Seebach: It seems that even educators have a hard time learning

Published October 21, 2006 at midnight

Text size

More Opinion Columns & Blogs

The idea that the most effective way to teach students is to give them problems to solve, with only minimal guidance, has been around for half a century or so, under a variety of names.

And it's always been wrong. In a paper published earlier this year in the journal Educational Psychologist, Paul Kirschner (Utrecht University, The Netherlands), John Sweller (Australia's University of New South Wales) and Richard Clark (University of Southern California) review the sorry history of this failed theory, and explain "Why Minimal Guidance During Instruction Does Not Work." It is bound to fail because of the way human memory functions. (The paper is available at www.cogtech. usc.edu/publications/kirschner_Sweller _Clark.pdf online.)

To simplify, people have two kinds of memory. Working memory is what you use when you've just looked up a phone number and dial it without looking at it again. Working memory is very small; probably seven bits at the most (which is why I have to write down phone numbers - I can't remember them long enough to finish dialing).

Long-term memory is vast, which is why excellent chess players can glance at a game in progress and reproduce the board without looking, but are no better than novices at remembering the placement of pieces put on the board at random.

And long-term memory is not just a scattering of isolated facts. Like the chess player's expertise, "it permits us to quickly recognize the characteristics of a situation and indicates to us, often unconsciously, what to do and when to do it," the paper says.

Learning, by definition, means a change in long-term memory. And whatever your favorite term for minimally guided instruction, whether discovery learning, problem-based learning, inquiry learning, experiential learning or constructivist learning, the problem is that it puts such a load on working memory that long-term learning can't take place.

If there were actual evidence that minimally guided instruction worked, then you'd expect it to have its defenders. But the evidence points the other way. The authors cite a 2004 review of the evidence, which suggests that every decade or so a new form of the idea pops up under a different name. "Each new set of advocates for unguided approaches seemed either unaware of or uninterested in previous evidence that unguided approaches had not been validated," they say.

When I write about this subject, I usually hear from teachers who tell me I'm wrong because what I'm saying disagrees with what they learned in ed school. Yes, that's the point.

The large number of studies they cite establish that strongly guided instruction, including examples, "resulted in vastly more learning than discovery." And the studies using a control group that reach an opposite result scarcely exist.

Studies going back at least 20 years have demonstrated that students who study worked examples learn more - that is, they perform better on subsequent tests - than students who are required to discover how to solve equivalent problems. "Learners can engage in problem-solving activities for extended periods and learn almost nothing," the authors say.

The worked-example effect eventually disappears as learners become more proficient, and by the time they are experts it even reverses, because studying examples they already know how to work is wasted effort. But by then, problem-solving is effective because they already have the necessary information and structure available in long-term memory.

Some studies have shown that lower-aptitude students might even know less after unguided instruction, although they tend to like it. "Higher aptitude students who chose highly structured approaches tended to like them but achieve at a lower level than with less structured versions but did not suffer by knowing less after than before instruction."

We're not talking just about children here. Some 60 medical schools, the authors say, have adopted problem-based learning over the past two decades. They cite an analysis that concluded the students in these programs, compared with students in traditional programs, do get better ratings for clinical skills (although that may be because they spend a lot more time in clinical settings). But the negatives include lower scores on basic science tests, more study time and a pattern of ordering significantly more unnecessary tests at a much higher cost per patient with less benefit.

Other research on medical education indicates that once medical students have learned something about basic science in the context of a particular patient's case, they have trouble separating the two. These students generated more elaborate explanations, but "they had less coherent explanations and more errors."

This would not appear to be a step forward in the practice of medicine.

These misguided forms of pedagogy, the authors suggest, come from confusion between how experts practice the skills they have acquired - research, obviously is a kind of problem-solving - and the most effective way to impart those skills to novice and intermediate learners. There's no particular reason to think they ought to be the same. Given what's been learned about memory and cognition, they are almost certainly not the same.

Teaching methods based on the assumption that they are are almost certainly going to be ineffective, yet enthusiam for them in educational circles continues undimmed. After all this time, why is that?

Linda Seebach is an editorial writer for the News. She can be reached by telephone at (303) 954-2519 or by e-mail at seebach@RockyMountainNews.com.

Subscribe to the Rocky Mountain News