- Scott H Young - https://www.scotthyoung.com/blog -

The Importance of Knowing What You Know

If I had to speculate what was the biggest obstacle to learning well, I wouldn’t guess reading speed, memory or even procrastination. I’d say it was metalearning. It’s important, and most people are lousy at it.

“Meta” is a prefix that gets added to things that are about themselves. Metaphilosophy [1] is the philosophy of philosophy. Metamathematics [2] is the study of math itself using mathematical methods. Metalearning is when you learn about how much you know and don’t know in a particular domain.

Metalearning is important, because it’s easy to delude yourself into believing you know more than you actually do. In this great article [3] by cognitive scientist Daniel Willingham, he explains why many students think they’ve studied well—and then later bomb the exam. The difference, he says, is that students confuse familiarity with recollection.

Familiar Things You Don’t Actually Know

Familiarity is the sense that you’ve seen something before. Most students who study (particularly those who study repetitively without building insight), will get the sense that all of the information they’ve previously seen before.

That feeling of, “ah, I’ve seen this before,” is commonly confused with actually being able to recollect said information. Recollection is much more than familiarity however. To recollect something, you need to be able to produce the answer without prompt.

The confusion stems because familiarity has a feeling, whereas recollection usually does not. We all know what it is like to feel something is familiar. But, without actually testing yourself, recollection doesn’t feel like anything. As a result, our internal self-monitoring tends to assume recollection when all we really feel is that some information is familiar to us.

This is a failing of metalearning. When something feels familiar, but you can’t actually recollect it when self-testing, you have an incorrect assumption about your own knowledge. Namely, that you think you’ve remembered something when you haven’t.

Tools for Enhancing Metalearning

Smart learners understand the frailty of metalearning. They understand that the feelings of familiarity are often deceptive and they build mental tools to better understand what they know and don’t know more accurately.

The simplest of such tools is self-testing. Doing practice problems, flashcard reviews or even creating your own questions gives you the chance to see past familiarity and get more accurate information about what you know and don’t know.

Another method is self-explanation. The Feynman technique [4], teaching others and reciting explanations out loud are also ways to help see how much you understand about an idea. Articulation is a difficult task, so weaknesses in your knowledge get exposed quickly.

None of these methods are foolproof. The Feynman technique works best when it is ultra specific, because using it on a broader concept allows you to skirt around the parts you don’t understand, only articulating the parts you do. Flashcards are good for pairing associations, but they can divorce knowledge from the context it is used in.

Strong learning strategies use multiple methods that balance each other. When I was doing the MIT Challenge [5], I paired Feynman techniques (which emphasize conceptual insights) and practice problems (which emphasize technical accuracy). Doing only practice problems might have caused me to memorize patterns to particular types of problems, instead of understanding the deeper concept. Doing only Feynmans might have caused me to collect insights but miss the details of actually solving problems.

Learning Faster Means Learning Flexibly

My current language learning experiment [6] has reminded me about the importance of metalearning. Learning well means being able to know what you know, and quickly adapt to fix the weaker points in your ability. This means learning needs to be flexible and there’s no one “superior” strategy that automatically dominates.

Right now, listening, speaking, reading and writing all form tests on my level of Spanish. In each, I get more information about how much I can understand and produce. From that I can direct deliberate studying practice or put emphasis on certain parts of my speech, to improve.

When I first arrived, for example, my main focus was on proper verb conjugation in the present tense. In Spanish, this is particularly important because misconjugating a verb doesn’t just sound funny, it completely changes the meaning of the sentence.

Now, I get verb conjugation right most of the time, so my focus has shifted more on correctly using the different forms of past tense in Spanish—a more difficult task. I’ll keep making this a priority until I’m getting it correct most of the time, at which point I’ll move onto something else (perhaps extending my vocabulary or mastering future/conditional tenses).

It would be a mistake to reduce this strategy down simply to “speak a lot”, “use Anki flashcards” or “listen more.” Because the real difference comes from having a tight feedback loop between metalearning and the actions I take to improve what I know.

How to Use Metalearning to Learn Faster

Many solutions can be reached simply by understanding the problem. Have a poor understanding? Seek out alternative explanations from peers, teachers or the millions of free resources online. Have a poor memory? Make vivid associations or use mnemonics. Lousy at a skill? Practice drills on the parts you’re weak at.

These solutions often become almost trivial once you’ve solved the problem of not knowing what you know. Although many students are precisely aware of the things they don’t understand or remember but still struggle to improve it, they are the minority. For most, the problem is simply pinpointing what you don’t understand or things you’re forgetting.

I recommend that a good chunk of any learning plan be dedicated to improving metalearning. In the MIT Challenge, about 50% of my time was spent doing practice problems. So far, in this challenge, that percentage has been even higher. I spend about 90% of my time practicing speaking, listening, reading or writing and perhaps only 10% on direct study.

The good news is that this burden on metalearning is rarely wasteful. The activities that promote metalearning also help you remember and understand better as well. Doing practice questions was listed as one of the most effective studying techniques [7] in a comprehensive research of various methods.

Be skeptical about what you actually know. Don’t confuse familiarity for recall. Most importantly, spend a large amount of your learning time on tests of your knowledge, so that you can direct the remaining time in an efficient way.