Following Don Norman's paper, Steve Tripp [5 Apr 96] and he discussed specific versus general skills versus knowledge. I have two points to propose. My second point is probably more important than my first.
Their discussion seemed to be driven by the issue of generalization, with them trying to fit ideas of learning, skill, and knowledge to that issue. Actually generalizability may be almost independent. Consider reflexes (as Rodney Brooks has made us consider in robotics and Artificial Intelligence) such as jerking your hand away when it touches something very hot. If forced to choose, you have to say that this is skill not knowledge. It is also very general, in that it is applied by your reflex system (your spine?) to all situations, and furthermore it is probably right more often than not (so it does "apply"), even though it probably has a higher error rate than a more thoughtful approach that understands the specific situation before acting. In fact skills are judged, even defined, by their fast execution times: if someone does something slower than you, you wouldn't call them skillful, although they also have to succeed (have an adequately low error rate). Conversely knowledge is defined by its general applicability: if it isn't true across many situations you wouldn't call it knowledge, although if the knower is very slow retrieving the information we often wouldn't count it (is it knowledge if it can't be accessed?). So the suggestion here, inspired by Rodney Brooks, is to focus on a tradeoff spectrum between accuracy vs. processing speed (knowledge vs. skill), to suggest that this is closest to the skill-knowledge contrast, but that it is largely independent of the generality (breadth of applicability) issue. One way to get a speed increase is to store ("cache") redundant facts that will come in useful: when I was taught my multiplication tables, this was the implicit rationale that knowing that 6*7 was 42 saves having to do 6 or 7 repeated additions. However given procedures for doing calculations such as 46*77, the cached fact turns out to be "general" and is re-used in many different cases.
My second point concerns another issue that cuts right across these distinctions in what I believe is an illuminating way, and certainly one that is relevant to education. In her 1993 book Rethinking University Teaching Diana Laurillard suggests that all subjects have two levels that should be taught and learned: the level of public, formal descriptions and concepts, and the level of personal experience and action. In chemistry this would be learning formulae vs. knowing what copper sulphate looks like. In English Literature, this would be knowing how to read and write critical essays vs. the emotional experience of seeing a good play performed. In language learning this would be knowing how to answer questions about the subjunctive form, vs. communicating successfully with a street vendor. I am more and more convinced of its usefulness, although even now I can still find this distinction quite hard to apply.
(For instance how exactly should it be applied to math? Is the personal level that of "conservation" issues: e.g., becoming convinced that 5 objects are still 5 even when moved into different spatial configurations? Or is it to do with the fundamental step of believing that written digits really can represent real quantities, and that after the purely abstract transformations of paper arithmetic, the answer will then really correspond to the world? Or is it just the speed-from-practice issues of learning multiplication tables, or getting hours of practice with different forms in integration problems?)
In standard teaching this approximately corresponds to lectures vs. labs. It corresponds still better to the issue, so prominent in the last decade or two in the education literature, of the pervasive nightmare in science teaching of producing students who pass exams in the conceptual level, yet reveal in simple interviews or other probes that they have made no connection between this conceptual knowledge and everyday personal experience and often still exhibit the same "misconceptions" as the uneducated do. Obviously the "neo-constructivism" (as I will call it) described in the Norman and Spohrer paper seems to address this in part by teaching through real world cases and personal learner activity. However I feel Laurillard's distinction is insightful because it points at a fundamental underlying distinction, too often a divide, between public knowledge and private experience.
Perhaps the most important implication is that learners are implicitly faced with two different arenas in which to exercise what they learn: two quite different kinds of task. The public formal aspect is exercised in public communication tasks: passing exams, writing reports, teaching others, getting advice and information from reference books and other experts. The personal aspect is exercised in perception and action: in accomplishing material tasks personally. If you consider professions such as medicine or civil engineering you can easily see the importance of both aspects.
Laurillard's distinction has many other aspects (for instance, its echo of the Piaget/Vygotsky contrast in perspectives on learning). Here however it could be used to give a penetrating critique of particular attempts at neo-constructivist software. Consider trying to teach basic economics, and the law of supply and demand, through specific real world cases. You could pick cases from the stockmarket and try to engage students with references to enormous sums of money and the attractions to them of a Yuppie lifestyle. You could pick the example of a loaf of bread: why does it cost 30 pence (say) in the UK at the moment? Or you could pick the example of dividing up household chores. The first might seem the most "engaging," and would probably attract a multimedia producer the most. The second will look boring, but at least the learner has daily experience of what a loaf of bread is. The third however is probably what should be chosen to connect to the learner's personal actions and experience: few will have fixed the price of a loaf, but they will all have had to negotiate chore apportionment in some local climate of supply and demand, and this skill of determining prices by taking full account of the particularities of the situation is what real businessmen need. These examples show how "engagement" and "motivation" may determine what people choose when going to a movie, but may not address the educational issues. They also show that simple personal familiarity may not be the issue in itself: learner action is what is important, and multimedia may struggle to make this connection however effortlessly it can achieve glamour.