Stephen Downes

Knowledge, Learning, Community

Jun 12, 1998

Posted to DEOS-L 12 June 1998

G.J. Burt asks, Is learning style a type or a dimension?

I think that's a good point, and I think he answers his question in the remainder of the post. Learning style seems to me almost certainly to be a dimension, or as Burt writes,

...people vary continuously in a variety of directions to form something like a multidimensional version of the normal distribution (or some other unimodal distribution)?

It would be interesting to see empirical research verifying this assertion. But I think we can obtain a prima facie case for it from reflection on our own learning styles. In my case, at least, I can learn at least to *some* degree in a variety of modes (good thing, too, otherwise I would have failed high school).

If so, then it is difficult to do other than to provide for the average position of the distribution. Most people are fairly near the average. People who are not near the average are just a small tail of the distribution and so it is difficult to justify on economic grounds making separate provision.

I disagree. This statement is based on the assumption that there is a single thread in learning. But that need not be the case. Indeed, in my opinion, the best learning materials are multi-threaded. They provide materials for each of the several learning styles at the same time.

From a software design point of view, I have called this approach 'tiering' (see http://www.umuc.edu/iuc/cmc96/papers/downes/tech1.html ).

From a design point of view, considering a learning material to be composed as a series of concepts to be learned, the typical representation might be:

concept1 -> concept2 -> concept3 ->

For each concept, however, the material should be seen as being composed of a series of distinct threads:

-> concept X ->
-> text-based representation of X ->
-> image-based representation of X ->
-> flow-chart based representation of X ->
-> audio representation of X ->

and so on. (Obviously these threads could be refined).

If learning styles, as Burt suggests, are a dimension, then presumably we could quantify the degree to which a person relies on one or the other. For simplicity's sake, assume 1-10. Then, provided each thread contains the equivalent of information required by a 10, all learners, and not just the average learner, can be completely satisfied by the material.

But what's *really* interesting is the possibility of tailoring this system. Which is why I'm writing this post.

Let's assign, to each learning element, a weight. Say, for example, if a picture is very important as a representation of the concept, give it a weight of 10. If it is less important, give it a weight of 2. Or for text based materials, a major heading or fundamental definition might be given a weight of 10, while a parenthetical remark may be given a weight of 3.

Then, for each learner, give them a learning styles profile. Assign a value of 1 to 10 for each identified learning style.

Now the presentation of any given concept will be composed of several threads, one for each learning style, and the components in those threads each have a weight. When presenting the material, present only those materials whose weights exceed the learner's profile value for that learning style. This thus tailors the presentation of the materials to the individual's learning style.

How do we determine the learning styles profile. I would suggest two methods:

1. Test-based, system generated. Have the learner take a series of diagnostic tests, which are evaluated by the system. The system them sets weights for the learner according to the test results.

2. Control-based, user generated. Provide the learner with a panel, on which there are sliding controls, one for each learning style. Have the learner set his or her own preference.

The hard part in all of this is not designing the mechanism to vary the presentation for each user. The hard part is in designing learning materials for different styles in the first place. Many learning materials out there are designed for one or two learning styles at best. Designing a multi-threaded presentation for *each* concept in a course is a large task.



Stephen Downes Stephen Downes, Casselman, Canada
stephen@downes.ca

Copyright 2024
Last Updated: Mar 29, 2024 03:50 a.m.

Canadian Flag Creative Commons License.

Force:yes