First of all, an announcement: The April issue of the Communications of the ACM (CACM)is now out. It includes this (apparently) controversial paper by me and Jim Spohrer, as well as the first set of articles on the topic. The rest of the articles will be published in the May issue of the CACM.
Second, it sounds like I may have made a bad choice of paper for this exercise. The paper was meant for a specific context: as an introduction to the papers that had already been selected for publication in a special issue of the CACM. We had no part in the section process. Moreover, the introduction was not a place for us to express our complete theory of instruction. Our goal was to provide a framework for understanding the papers. No more, no less.
Finally, let me try to provide a brief review of my views on instruction.
The most important thing is understanding. "Facts" are not the critical part, with mild exceptions. Yes, it is important to have memorized: the order of the letters in the alphabet, the addition tables, the multiplication tables, and a few other things here and there. But on the whole, I don't believe in facts. I believe in understanding. One can always look up facts. (And most facts aren't facts anyway. History, for example, is really a set of points of view, not "facts.") . With a good conceptual framework and understanding, usually the stuff that one needs to know is easy to remember.
As I have pointed out in another spot (in my book Learning and Memory), when you met someone in the hall who says, "I'll meet you at 5:30 for dinner," you usually do not have to go around doing a rote memory exercise on this piece of information: it fits a well established conceptual structure, and you remember it without effort. That's how learning ought to be, and it is the instructor's responsibility to provide that rich, functional conceptual structure.
How does one learn? How does one instruct? Answer: there are many kinds of learning, many different kinds of students, many different situations, and therefore, many different educational procedures. Dave Rumelhart and I once argued that there were three different kinds of learning: Structuring, accretion, and tuning. Each required very different instructional procedures.
Structuring is where one develops the proper internal representations--the structures. This requires reflection, deep thought, This is where simulation-based instruction is superior, where tools that help students contemplate different courses of actions and different approaches are needed. Tools that let students reflect upon this and that. This was the focus of my discussions in my book Things That Make Us Smart.
Accretion is the sheer accumulation of knowledge. (Not facts--ugh!) This can be done by demonstration, or lecture, or by reading, or watching films. Here is where I show you how to juggle, or that 5 + 3 = 8. (The understanding is part of structuring.)
Tuning is what is required to transform knowledge into automatic skills. This requires practice--considerable practice. I can teach you 3-ball juggling in 30 minutes. It will then take hundreds of hours of practice to be comfortable. I once argued (and still believe) that it takes 5,000 hours of practice to become expert. A little over two years of full time effort. This is tuning.
Learner-centered education means that the learner is the center--an active center. Kent Thomas, in his reply to me, seemed to think that learner-centered somehow could be done through lecture and demonstrations.
No.
Kent tells me he is a trainer. What an interesting term. I guess I am not sure what that is or how it differs from teachers. He said:
[quoting Thomas, 1 Apr 96] As a trainer, I believe in "learner-centered" simulations in CBT/MM absolutely, but I use simpler tutorials (equivalent to demonstrations or lectures) for most of the content. A typical skills training course might include:
50% "linear tutorials" presenting introductions, basic facts, background information, demonstrations, etc.
30% "path simulations" or "prompted simulations" (actually a complex, branched tutorial in most cases) providing guided practice as the learner begins applying the information, and
20% "free-play simulations" to provide the learner independent or unguided practice in applying the information, with a critique of performance at the end.
By the way, the second method costs roughly twice the amount of development time as the first, and the third costs roughly twice the development time as the second.
Sounds horrible to me. That's the kind of course that bores the hell out of students. As for the cost structures--well, the whole point of the CACM articles was to try to develop instructional methods that reduce the costs. Just because the first method is cheaper than the others doesn't mean we should do it. Actually, I would never use the second or third methods. Branched tutorials are murder to develop and prepare. And it is the rare one that truly stimulates: they are just too hard to do. As for the last, free-play is seldom appropriate. I believe in adding much more structure. I so structure the situation that the students have to discover exactly the information I want them to learn. "Free-play" is not learning. (In skill training, we distinguish between playing a sport and practicing a sport. In the former, little learning may occur.)
My philosophy is to instruct through problems. Present a problem that the students are to solve, one that is intrinsically interesting. Structure the problem so that along the way, they have no choice but to encounter exactly the issues you want them to learn.
Provide learning material: demos, lectures, reading material for them to use when they have become motivated to need that material. Provide them with environments that support their exploration of the problem space--this is where simulations might be useful, but only where appropriate.
Clark Quinn [1 Apr 96], from down there in Australia, revises our classification scheme. Clark's proposal makes a good deal of sense. But once again, we have the problem that Jim and I were trying to make sense of the papers and to provide a gentle critique along with our encouragement: hence our choice of engagement, effectiveness, and viability. We were very concerned that these papers were, yet again, promising the moon but, in fact, simply demonstrating that they could climb trees. We were extremely concerned with how well the ideas would actually work in a real instructional environment, with real material. Hence our classification. Clark's is superior for overall instructional purposes.
Back to Kent Thomas who says: "As a trainer, I believe in "learner-centered" simulations in CBT/MM absolutely" but who also states "a total "swing" toward learner-centered education is a big mistake."
Well, who ever said we wanted a total swing toward anything? Moreover, our views of LC education seem to differ. I believe strongly in providing a lot of structure, and in making the environment guide the student to exactly the points that need to be explored and learned. Exploratory learning works if the teacher has set up the situation that the exploration always happens upon just the concepts to be acquired. Otherwise it fails.
As my colleague Alan Kay loves to point out. Students won't invent science, or the calculus, or Beethoven, or Shakespeare. We have to guide them to discover these, to study them, and to enjoy them. Traditional education, however, rubs the students' noses into the material, often diminishing the enjoyment, and thereby minimizing what gets learned.
[quoting Thomas, 1 Apr 96] In my humble opinion, we as practitioners desiring to consider ourselves professionals should not limit ourselves to any one theory, strategy, method, or technique. We should select what we think is most appropriate from as large a "tool kit" as possible.
Yup. Absolutely correct.