Content-type: text/html ~ Stephen's Web ~ Simulation as Education

Stephen Downes

Knowledge, Learning, Community

Jan 18, 1999

Posted to the Node 19 January 1999

Kim McShane wrote, In Section 3 of your paper 'New Technology in Education' you hold that online education will move towards simulation-type environments. I am wondering if simulation alone can prepare a pilot, for example, for ALL she can expect to encounter in her occupation. You know, can simulation account for everything?

I guess I am thinking that the trainee pilot should also have to be exposed to some of the theoretical instruction/background so that she can understand the principles behind key manoevres and situations. Once one analyses, understands and can synthesise the principles in multivariate ways, one is able to adapt to different and indeed even unpredictable circumstances.

The theory behind instructive simulations needs to be exposed and made explicit to the learner/trainee I believe. I value reflective students who can engage in flexible decision-making, without resort to a coda or recipe. Such students are also less likely to be manipulated unquestioningly by their employers and indeed by those who produce simulations (!). My point is that simulations may narrow the range of experiences trainees have rather than develop the micro and macro perspectives which are both needed. Is theory and reflection out...going...going...gone?

This question displays deep insight into one of the more fundamental problems of learning, and I confess, I had not considered these questions when arriving at my predictions regarding simulations. That said:

The issue in question centres around the proposal that one can not (or would not tend to) arrive at general principles on the basis of concrete instances. There are those (Chomsky, Pylyshyn, Fodor, who would argue that, in order for a general principle to be learned, the general principle must in some way be made explicit, because the general principle is underdetermined by the concrete instances.

Against this argument is the view, advanced by Kosslyn and the PDP/connectionist crowd, that what we think of as general principles (and there is considerable debate, stemming from such empiricists as Hume and Mill, as to exactly what constitutes a general principle) can be dervived from concrete instances.

In the end, in my view, the debate boils down to whether we think our response to new situations is going to be rule-based, as the Chmosky crowd thinks, or exemplar-based, as the latter crowd would have (they would not all use the word 'exemplar').

Consider, for example, what an airline pilot would do if confronted in the cockpit with a large furry lat-like creature with extended claws and gnashing teeth. On the rules based theory, the pilot would consult some sort of general principle, such as 'If confronted with large furry lat-like creature with extended claws and gnashing teeth, shoot it'. On the exemplar-based theory, the pilot would think, 'It looks like a tiger, and tigers are dangerous, so I should shoot it' (the latter is a gloss, because the pilot would not even think in such a formal way).

Now of course, pilots are unlikely to encounter tigers in the cockpit (thankfully), however, they will encounter novel situations. And the weakness of simulation-based training is that, no matter how many simulations are run, it will not be possible to simulate all possible (or even all likely) novel situations. The advantage of employing generalizations and theory is that the pilot will be able to apply the general principles of the theory to the specific case in hand.

So the question boils down to: could the trainee construct the required theoretical principles solely on the basis of simulations? And even if so, is it better that they arrive at these principles as a consequence of learning by simulations?

My answer to both questions is in the affirmative, with a caveat. I believe that it is possible do so because, in fact, we have done so. Nature does not come equipped with a manual detailling all its various general principles. Each general principle that we have ever derived, we have derived on the basis of experience. True, these principles are often subsequently taught as general principles, but that does not mean that the students could not have learned them independently of such teaching.

Moreover, I believe that it is better to enable students to derive general principles on the basis of their own experience, rather than to be explicitly taught the principles. The reason for this is that we want these principles to be embedded as deeply into the pilot's knowledge as possible; to paraphrase Dreyfus & Dreyfus, we want the pilots to be "experts", where the knowledge is "intuitive". A pilot who, when crashing the plane, must consult a stock of internalized rules before deciding what to do is a bad pilot; we want the pilot to react instinctively. But instinctive reaction is possible only possible if the principle in question is deeply associated with concrete instances in the pilot's mind, and such a deep association is only possible though simulation.

Now for the caveat, and it is this: humans notoriously get the general principles wrong when they learn from experience; inductive inference is often a trial-and-error sort of thing. However, in the case of pilots, we do not have that luxury; it is the sort of environment where you get to make only one mistake. Therefore, I would urge the teaching of general principles as a failsafe. We would want to construct our simulations in such a way as to attempt to evoke in the pilot's mind the construction of a general principle, and then compare the principle thus evoked with the principle representing the state of the art in the field, and correct the pilot's perception if necessary.

It should go without saying that people who consistently generate incorrect general principles in cockpit simulations would probably make bad pilots.

Related to this please tell us more about the kinds of learning experiences you anticipate and (if education prepares individuals for social participation...?), what kinds of social beings are we preparing through/with/via these technologies?

I studied mathematics and logic in the time-honoured manner of ages past: the instructor would propose a rule or principle (possibly derived from other rules or principles, which had been covered previously). The instructor would then provide some examples of the rule or principle in operation. We would then be invited to apply the rule or principle in other applications, thus mastering the principle by means of (as Kuhn puts it) "solving the problems at the end of the chapter".

In my view, this is why we have so few mathematicians and logicians in the world, and why high school (and even college) graduates today have such difficulties seeing through the fallacies contained in, say, advertising. To apply a rule or principle to a novel situation requires a mastery of the art of formalization, the capacity to subsitute concrete terms for variables. But formalism per se is, it seems to me, rarely taught, with the consequence that only those students with an intuitive understanding of formalism can learn in this way.

That said: it was the only practical way of teaching. It is not possible for a single teacher to run a class of 30 or more through a variety of experiences designed to induce a general principle. In my own classes - teaching adults with little or no educational background, and no intuitions about formalism at all - I found it necessary to sacrifice progress for the sake of retention; I focussed on a pew principles which could be derived from experience. For although I felt the students could probably memorize the principles in question, I also felt that they would never use them.

New technology allows us to draw into the educational experience in a much greater degree the experiential component, because the intervention of the instructor is no longer required to mediate the experience, and because the availability of quality learning tools will relieve the instructor of the need to create the experience. And because (as I argue above) experience is a better teacher than principle, it is my belief that, in the interests of better learning, instructors will tend toward the use of experience in teaching, and away from the use of theories and principles.

This - interestingly enough - applies as much to social interaction as to any other component of learning. It is interesting that the defenders of traditional schools defend them on the ground that they teach students the essentials of social interaction, and yet almost all of that instruction is based on experience (there being few etiquette classes in schools today). Future learning will see this experience enhanced and extended.

Even today, we are seeing students' social interactions with each other spanning the country, and the globe. I believe (though this would need to be reserached) that this is deepening their understanding of social interaction. It should, in theory, because a wider range of samples produces more reliable inferences than does a narrow one. Moreover, as I mention in my essay, students will be encouraged to be more mobile, thus extending the range of their interactions to offices, shops, and the community at large.

Stephen Downes Stephen Downes, Casselman, Canada

Copyright 2024
Last Updated: Jul 21, 2024 07:19 a.m.

Canadian Flag Creative Commons License.