tying your shoe and would now need to watch your hands tie your shoes to describe how to tie your shoes. Similarly, I have asked many people which way to turn the wheel of a bike when falling. Everyone guesses wrong, even though they would turn in the correct direction when riding.

Encoding. Suppose someone tells you how to tie shoes, and those instructions become lodged in your memory. You can then recall that information and tie your shoe. Suppose also that you also practice tying your shoe, forming a habit. That habit now exists in addition to your memory. Your habit can't contain exactly the same information as your memory, but the two can be close. Suppose for the moment that your memory and your habit contained exactly the same information. Then you could tie your shoes from memory, or you could tie your shoes from habit, and your shoes would end up being tied exactly the same way.

However, you would still have two completely different types of knowledge, lodged in a different places in you brain, and encoded in different ways. If you wanted to think about something else while you tie your shoe, you would need your habit -- using your memory requires conscious recall and conscious application. Conversely, if you want to tell someone else how to tie their shoes, now your memory is useful. If you just have a habit, then you have to do the watch-your-hands-tie-the-shoes trick. The two also differ in how well they are retained -- habits are retained a lot longer than memories.

In education. Habits play an important role in early education because the 3 R's are essentially automatic responses. For example, you want a child to think 7 when given the stimulus of 4+3. The educational system seems to handle this well. Note that you would not give a lecture to teach your student a habit. Instead, you would have your student practice. The point is, two different types of knowledge are taught in two different ways.


A third type of knowledge is prototypes. These were discussed in the chapter on felt senses. Briefly, the knowledge in a prototype is not easily made conscious. For example, my research has shown that people can use information in deciding whether something was a cat or dog even though they could not consciously describe that information. They even left that information out of a picture. For example, they would draw dog and cat eyes as being the same even though they would use differences between the eyes in deciding whether something was a dog or cat. When subjects in an experiment were taught a prototype, by example, they could recognize it better than they could consciously say the rule. (Conversely, when subjects were consciously told rules, so the information was stored in memory, they could report the rules better than they could use them for recognition.)

Encoding. The basic idea is the same as for habits. You could tell students the rules for recognizing birds, putting that information in memory. Or you could just present birds and let the students form a prototype. The information in memory and the information in the prototype won't be exactly the same, but let's suppose they are close. The prototype and the memory would still be two different things. They might be in different parts of the brain, and they almost certainly would be encoded in two different ways. Because they are different types of knowledge, they are learned and used in different ways. The prototype provides effortless recognition, and it could probably be added or subtracted from other prototypes. The memories take effort to use -- they need to be consciously remembered and consciously applied. Conversely, if you wanted to tell people the information, the memories would be perfect. Extracting information from a prototype is difficult, and you almost certainly would not exract all of the useful information.

In education. I once thought that students could be given examples, then they would abstract the general concept from the examples. When I wanted my students to learn what a theory is, I gave them examples. I hoped they would extract the regularities and form the concept of a theory. They didn't.

Examples can be very powerful, in the right circumstances. But people do not form new concepts from just seeing examples. One problems is that students will extract only the regularities that they can see. When I present examples of theories, they notice that theories are long and difficult to understand. They can see that. However, they don't have the concept of a theory, so they don't see the theoriness in the examples. Therefore, they don't extract that as a regularity. Put bluntly, if students can't see blue, you can't teach them the concept of blue by showing them blue things.

Both the power and weakness of examples was illustrated by this classroom exercise I once did: I was trying to teach a concept called "basic level concepts". I explained to my students how to identify basic level concepts. (A basic level concept is easily communicated in charades or drawing, and it is learned early by children.) I also gave examples. Then I tested my students. Half of the students got the correct answer. They used any one of the easy methods I had described. The other students were wrong. They extracted a regularity from my examples, a regularity I had never mentioned and one that in fact was not reliable. Then they ignored everything I had said and instead used the regularity they had noticed. So examples are powerful -- on their own, the students extracted a regularity and used it in preference to everything they had been taught. But the examples were also impotent -- the students extraced a superficial pattern, not any understanding of the concept I wanted to teach.

Think for a moment how little you have learned from examples. You probably have listened to many songs without learning how to compose a song, seen many shirts without learning how to design a shirt, seen many buildings without learning architecture, and so on. Beethoven could listen to a new symphony and learn a lot; you can't. This is not to say that examples are useless. The point is, you cannot magically form a new concept using examples if you cannot see the concept in the examples.

Mental Models

Once I was in a toy store searching for Christmas presents. I decided to buy my nephew some Legos. The store had many sets of Lego look-alikes, including one set that had some phantasmagoric shapes. I felt my nephew would prefer the phantasmagoric Lego set. I even wondered how the other manufacturers would stay in business -- who would buy their product when the more interesting phantasmagoric set was available?

This story is completely mundane. This happens to you a thousand times a day. You have a feeling that someone won't like something, a feeling that something isn't going to work, or a feeling that something will happen. You have a feeling that you should say one thing to someone instead of something else.

But if you try to fit this story into the idea that all learning is memory, it just doesn't fit. Yes, I have had experiences with my nephew. Yes, those experiences led to learning, and the learning led to the feeling that he would like the phantasmagoric set best. But what kind of knowledge was I forming from my experiences? I wasn't using memory to make my judgment. I did not recall past experiences with my nephew. I did not recall things I had learned about my nephew. I did not recall anything. I just felt that he would like the phantasmagoric set better. I could later recall events suggesting he would like the phantasmagoric set, but I didn't do that when I was having my feeling and making my decision. Anyway, I can equally well remember events suggesting he would not like the phantasmogoric set. In fact, if I consciously think about which type of Lego set he will like, using memories of my experiences with him, my conscious conclusion is that I don't know which set he will like. But I still have my feeling that he will like the phantasmogoric set.

There is another type of knowledge. I call this type of knowledge a mental model. It corresponds well to psychogists' understanding of mental model (except of course they assume the mental model is conscious and built using memories). I have a mental model of my nephew, and I used that mental model to predict what my nephew would like.

My story had an interesting sequel. After deciding on the phantasmagoric Legos for my one nephew, I went shopping for another nephew. After some fruitless searching, I decided to buy him a Lego set too. But I felt he would probably prefer the standard set of Legos, so I bought that for him. It's the same old story. I had a mental model of my second nephew, and I used the mental model to decide what he would like. The answer came as a feeling. You do this so often every day that there is nothing remarkable about this story.

But if you try to fit it into the idea that all knowledge is memory, the fit is even worse. I want to know if my second nephew will like the phantasmagoric Legos. Suppose I really did somehow start accessing my memories. The first memory I am going to find is my conclusion, make minutes ago, that all boys will prefer the phantasmagoric Legos. We know from all research on memory that recency is very important. I am not going to have a stronger memory than that. And I cannot have a more relevant memory. Every thing else is kind of fuzzy, but this is straightforward: All boys prefer the phanasmagoric Legos, my nephew is a boy, therefore he will prefer the phantasmagoric Legos. Over and done.

So why didn't I come to that conclusion? The point is, I didn't use memory, I used my mental model. In fact, after I decided to buy the normal set for my second nephew, I recalled my conclusion that everyone would like the phantasmagoric set. Even then, I didn't use my memory. I laughed at my foolish conclusion and trusted my mental model.

The Function of Mental Models: As Models. Let's focus on the idea of a model. An engineer can build a model of an airplane wing, put it in a wind tunnel, and learn what will happen to a larger wing on an airplane. An engineer can build a model of a bridge and use that model to learn the weakest parts of the bridge. These models are used for prediction.

Your mental models work the same way. I put my mental model of my nephew in a room with some phantasmagoric Legos, and my mental model predicts his reaction. Of course, that prediction could be wrong -- nothing in the brain works perfectly, not even perception. The prediction is just useful -- it is better information than nothing, and with enough experience the prediction can be very good.

You have thousands of mental models. You have mental models of the people you know, and you have a mental model of people in general that you use for people you don't know. You have a mental model of your computer that governs how you interact with it, and you have mental models of the other machines in your life. You have a mental model of solids, liquids, air, your digestive system, the post office, and so on. Some of these mental models are simple; most are complex. You use these mental models all of the time, to predict what will happen in different situations, and you use these predictions to guide guide your actions.

Mental models are not useful for reporting information -- it is very difficult to extract the information in a mental model. You can try to say everything you know about someone and see the difficulty. Even experts have this problem -- they can have sophisticated, accurate mental models, but they can't say what they know. Sometimes the conscious report of experts is simply wrong. For example, researchers found that doctors taught their students one method of reading x-rays and then used a different method when they read x-rays.

Encoding. Suppose you had a mental model, and you also had memories with the exact same information as the mental model. The memories and the mental model would be encoded in the brain in different ways. They would have different functionalities -- it would be relatively easy to tell people the information you have in memory, and difficult to tell people the information in your mental model, but it would be easy to use the information in your mental model and relatively difficult to use the information in memory. Research also suggests that the mental model will be retained a lot longer than the memories.

Mental Models Vs. Conscious Knowledge: Evaluation

Is this theory really true? That' a fair question.

I said there were four types of knowledge. I am not strongly attached to the number four. If memories and habits turn out to both be forms of classical conditioning and you want to count them as only one type of knowledge, fine. There are also some similarities between prototypes and mental models. If there turns out to be two or three types of knowledge, fine. I doubt that anyone will discover a fifth type of knowledge, But if they do, fine.

I am strongly attached to idea that there are different types of knowledge. Again, I see the bias of conscious awareness. Anyone who assume all thinking is conscious will probably fall into the trap of thinking that all learning is memory. (This means you, at one time, and it meant me at one time too.) But any thorough examination of human learning reveals that there are different types of knowledge, learned and used in different ways.

The last question is if mental models are one of these types of knowledge. Frankly, it is difficult for to imagine how the brain could encode a mental model. It is relatively easy to imagine that the brain can learn memories, habits, and prototypes. However, no one knows exactly how the brain encodes memories or habits, or the details for how the brain encodes prototypes.. So it is reasonable that mental models are simply a bigger mystery. And if there is no such thing as a mental model, it is difficult to explain why there seems to be this fourth type of knowledge, learned and used in a way different from the other three types.

So if you want to maintain a piece of skepticism about this, fine. I do. If you want to say that the evidence is not convincing, okay. If you want to say that it is crazy to believe in different types of knowledge, you are wrong. It is crazy to firmly believe there is only one type of knowledge. In particular, mental models are do not behave like some disguised collection of memories.

We can treat this like an exciting issue for scientists to argue about, but there is also a very important practical issue -- trying to make sense of our educational system. My four-types-of-knowledge theory came from laboratory studies. But it makes sense of what I see in the classroom, and I have used it to better understand how to teach. If this theory does not help you make sense of what happens in the classroom, if it does not help you teach, then it is of no use to you, right or wrong.

Mental Models Vs. Memory: Summary So Far

I brought up habits and prototypes because there is very good evidence that they are different from memory. I desperately want you to understand why I think there are different types of knowledge. But now they drop out of the story. The story now is memory versus mental models. For the purposes of understanding education, the critical point is that memories and mental models are two different types of knowledge, learned in different ways, and used in different ways. We can now ask, which type of knowledge do we want to teach? And if we want to teach mental models, how are mental models taught?

It is really very easy to distinguish mental models from conscious knowledge. To use conscious knowledge, you conscious remember it. Then you make inferences using it. You don't get feelings -- I didn't have a feeling I had appendicitis, I had the logical conclusion that I might possibly have appendicitis. To use a mental model, you do just have a feeling.