Part of the general topic of control in computational systems; the order in which subtasks are performed. I mentioned this in the context of inference; it can be applied in many other contexts. E.g. learning systems. Do they try to generalise whenever new data comes in, or do they wait until they have some goal in mind, and then start generalising from stored cases? Case-based learning is largely an example of the second; inductive logic programming, of the first.