Spaced repetition helps, but how do spaced-repetition researchers have their subjects practice within a single practice session? I’d expect optimized practice to involve not only spacing and number of repetitions, but also an optimal way of practicing within sessions.
So far, I’ve seen a couple formats:
Subjects get an explanation, examples, and a short, timed set of practice problems.
Subjects practice with flash cards. Each “round” of flash cards involves looking at only the cards they haven’t seen or got wrong last time. They do back-to-back rounds of flash cards until they get every card in the “round” right. In the end, they’ve gotten every card correct once.
This helps me answer a lingering question. What sort of cutoff for “good enough for today?” are these psychologists using for their practice sessions?
The answer seems to be either
Completed a pre-defined, do-able workload or
Was able to memorize everything, even if they’d only referred to the flashcard just a moment before.
I adapt this for my math studies by listing concepts, theorems, and proofs from 1-2 sections of a chapter in a Google Doc. I mark the ones I can’t remember/got wrong in colored text, and the ones I got right in black. Then I try to rehearse the marked ones again, like with flash cards, until I have recited each proof/concept from memory once.
I also will try to solve one or two relevant problems from the back of the book. They’re typically much more involved than anything I’ve seen in the psych literature.
One way I suppose psychologists could try to study the “optimal spaced repetition study session” would be to determine to what extent overlearning bolsters retention in conjunction spaced repetition. So for example, if groups 1 and 2 do practice problems in week 1 and week 2, then get tested in week 3, but group 1 does 5 problems/session and group 2 does 10 problems/session, how much better does group 2 do on their test?
Theoretically, the concept of overlearning seems to posit that there’s a point at which a topic is “learned,” beyond which it has been “overlearned.” Is the benefit of overlearning to provide a safety margin to ensure the subjects actually learned it? Or is it to give extra practice even to subjects who successfully learned the topic in a smaller amount of time?
It seems like we want to know if there’s a sort of “minimal unit of learning.” Intuitively, there must be. If I just skim through section 7.1 of my linear algebra textbook, but never read it in depth, I probably haven’t actually learned anything. At best, I might retain a couple key phrases. But if all I ever do is skim the text, I probably never will learn the material.
So the relevant question is “what is the minimal unit of learning?” and “how do you know if you’ve achieved it?” There are probably multiple answers to the latter: can you solve a relevant practice problem? Can you recite the definitions of the concepts and the proof in your own words?
The former question, though, is more like “given that your level of understanding of topic T is at level X, what’s the least you need to do to get to level X + 1?” And that will depend on T and X, so it’s hard to specify in general.
But I think it’s helpful to frame it this way, since I at least often think in a pretty binary fashion. Like, to have “learned something” means that I can execute it effortlessly and accurately, and can also explain the underpinnings of why the process works.
It seems helpful to break down the multiple factors of what it means to “learn something” and put them on a scale.
Practice sessions in spaced-repetition literature
Spaced repetition helps, but how do spaced-repetition researchers have their subjects practice within a single practice session? I’d expect optimized practice to involve not only spacing and number of repetitions, but also an optimal way of practicing within sessions.
So far, I’ve seen a couple formats:
Subjects get an explanation, examples, and a short, timed set of practice problems.
Subjects practice with flash cards. Each “round” of flash cards involves looking at only the cards they haven’t seen or got wrong last time. They do back-to-back rounds of flash cards until they get every card in the “round” right. In the end, they’ve gotten every card correct once.
This helps me answer a lingering question. What sort of cutoff for “good enough for today?” are these psychologists using for their practice sessions?
The answer seems to be either
Completed a pre-defined, do-able workload or
Was able to memorize everything, even if they’d only referred to the flashcard just a moment before.
I adapt this for my math studies by listing concepts, theorems, and proofs from 1-2 sections of a chapter in a Google Doc. I mark the ones I can’t remember/got wrong in colored text, and the ones I got right in black. Then I try to rehearse the marked ones again, like with flash cards, until I have recited each proof/concept from memory once.
I also will try to solve one or two relevant problems from the back of the book. They’re typically much more involved than anything I’ve seen in the psych literature.
One way I suppose psychologists could try to study the “optimal spaced repetition study session” would be to determine to what extent overlearning bolsters retention in conjunction spaced repetition. So for example, if groups 1 and 2 do practice problems in week 1 and week 2, then get tested in week 3, but group 1 does 5 problems/session and group 2 does 10 problems/session, how much better does group 2 do on their test?
Theoretically, the concept of overlearning seems to posit that there’s a point at which a topic is “learned,” beyond which it has been “overlearned.” Is the benefit of overlearning to provide a safety margin to ensure the subjects actually learned it? Or is it to give extra practice even to subjects who successfully learned the topic in a smaller amount of time?
It seems like we want to know if there’s a sort of “minimal unit of learning.” Intuitively, there must be. If I just skim through section 7.1 of my linear algebra textbook, but never read it in depth, I probably haven’t actually learned anything. At best, I might retain a couple key phrases. But if all I ever do is skim the text, I probably never will learn the material.
So the relevant question is “what is the minimal unit of learning?” and “how do you know if you’ve achieved it?” There are probably multiple answers to the latter: can you solve a relevant practice problem? Can you recite the definitions of the concepts and the proof in your own words?
The former question, though, is more like “given that your level of understanding of topic T is at level X, what’s the least you need to do to get to level X + 1?” And that will depend on T and X, so it’s hard to specify in general.
But I think it’s helpful to frame it this way, since I at least often think in a pretty binary fashion. Like, to have “learned something” means that I can execute it effortlessly and accurately, and can also explain the underpinnings of why the process works.
It seems helpful to break down the multiple factors of what it means to “learn something” and put them on a scale.