Generalized Domain Knowledge

In technical subjects such as chemistry, students quickly get lost in the details. Those details are important, but they support a larger edifice, which in turn supports those details. Students need the whole shebang to grasp what’s going on. But how should we sequence the approach so that students aren’t drinking from a firehose and overwhelming their precious cognitive resources?

 

Here’s one study with a tantalizing, but jargon, title.

 

I’ve only shown half the abstract, because the rest is technical jargon that won’t make sense unless you’re in the field or you’ve read and understood the paper. But in the last line shown you can see that it’s a very small-scale study so the result should be taken cautiously.

 

First, we need to clear up some jargon.

 

Domain-specific knowledge is what helps those with expertise in a domain to efficiently solve problems in that field. It typically requires explicit instruction in the early stages, and requires conscious use of working memory to help consolidate the new knowledge. It consciously feels like your brain is working hard and thinking. Schools focus on teaching domain-specific knowledge because acquiring it without being explicitly taught is difficult and highly inefficient. Once you have some background expertise, it becomes easier to teach yourself more in that area, but you’ll still have to work hard at it.

 

Domain-general knowledge is what we seem able to acquire unconsciously simply through repeated exposure; it likely has an evolutionary component. It is not tied to a specific area of expertise or domain. You don’t need to be explicitly taught the material. You probably don’t remember how you learned it; and you’d be hard-pressed to articulate the process of learning without resorting to (weak) post-hoc rationalization. It seems automated and doesn’t feel like you’re having to think hard when you acquire and use such knowledge. However it isn’t useful for solving complex domain-related problems – any technical problem or any problem requiring some depth being prime examples.

 

Educators focus on the former rather than the latter for good reason. The first needs to be taught. The second does not. Given the increasing technical complexity of the world we live in, this explicitly taught domain-specific knowledge becomes even more important. Not that you can’t acquire such knowledge on your own – it’s just highly inefficient to do so on your own through seemingly passive observation without conscious engagement of what feels like hard thinking.

 

Kalyuga, the author of the article, splits domain-specific knowledge into two categories: specific domain knowledge and generalized domain knowledge. Note that both focus on the domain area. Here are the definitions. “Specific domain knowledge is applicable to a narrow range of tasks in the domain. Generalized domain knowledge is applicable to a wider class of different tasks in this domain; however it remains a part of domain-specific knowledge… [it] may provide a compromise in the trade-off between the generality and power of knowledge.”

 

The study aims to test whether constructing and providing a generalized domain knowledge schema is better at helping learners acquire and retain domain-specific knowledge. There is some evidence from their small study that providing a general-to-specific schema is slightly superior to a specific-to-general schema. So perhaps sequence matters. Both these outperform not providing a schema (which you can think of as scaffolding statements related to learning the material). Interestingly, the participants do not self-report significant differences in cognitive load in any of these. The technical task was to learn how an air-conditioner works, and participants were pre-screened to assess prior domain-specific knowledge.

 

I’m less interested in the result itself, but rather I’d like to consider what providing schemas for generalized domain knowledge looks like in areas that I teach.

 

First-year college chemistry textbooks (at least in the U.S.) have moved to what is called an atoms-first approach. Instead of stoichiometry, balancing chemical reactions, and doing lots of calculations involving moles and masses, early on in the first semester, these topics are now shifted later in the semester. Taking their place early are the interaction of light and matter, electronic configurations, and chemical bonding. It’s hard to say whether these broad moves constitute general-to-specific or specific-to-general schemes. If one argues that the macroscopic is central, then the move feels specific-to-general. If one argues the microscopic (or nanoscopic) is central, then it feels general-to-specific. The tricky part is that in the grand scheme of things (pun intended), both are equally important in my opinion. (See Johnstone’s Triangle.)

 

One of my areas of expertise (as a quantum chemist) is chemical bonding, so perhaps I’m biased in putting chemical bonding center-stage. The atoms-first move works well for my pedagogy overall, although I still quibble with the arrangement and flow of topics in any textbook out there. (No, I haven’t had sufficient motivation to write my own textbook.) A few years ago, I made a significant change to how I approach chemical bonding by spending an entire hour-long class session on what I think would squarely fall under providing a schema for Kalyuga’s generalized domain knowledge. You can see the scope in this previous blog post. Too bad that’s the only clear example I can think of that I use in G-Chem.

 

Throughout the semester, I’d like to claim in every class session (although I’m not sure if that’s true), I make statements in reference to generalized domain knowledge – but these are scattered throughout mainly as context reminders rather than explicitly teaching a schema. I do use a mind-map-ish activity in second-semester G-Chem for students to connect concepts in energy and thermodynamics, and there was once when I devoted an entire class session to discussing the idea of energy in conjunction with a Feynman reading. But otherwise I haven’t worked on designing an explicitly taught general-to-specific schema of the type that Kalyuga refers to. I’m not worried about this. I think the approach I’ve refined over the years works well, although I will be sorely put to the test going all-remote this semester. My schemes may or may not work. I’m likely to blog an end-of-semester update reflecting on how things went.