who has done the highest quality research on learning (and transfer learning in particular) in humans? specifically, i’m curious to answer questions like:
how much does doing things make you good at other things of varying degrees of similarity? how much of the value of having done things different from the thing you care about is (a) signaling that you are competent in general, (b) learning extremely general things like how to manage your time well or how to update on evidence, (c) extremely specific and ungeneral facts like a particular theorem or debugging technique, or (d) literally everything else in between.
if your goal is to be good at X, under what circumstances is the most efficient way to become good at X not just trying to do X (and instead, to learn from a curriculum, do some other thing with a tight feedback loop, etc?)
I looked into transfer learning a while ago, resulting in this post, it contains some pointers to further literature. I was not particularly impressed by the literature, but it’s a thing that’s hard to study. Open loops were investigating error-based learning, video/audio self modeling, self-explanation (talking to oneself (an LLM?) and explaining something while learning/thinking). Some thoughts about feedback loops here.
who has done the highest quality research on learning (and transfer learning in particular) in humans? specifically, i’m curious to answer questions like:
how much does doing things make you good at other things of varying degrees of similarity? how much of the value of having done things different from the thing you care about is (a) signaling that you are competent in general, (b) learning extremely general things like how to manage your time well or how to update on evidence, (c) extremely specific and ungeneral facts like a particular theorem or debugging technique, or (d) literally everything else in between.
if your goal is to be good at X, under what circumstances is the most efficient way to become good at X not just trying to do X (and instead, to learn from a curriculum, do some other thing with a tight feedback loop, etc?)
is habryka’s 4 factor model of skills accurate?
i’m sure all of you have takes on these. but i’m specifically interested if anyone has gone out and done really high quality studies.
I looked into transfer learning a while ago, resulting in this post, it contains some pointers to further literature. I was not particularly impressed by the literature, but it’s a thing that’s hard to study. Open loops were investigating error-based learning, video/audio self modeling, self-explanation (talking to oneself (an LLM?) and explaining something while learning/thinking). Some thoughts about feedback loops here.
Just an obvious thought, not sure how useful: https://en.wikipedia.org/wiki/Evidence-based_education