[quoting Wild, 3 Apr 96] Without wishing to step on Don Norman's toes here, its probably of some value to point out that Doris almost answers her own question: the terms she has highlighted as used by Don, are generally, alternative descriptors (albeit less precise) for "mental models" (or at least the mental models of Johnson-Laird's characterization).
I'd have to disagree. Not to say that Martyn's characterization of Johnson-Laird's use of the term "mental models" is wrong, but to argue for a tighter meaning for mental models. Obviously, I'm not speaking for Don (he's welcome to agree with me, of course :-), but I'd reserve mental models specifically for cognitive (qualitative) models of dynamic systems, and use "conceptual frameworks" as higher-level representations that organize disparate bits of knowledge, including mental models.
For instance, your conceptual frameworks for nuclear reactors may include a mental model of heating of a liquid to a gas to drive a turbine, but it may also include knowledge about the safety factors, a few specific incidents, the potential for spent fuel conversion to weapons, etc. As such, conceptual frameworks may incorporate a variety of knowledge, including declarative knowledge, causal knowledge, knowledge of common events, etc. In short, these conceptual frameworks are the organizing structures known variously as "schemas" or "frames" in the psychological literature.
While the overarching schemas are necessary, I'd like to argue that good mental models are a most powerful reasoning mechanism. They allow us to understand systems, and as such they provide qualitative bases for prediction and explanation. And this is a capability that's been hard to address in the past but technology gives us new means to address.
Many systems are too dangerous, expensive, etc., to experiment with. With technology, however, we can create a simulation that allows safe exploration. Even more, we can provide building tools to construct our own models, which allows us to test our understanding in a different way (see my comments about exploration vs. construction in the my previous message [1 Apr 96]).
As a second issue, Don's notion of assessment is an issue I'm coming right up against.
[quoting Norman, 3 Apr 96] The best measure of performance of a person is performance itself, not retention, not artificial, abstract tests.
Yes, the idea of a nice competency-based assessment of performance sounds great. However, the biggest problem (which echoes what Kent was saying about training vs. education) is far transfer. In training (in my definition), we have a clear context and performance indicators. But for certain areas, design for instance, what is the performance indicator? What is a good measure of ability? In teaching my HCI subject, I can't know whether my students will be facing designing small utilities, large industrial systems, or household appliances. I can't train them, I need to educate them. What are useful measures of performance (for that matter, what are useful practice tasks) to ascertain whether they can transfer to many different areas? I can't test them all. I hardly have time to provide good practice, given 14 weeks and the need in such iterative design to provide for a cycle of feedback and revision, as well as have a finite amount of marking to do (I have 80 students, not 15).
At graduation time--which would have some minimum number and assortment of modules as a requirement--one could evaluate people by what kinds and how many modules they finished. The good mathematician or scientist would have finished more modules of math or science than the person no t so good in these topics. But everyone would have the same grade on the modules they did pass.
But is this right? Do we want students to do this without some interaction with people (currently known as counselors, tho' there are lots of problems with the current version thereof) who can advise on how the different choices mesh with desired areas of endeavor? Certainly high school graduates shouldn't be expected to know what they want to do. I changed major several times while in uni (and took six years), let alone the diversions in post-grad study (Don may painfully remember this :-).
As an aside, the Open Learning agency here in Oz almost meets this criteria. It packages a number of subjects from different uni's and makes them available through distance mode. You can actually get a bachelor of arts degree from one of the uni's after completing a set number of any of the subjects! You get marks, but you only have to pass the subjects. The question then becomes, what is the degree worth? Will industry recognize it? Etc.?
We're traipsing all over the landscape here (which is what makes this such a great mailing list). We're addressing the realities of the learning environments (hey, I still lecture), the underlying components of learning, the changing context in which this is happening, and how technology can be adapted to the pedagogical goals. I'd argue that the bottom line, however, is to design good learning activities, and then see if technology can make them happen. We'll end up compromising, but the technology can make a lot more possible if we throw off our blinders about what'd been done and start with an approach I call the "no limits" approach: if there were no limitations at all (except mind-reading, for hopefully obvious reasons) what would be the ideal activity-reflection cycle? It'll be different for all sorts of learners/content/contexts, but from that spec, we are increasingly able to approach it with the increasing capability of technology.