Another idea that I had is to model the student as a computer. For instance, how big are student's caches (short-term memory)? How much long-term memory do they have? How fast can they put things into memory? What is the new concept processor speed? These things might be interesting to consider when designing classes and lectures. I think these ideas are probably in other fields (like cognition of learning and memory and neuroscience), but casting them into the computer model might be enlightening, plus it can help to teach TAs about students -- it can even be a kind of educational crutch for computer-science students to understand computers.