Eight of My Unanswered Questions About the Best Way to Learn

I spend a lot of time thinking about how we learn.

I’ve read hundreds of books and articles on the subject and written scores of essays—and a book!—attempting to summarize and synthesize expert perspectives on the best way to do that. Yet there’s still a ton I don’t know!

Today, instead of offering answers, I’d like to pose some questions I still have. I’ll frame each question in terms of a specific topic, but each likely speaks to a broader issue that impacts other skills and subjects.

My Unanswered Questions

1. When learning a broad subject, like history, how much time should be spent on retrieval practice versus new reading?

I’ve written before about the value of retrieval practice over re-reading. Give students an essay to read, and they learn more when they practice free recall of the essay’s contents than if they re-read it again.

But the research here focuses on re-reading the same essay again. What about reading a different source? If you’re studying a broad topic like history, do you learn more from retrieval activities or reading many other books and sources?

2. Is it better to first watch a foreign language film with subtitles or without?

Say you’re watching a film in a language you are learning. You aren’t to the point where you can understand what’s happening from the first viewing. If you’re going to watch more than once, should you watch with subtitles on the first or second viewing?

The argument for the first viewing would be that you’d understand the plot and thus could better assimilate language on the second viewing. The argument for the second viewing is that you’d attend to the words better and process them for meaning.

3. Should you start by struggling on challenging math problems or simply view the instructions?

This isn’t just my question—it’s a hotly debated area of educational research. Manu Kapur’s productive failure experiments argue in favor of presenting difficult-to-solve problems first. In contrast, others’ experiments on cognitive load theory argue in favor of starting with explicit instructions.

Both theories support problem-solving as you gain expertise, as it provides needed practice. But the question is whether beginners, who are unlikely to know the correct solution beforehand, benefit more from problem-first or examples-first approaches.

4. How many words in a new language should you study via flashcards?

I’m a big fan of learning vocabulary with flashcards. At my peak, I had around 16,000 cards for Mandarin and several thousand for Korean and Macedonian.

Memorizing the central meaning of a word is much faster to do via flashcards than incidental exposure. Yet a flashcard only teaches you part of what you need to know about a word. You can’t get a feel for when the word is likely to show up or what words it goes with, and you need practice to recognize and use it in context.

My hypothesis is that flashcards are useful—but only to a point. Common words are good to learn, as are some uncommon ones. But how far should you go? The most common 100, 1000, or 9000 words? Should you focus on single, core meanings or quiz yourself on variations (e.g., one card for “divide” versus cards for “divide,” “division,” “divisive,” etc.)

Learning a new word incidentally is almost always slower than learning it deliberately. Still, it has the added advantage of the complete context, plus additional practice on the other words in the context. At some point, learning words “in context” will probably become more efficient than rote memorization. The question is where this trade-off point occurs.

5. If you’re learning to ski or snowboard, do you learn faster if you stick to the easy slopes and perfect your technique or take on the steepest slope you can comfortably manage?

For cognitive skills, there’s support for getting the method “right” from the beginning. If you learn a mathematics algorithm incorrectly, there are few opportunities to self-correct that knowledge later.

However, there seem to be different schools of thought for physical skills. One school would argue for perfecting the “correct” technique before moving on. This is the view that sloppy technique is hard to fix later, so it should be mastered before taking on harder challenges.

Another school would argue that once you’ve intellectually understood the technique, you benefit more from practicing in difficult/varied situations rather than trying to perfect it in an easier setting. Yet those complex settings likely to introduce errors and thus lead to long-term headaches.

6. How useful is learning academic computer science for a successful programmer?

This question has been on my mind since my MIT Challenge. On the one hand, knowledge of number theory, analysis of algorithms and logic play a role in programming practice. On the other hand, who has ever sat down to prove the complexity class for a problem before they begin working on it?

This question of the relationship between academic theory and practical skills shows up in many fields. Opinions on academic relevance range from, “Of course, you need to learn that!” to, “Nobody has ever used this on the job.”

Possible answers include:

• It helps, but mostly with advanced work, not entry-level work.
• It helps by making the practical skills “make sense,” assisting with future learning.
• It doesn’t help much with programming, but it helps you land a job.
• It’s less efficient than simply learning programming directly.

7. Do you gain greater chess proficiency by playing slower or faster games?

Say you want to increase your chess score rapidly. Should you focus on longer games, where there is more time to think for each move, or faster games, which have less time to think, but you end up seeing more board positions for each hour spent playing.

I suspect both ends have an extreme where learning benefits break down. Spending a year per game, or spending less than a second per move, is probably not efficient. But where does the optimal point sit in between?

While this question matters for chess, of course, definitively answering it would say a lot about other skills which can be done at varying “speeds” and have trade-offs between depth of analysis and quantity of exposure.

8. If you want to paint in a “loose” style, should you first learn to paint with “tight” control?

Similar to playing chess fast and slow is the question of cultivating an artistic style. Suppose you want to paint in a “loose” or “painterly” fashion. Would it make sense to learn with greater control first and then loosen up as you gain experience, or should you paint relatively loose from the start?

The argument for starting with greater control is that much of what we like in skillful, loose painting styles is that the artist can communicate much with an economy of movement and brushstrokes. Beginners can’t do that, so they’re better off learning the fundamentals and loosening up as they get better.

An analogy for this might be math: An experienced mathematician can simply shout out the answer to a complex question, whereas a beginner would need to do the calculations ploddingly by hand to get that same answer. We would never expect just shouting out an answer to be a way for a beginner to gain the expert’s knowledge.

The argument for starting looser is that the skill of controlled painting is different from looser works. So stylistically, an early focus on control may make it harder to develop a successful, loose style later.

What are Some of Your Unanswered Questions?

Those were some of my unanswered questions—what about yours? What are you unsure about in the process of learning? Do you have any opinions about the questions I’ve posed–share your thoughts!