Last week, I sent out a call for questions to subscribers to my newsletter. I got many good ones, so I decided to split my responses into two separate posts.
Some questions have been lightly edited, and I merged similar questions that a few people asked. Chatting with readers is always fun—hopefully, we can make this a regular feature!
Q: From the bits of work I’ve read of yours, you seem to be an advocate of focused, targeted, and possibly quite isolated (individual study) learning. Also, learning things that generally have answers—technical maths-based subjects, languages, etc.
Have you read Range by David Epstein?
What’s your view on learning through trial and error, experiment, sense-making of existing past experiences that were not targeted but they happen to already be a part of you—(constructing them into your own personal specialism)?
Range is a funny book for me because, in some ways, I’m the epitome of its thesis. I’ve learned a great deal of unrelated things and done a lot of random self-exploration. I write a blog about learning, in part because I’m too fickle to stick to any one thing and master it completely.
That said, I think there are some broad empirical findings from educational and cognitive psychology that we need to keep in mind:
- Students tend to learn better in more guided and structured programs. Despite their appeal, unstructured, discovery-based approaches tend to underperform against more direct teaching methods unless you’re stacking the deck in their favor. That doesn’t mean learning through discovery is useless; simply that the effectiveness of learning through pure discovery doesn’t have the evidence base people often claim it does.
- Transfer between unrelated skills tends to be fairly low. Learning tons of random things might have some benefit, but given the almost complete lack of “far” transfer in carefully controlled studies, I would say that the burden of proof is upon the person claiming transfer is large rather than the reverse.
- Ill-defined skills (I think that’s what you’re talking about when you’re contrasting it to learning things that have an “answer”), I don’t believe are different from well-defined skills. Indeed, a lot of what learning is is making a skill well-defined. If you don’t know anything, the problem space is huge, and you are unsure of tons of possibilities. As you learn more, you find better ways to represent problems and better ways to solve them.
I don’t think there’s anything wrong with being a generalist and learning lots of things. Life’s about more than optimizing for a singular goal.
Q: Do you have any tips or strategies to learn more efficiently from books you read?
I think the most important thing is just to read more. Reading more improves your fluency and background knowledge, which increases your reading rate. People really underrate how much reading a lot improves your reading speed.
Q: What about learning styles? Is being a “visual learner,” etc., a real thing?
Daniel Willingham has a great summary of the evidence against the idea that people have learning styles. To be a true “style,” it should mean that people learn better when information is presented in their preferred modality. Studies generally don’t find increased learning gains from tailoring information in this way.
Q: How many things can we learn at one time? Should we just pick one thing and get to a certain level and then move on to others, or can we learn multiple things at the same time? Does it overload our brain, or is it capable of handling so much new info?
You can definitely learn multiple things in the same timeframe. Most of primary and secondary schooling is based on this. I tend to discourage multiple, concurrent self-directed learning projects because most people overestimate what they can accomplish and end up with several weak projects instead of one effective one. But there’s no cognitive reason preventing it.
I think learning highly similar subjects in tandem can create issues. I’ve never attempted to learn two new languages simultaneously, although I recommend practicing multiple languages in close succession for maintenance.
Q: Do you have any supplements that you swear by for focus and productivity?
Caffeine? Honestly, not really anything.
I’m not convinced about nootropics as something that could plausibly work with no downsides. For nootropics to make sense, you’d have to believe that the brain could work better than it does, but the internal dials dictating its performance are simply set suboptimally. But why would evolution build a brain like this? If turning a dial could increase performance with no downside, evolution would have done so already.
My guess is that the true learning/focus enhancers will either have clear downside risks (e.g., amphetamines), or they’ll be targeted toward a particular deficiency/disability but not effective for the general population. I’d be happy to be proved wrong here, but I’d need to see a lot of good evidence first.
Q: I know some French as I lived in Quebec but left 30 years ago. What is the fastest way to learn it again in preparation for a trip to France in a few months?
Book some tutoring time with iTalki and try chatting again.
For relearning, most of the issue is low fluency and self-confidence. It’s less work since the knowledge is probably already in your head—it’s just not as automatic. A few lessons should help you feel more fluent, which will also help with confidence which can be a big part of speaking a language.
Q: If you had to do the MIT Challenge all over again knowing what you know now about the scientific literature on learning, productivity, etc., what would you have done differently and why?
I initially dreamt up the MIT Challenge as, “I wonder if you could learn all the CS courses MIT teaches?” and then further, “I wonder if I personally could do it in 12 months if I just focused on the exams.” As such, it was somewhat contrived since the aim was passing exams rather than acquiring the skill for a particular purpose. I’ve said as much over the years, but the project would have looked entirely different if my end goal had been landing a programming job, launching a start-up, or becoming a researcher.
In terms of what I know now, there are a few things I can see as weaknesses that could have been remedied with a different approach. I passed a lot of exams by having a decent conceptual understanding. Still, my procedural knowledge of algebra/calculus was weak, which led to me struggling with the physics-based classes with a lot of math to wade through to compute the answer. If I did this project over, I’d do massive drills to build up that fluency, maybe as a prerequisite project to starting the challenge itself.
Similarly, while I did the programming projects, the MIT CS degree was light on actual programming. It was a lot more theory and math. That’s fine, but if I had intended to land a programming gig right after, I’d have spent a couple months going deep on a particular language, ensuring I could fluently implement a lot of ideas I had covered.
Overall, I’m not sure I would have changed much about how I studied. The timeframe constraints didn’t leave much wiggle room to do things wildly differently. But if I had the broader goal of learning programming or computer science, I could imagine a lot of different approaches that might have better results for some purposes.
Q: Why do I enjoy reading nonfiction since most is forgotten? Why do we test high-school students on content “finals” when we know most of the material will be forgotten?
Things are forgotten, but they often still influence us in an impressionistic way. I often find my beliefs have been shifted in a direction even though I no longer recall the exact source of the argument. Some of what reading books does is shift your intuitions about situations in ways that don’t always leave a trace back to their source.
_ _ _
That’s it for this week. Next week, I’ll tackle some more questions including overcoming procrastination, my thoughts on the contentious input hypothesis in language learning, motor skills and more!