More on Learning

Addendum to my previous education posts: In general, the best way to learn anything is to simply jump in, do some wild and crazy stuff, make mistakes, get quick, accurate feedback about how well you are doing, and benefit from the previous experience of others in the environment. This goes for both humans and computers, and is essentially true in all environments.

It is in theory possible to learn without direct interaction with the environment, but the learning rate takes a major hit, a minimum of a factor of three to five slow down. Again, this applies to both humans and computers, across all skills. One of the few interesting unifications of AI and human psychology has been a partial empirical theory of learning. This is about the only result it has, but it's worth knowing.

In fact, there are some classic experiments that demonstrate the counter-intuitive result that if you have two learners, one in control of the environment, and one wired to receive the exact same incoming data (obviously only possible for computer programs), using the same learning algorithm, but not allowing the second any interaction with the environment, the second learner will be much slower then the first. There is great value in being able to form hypothesis and test them directly. (I'm eliding a few of the details as irrelevant; obviously in the computer scenario in a deterministic environment the second learner could think it was interacting, even though it wasn't, and seem to learn just as quickly, but that's not the point of the experiment... and the real world is less predictable then you think.)

In light of this virtually undeniable fact (in terms of the scientific and mathematical backing it has), it is obvious that the current educational system is very flawed, in that none of the conditions necessary for good learning are present.

• Feedback for a given performance trial (test, quiz, whatever) is rarely faster then a day. On those rare occasions where a person is using a computer, the feedback is instant but the experience is ruined by other problems with computerized drilling, such as the terminal boredom the student experiences. Accuracy is pretty hit-or-miss in the subjective environments, too... I never did figure out how my Art grades were determined.
• The learner is held back from exploring the environment, and instead must learn through listening to somebody else talk about it, or reading somebody else talking about it. Obviously that criticism applies somewhat less to fields like history, but even so, the learner should be encouraged to explore on their own, which the system generally discourages, because it's impossible to evaluate the results of such exploration numerically.
• Similarly, the learner is heavily penalized for all mistakes on all evaluations; only rarely are there "test runs" where feedback is given, but does not count directly towards "the grade". The truth is that a person who gets a C, then an A on a "re-test" may very well understand something better then someone who gets an A both times, then immediately forgets everything, but the second person gets the better evaluation.

About the only thing the current school environment gets right is providing the experiences of others, but then, without that, what would be left?

Once you learn about the process of learning, it becomes difficult to imagine a more backwards educational system then the one currently in use. People learn in spite of modern schooling, not because of it.

Reformations of the school system are actually quite easy to imagine; just directly attack the problems mentioned above. But the problems are so deeply ingrained into the symbols with which people think about "schooling" that the solutions seem too radical to work. Ironically, those are the only ones that will work; the simple tweaks here and there which sweep the educational community like ocean waves are doomed to failure, because they do not attack the root.