Learn By Doing, Not Watching
Decades ago the famous “gondola kitten” experiment demonstrated that one must actively explore if one is to learn. One littermate in the set-up was free to explore its environment while another hung passively suspended in a contraption that moved in parallel with the exploring kitten. The gondola passenger saw everything the exploring kitten did but could not initiate any action. The mobile kitten discovered the world for itself while the passive kitten was presented a fait accompli-world in the same way that screen images are passively delivered to us. The passive kitten learned nothing. Since this classic experiment we have come to appreciate how crucial self-directed exploration is to understanding the world.
This holds true for humans as well as kittens. In an update of the gondola kitten experiment, researchers recently videotaped an American child’s Chinese-speaking nanny so that a second child saw and heard exactly what the first one did. The second child learned no Chinese whatsoever, whereas the first child picked up quite a lot. (more)
This supports my suggestion to Chase Your Reading; you more learn to figure things out by trying yourself to figure things out, and less by passively listening while writers figure things out in front of you.
P Halmos teaching method? He would give his students a list of theorems and say: prove these and you have learned the material.
I believe this greatly oversimplifies learning. We learn all sorts of things, with some things relying more on procedural knowledge, some relying more on declarative, some more on observational, and many things being the result of a complicated mish-mash of modes. If you're lost in a strange city, you're much better off observing the directions on a map, or asking for them, then you are just trying to find your way on your own. People don't learn to tie shoes by trial and error.