JOSH: Hello, and welcome to Clearer Thinking with Spencer Greenberg, the podcast about ideas that matter. I'm Josh Castle, the producer of the podcast, and I'm so glad you've joined us today. In this episode, Spencer speaks with Scott Young about what schools get wrong, learning techniques and technologies, and building confidence as a learner.

SPENCER: Scott, welcome.

SCOTT: It's great to be here.

SPENCER: Many people, most of their experience with learning comes from their school years. So let's start with talking about how learning works in schools, and what it does well and what it does badly. So first of all, what does school get right about how to learn?

SCOTT: I think the main thing that schools get right is that you have someone breaking down a complex skill and teaching it in parts. We often take for granted how difficult it is to learn skills that we've already acquired. Think about learning to read. Most of us here, probably listening to this, already know how to read, but if you think of what has to go on in your brain to read, you have to take all these little black squiggles, immediately recognize the letters, figure out what sound that corresponds to, and blend them together in a word. You have to do this so fast that you can read a couple of words per second. This is a very difficult task that I think most people underestimate how complex it is and why it's often difficult for people to read.

SPENCER: Okay, so they do a good job of breaking down tasks. Anything else you'd point to that you think schools do really well?

SCOTT: Yeah, I think the other thing is that they create an environment focused on learning. In the workplace, for instance, you learn a lot of valuable skills, but the job of the workplace is to get stuff done, not to teach you things. People like apprenticeships, but apprenticeships are often hard because you're in a job environment where they don't want to waste time training you, whereas school is all about spending a lot of time training you. So I think there are a lot of things that schools get right, but obviously there can be some things that they don't get right either.

SPENCER: And in our lives as adults, when we're out of school, there are lots of things we may be interested in learning just for our own sake as lifelong learners, or because we have to do it for our job, et cetera. But much of our idea of what learning looks like comes from school. And so that means if school has problems with the way it teaches, we may then adopt those in our own lives, thinking, "Well, that's how you learn." So let's dig into what schools are getting wrong.

SCOTT: I think some of it is just the format of school itself. You get people, you remove them from the real world, and you teach them a bunch of things that you hope one day might be useful. This can create some kind of drift in the system where you spend a lot of time teaching people things because that's what you're supposed to teach them, and there's no real deep investigation of what should be prioritized and what should be put first. We have a lot of subjects in school that are just taught because those are the kinds of subjects you teach in school. I think that's one advantage of self-education: you can be more directed in choosing things that you're interested in learning that might benefit your life or might teach you some useful skills. School has some of that, but it also has a bunch of other stuff.

SPENCER: A lot of schooling focuses on memorization. I remember in history class in high school, we had to memorize dates of different events, and in our history class in college, we had to memorize different paintings. What do you think about that?

SCOTT: Memorizing is, I think, kind of, there's a bit of a double-edged perspective on it. One perspective on memorization is that a lot of learning is actually just based on memory. As you acquire more knowledge, as you have more knowledge in your head, more stored memories, it makes it easier for you to learn new things. You can have more hooks to put the new knowledge on. A lot of what we consider to be expertise is really just this intense library of patterns you've learned from experience. Chess grandmasters have these libraries of huge amounts of different possible chess positions, and that's what allows them to see a board and quickly make sense of it. Memory is super important for learning. On the other hand, you can learn to memorize things in a way that is kind of detached from its context, where you've just learned a fact and you don't understand why it matters. I definitely think that can be a problem where, if you really focus on memorizing but you don't link that to anything useful, you just end up with a bunch of dates that sit in your head, but they're not applied to anything.

SPENCER: When I think back to my own school, I wouldn't be surprised if I forgot over 99% of the facts I learned in school. And then it raises the question, what was the point of that?

SCOTT: Yeah, there are a lot of critics of education that make exactly that point. That we spend a lot of time in school, and if you do social surveys where you look at the knowledge people actually possess as adults, it's frighteningly bad. And so there are a lot of people who raise that obvious question. Now I do think that the benefits of learning things, even things that you forget, are not so straightforward. Often we forget something that we can't recall. If I just ask you, "What do you remember from history class?" And then you draw a blank. But if I started talking about historical events, you've heard of them before, you know little bits about them, maybe not as much detail as you need to pass the exam that you took when you were in history class, but enough that if, let's say, you read a New York Times article that mentioned World War II, that we fought against the Germans and the Italians and things like that. So I think the idea that you can't recall, at will, all the stuff that you learn from school overstates how bad the problem is of memory. But certainly, given the investment that we put into education, I think it's useful to criticize how little we seem to retain from it in the long run.

SPENCER: If I were redesigning school, starting from scratch, I think there would be some set of facts that I think everyone should know, because there are facts about the world that you don't learn just through direct observation. Because we learn a lot of things just by trial and error or direct observation in our everyday lives. But there are other facts about the way the world works, like maybe the fact that things are made of atoms; that seems like a pretty good fact to know. And there will be some set, but it's probably a much, much smaller set than what is actually taught in schools. And so then the question is, "What do you do with the rest of the time?" I think I would focus much more on skills and on learning ways of thinking. So picking specific skills, like reading, which is one of the most valuable skills, certain basic math skills, certain kinds of critical thinking skills. What do you think about that?

SCOTT: Well, I hate to push back on you on that one, but I actually think almost the opposite. I think there is a lot of knowledge in the world that people don't possess, and if it were taught to them, it would benefit their lives in some way. Now obviously this creates a burden. We already are in school for an incredibly long period of time, so I'm not necessarily suggesting that we do that even longer, that everyone should be in there for 20 or 30 years. But I think my mental model of how the world works is that if you think of all the information that could be useful for understanding things and making decisions, it's probably a much larger amount of information than most people can reasonably learn in their lifetime, which creates this kind of prioritization dilemma. In terms of skills, I think there are quite a few important skills, like you said, reading and basic math are central skills that everyone really should learn. But a lot of other research that's been done on other types of skills has found that they're often knowledge-based. They're often based on knowing more things. The person who is an expert in, let's say, chemistry or computer programming, doesn't have some sort of highly abstract skill of programming; they just know a lot about programming. And that means when they encounter problems, they have all these different pathways to solve those particular problems. The pessimistic take you could have from this is that acquiring all that knowledge takes a long time, and so these general-purpose skills, like reading, tend to be the exception rather than the rule.

SPENCER: There may be a bit of a semantic question here, because if you take something like computer programming, which I learned when I was very young — I started programming when I was 13 — yes, some of it is knowledge, but knowledge broadly construed, when I see this kind of problem, here's a potential type of solution, or here are different data structures I can use. It blends knowledge in a very applied way, where if you just memorize the knowledge, you almost certainly would not be able to do it. It's kind of like with math. In order to solve math problems, yes, you need knowledge, you need to know certain formulas, you need to know certain concepts, but you better practice doing the skill, or you're going to be totally worthless. Nobody can do math just reading about it.

SCOTT: Absolutely. There's a whole branch of cognitive science that looks into how these sorts of complex skills actually form. It's pretty clear that to actually do something with the knowledge you have, not just recite it, you have to practice a kind of procedural knowledge, which is being able to bring out the right knowledge at the right time and do the right thing with it. That requires a lot of practice to get right. Indeed, maybe the practice component is a lot larger than the simply memorizing component, which is why you can be in a language classroom and memorize lists of vocabulary, but then you have to go up to someone and try to speak Spanish, and you're stuttering and halting just because you haven't learned the procedural part to fluency. I definitely agree with you there.

SPENCER: It seems like maybe if we do disagree, that part of the disagreement is around how many useful facts are there to know, or how many useful things are there to memorize when you're young. What are some of the things you'd point to that you think kids maybe should be memorizing that they're not usually in schools?

SCOTT: There are so many subjects that are probably useful that are not part of the typical curriculum, even in high school. We have this whole on-ramp to calculus that takes place in high school. In doing so, we sort of sideline statistics. Maybe you cover some statistics a little bit, or maybe you cover it in college. But arguably, statistics is the one that is more relevant for daily inference in life. People talk about scientific studies, and there's always this news about interpreting trends and all this kind of stuff, where if you had basic intuitions about statistics, you'd be able to make sense of that, whereas basic intuitions about calculus are an important set of analytical tools for going into physics, engineering, and some of these advanced STEM fields. As a practical matter, I definitely use the law of large numbers more than I use knowing how to do the chain rule in calculus. So that would be one example. Another one is economics. I think economics has provided me with more useful mental models than any other discipline I've studied, and it's not usually a mandatory class in school. That's just for me, but I think you could probably come up with a list of some other ones too.

SPENCER: It's funny, so I agree that both basic economics and statistics would be really valuable for people to learn. I was really happy that I learned those, but it's funny because when I think about something like statistics, I feel like it's done in a way that is too focused on memorization. It's often focused on memorizing a whole bunch of statistical tests and when to use different ones, which I think is almost totally wrong. Instead, it should be focused on the key ideas, like large numbers, the central limit theorem, the idea of a p-value, the idea of a probability distribution. To me, that's the core of it. I would actually focus way more on a small number of concepts and really give people an intuition for them and how to use them in practice, rather than memorizing lots of stuff, if that makes sense.

SCOTT: You're probably right. Maybe some of the areas where we're having a seeming difference that may be more apparent than actual is that there is a kind of depth-breadth trade-off when we're learning things. If you're learning something like statistics, you can cover a huge amount of material but understand it at a very superficial level, so that when you encounter situations in real life, the superficial knowledge is just not enough for you to do something profitable in that situation. And I agree with you, I think most of the useful statistics you'd learn would be the kind of thing that, at least in theory, you could cover in an intro statistics class. But to get it to that level of depth, that level of expertise, so that you could see a situation where someone is citing a study and you could be like, "Oh, that's BS, and this is why," requires a lot more work. There is some interesting research showing that people often fail to transfer knowledge they learn in, let's say, math classes to real-life situations. It should be a trivially obvious explanation that adults who learned how to do fractions obviously know how to do them, but when they have to do fractions in real life, they resort to all these kind of cockamamie schemes to avoid actually having to do the math for fractions. They know fractions and can do the math test version of fractions, but they haven't learned it deeply enough that they can quickly and automatically do it in real life. I think statistics is the same way. Even among researchers, we had the whole replication crisis, with people not really understanding a lot of the statistical tests they were employing. They just learned, "Oh, in this situation, you do this statistical test," but maybe not having a deep intuition of whether that's appropriate or what kinds of inferences are safe to make from that. I think that's true even of educated people.

SPENCER: That's a great point that you could spend a huge amount of time memorizing aspects of a thing and not understanding it at a deep level. It reminds me of something Richard Feynman said about students he was with. He realized that they knew lots of formulas, but they didn't really understand what those formulas were about. They could quote all these formulas, but they didn't get the underlying concepts. I think the same thing very often happens in statistics. You memorize a bunch of statistical tests, but you don't really get what it is you're even talking about or where the tests come from. This idea of shallow but broad learning, rather than really deep, fundamental learning.

SCOTT: I think if you talk to anyone who works in an educational space, doing research and this kind of thing, this is the holy grail. We want students to have deep intuitions about hard-to-grasp concepts that are useful for the world. That is what we're trying to achieve, and it just turns out to be very difficult to do. Some of the reason that it's difficult is that deep intuition seems like a holistic thing that you just suddenly get an idea about, but that's probably the last little piece of the puzzle falling into place. It's probably more like a structure that you're slowly building over time. People who understand physics knowledge, this is an area where there's tons of research. You look at physics novices, and they'll get the question right in one situation, and then you slightly change the question, and they get it wrong again. Or they use the physics knowledge in the wrong way, predicting things that even normal people would get right. If they didn't know any physics, they'd be like, "Oh, well, you're not going to make that mistake," but they make that mistake. It takes a long time to get this intuitive expertise. It's built out of a lot of pieces. I think it is the ultimate goal of a lot of fields. There's debate about what's the best way to go about getting it.

SPENCER: I found myself really useful when learning math by looking at the formula and saying, "Why is this true?" Once I know the formula, asking, "Why is it true?" The why question could have lots of different answers. There's not necessarily one unique answer, but it's really about having some intuition for why the formula is the way it is, not some other way. Later, if you forget pieces of it, the why fills in the gap for you; you're essentially able to triangulate it back. Because you're like, "Oh, it must be this way because here's the reason it is the way it is." I think that applies outside of math too. For example, in economics, asking, "Why is it true that when this happens, inflation goes up?" If you're able to understand the why, at least on an intuitive level, then even if you slightly misremember it, you can often fix the formula or re-derive it because you have the why underneath it.

SCOTT: Yeah, I completely agree. A lot of what we're talking about is what it means to understand a formula. We can talk about memorizing a formula, but usually, in the way that we're using it, absolutely memorizing the formula means you know how it's stated verbally or symbolically. But that's about it; that's all you know about the formula. Maybe you've applied it to one or two equations. To really understand a formula, you have to have quite a robust picture from all these different snapshots of how the formula works. You have to see, "Why is this symbol over here? If it goes up, then this goes down. Why does it have to be this way?" People who really understand formulas or mathematics in general intuitively understand this idea of seeing the formula from all these different angles, seeing how you would change it, why would it be this way? Why would it be that way? People who have difficulty doing that, or who don't gravitate towards doing that, may memorize it. But because they don't have all these different pictures, that knowledge is very fragile. If you get into a situation where the formula needs to change slightly, maybe even just a very superficial change, you can't handle that adjustment because all they have is just the text of the formula. They don't have any deeper intuition or deeper representation of what it implies. Again, getting to that deeper understanding of a formula is very important if you're going to make math useful for real life.

SPENCER: I think you touched on something very important, which is that ideas can often be understood from different angles, and there's often no final understanding. You can just keep getting a deeper understanding. Math illustrates this really well. Take the exponential function; you can know what it looks like on a plot, and that's something to know about it. You can also know important facts about it, like that the exponential is its own derivative. You can also know what it can be used for, like modeling compound interest. Each of these is a partial understanding, but there's really an infinite understanding of the exponential function. That's just one function. Each of them deepens your perspective about what this thing really is.

SCOTT: Yeah, and I think this is exactly the point. Getting an understanding of understanding itself: what are we doing when we understand? I think the science here is not firmly established. I don't think anyone would claim that we have a robust understanding of understanding itself, but there are a lot of interesting ideas. Some pieces that have come up from this research suggest that one: understanding is probably not a unified thing. It may feel like you get it all at once, but it's probably lots of little bits and pieces of knowledge that assemble together. The second idea is related to this mental picture: you have some sort of mental model of what you're trying to think about. If you can form this kind of model in your head, you're able to pick it up, turn it around, look at it, and inspect it from different perspectives, metaphorically speaking. That's very valuable. It's certainly a much more complicated thing in your head than just the text of the formula. To go back to our educational question, this is one of the driving questions about a lot of educational ideas: how do you get someone who doesn't have that in their head to the place where they do have that in their head? What is the most efficient path to go from A to B? I think it can be challenging at times.

SPENCER: Something you touched on a little earlier, which I think is really important, is this idea of transfer of learning and the challenge of it. You want to tell us, what is transfer of learning and what is the evidence on it?

SCOTT: Yeah, so this is a huge topic. It's pretty age-old; it goes back to the turn of the 20th century. Edward Thorndike was doing investigations and found that a lot of the supposed reasons for teaching Latin in school were that it would help with memory or teaching geometry because it would improve reasoning ability. These justifications were frequently found to be wanting; teaching these particular subjects didn't do any better than teaching ordinary subjects like bookkeeping. The basic idea is that this vision of transfer is really a question of how the mind works and how learning works. I think the research pretty strongly favors this bits and pieces idea that when we have a skill, it's formed out of lots of little pieces. You can transfer abilities you learn in one area to another through a variety of ways. There can be actual overlap; for example, when you think about people who studied physics in university and then go work on Wall Street. They can make that transition because a lot of the math is the same. You learn a lot of advanced calculus and statistics and all these kinds of techniques, and it turns out that they are the same ones you might want to use in finance. So you just have less to learn. In other cases, it can be a little more elaborate, like if you have analogies or metaphors, you can find some way to use something you already know to learn something else. The basic idea is that it's these bits and pieces that are what learning is really made of. This is in contrast to the idea that the mind is like a muscle. If you train your critical thinking abilities in one area, it's going to make you a better critical thinker in another area. I think that's only true to the extent that the way you're thinking critically in that very specific way happens to be exactly the same, which is not usually the case.

SPENCER: Would you say that the literature is quite pessimistic that people just generally can't transfer successfully? What do you make of that?

SCOTT: There are certainly elements of pessimism within the literature. There's a whole kind of cottage industry of people being surprised that transfer doesn't happen in certain situations. But I wouldn't want to paint it with too broad a brush. I think it's a question of what your baseline is: how much transfer do you think should happen? If you have the viewpoint, I would say the really naive viewpoint that when I practice Sudoku puzzles, that's making me a more critical thinker, and then I should go to my programming job or my finance interview and crush it because I've been doing all those Sudoku puzzles, the research is pretty pessimistic on that version of events. On the other hand, there's also an opposite view, which is that as you learn more things, it's easier to learn more things. There are a lot of views that if you have a lot of prior knowledge in a domain, on a lot of tests of reading comprehension and retention of information, you do much better, even if it's new information. That's an example where you've learned something, and it makes it easier for you to learn things. I guess it depends on how you define it, but I would say that if your viewpoint is this somewhat naive perspective that, "I'm going to do brain training games, for instance, and it's going to make me generally smarter," that's something that I think the research looks at pretty pessimistically.

SPENCER: In your book UltraLearning, you've got the set of principles, and one of them is directness. And I think that might be related here. Do you want to tell us what that is?

SCOTT: Yeah. So directness is this idea that if you're trying to learn something because of this specificity of skills, you're going to be more efficient if you try to figure out what the actual practice you're trying to get good at is and align the efforts you're making with that kind of practice. So for instance, if you're learning another language and your goal is to have basic conversations because you're traveling, then practicing those basic conversations as if you're traveling is going to be a much more efficient way of learning that language for that purpose than, say, reading comic books or doing something that's sort of on topic but not quite related. This is important because often the subject we're tackling is really broad and vast, and if we're non-specific in our efforts, it can take an enormous amount of time to reach the broad coverage of the entire area when there's just one particular thing we happen to be good at. So I think it's a good thing to keep in mind that if you want to train or improve in any particular kind of thing, doing practice that's aligned with that is going to be very helpful.

SPENCER: It seems like it's a challenge of school as well because very often school is teaching you in a context that's devoid of the final way you're going to use the information if you're going to use it at all. I've talked to lawyers about law school, and they often say, "Yeah, maybe I use 10 or 20% of what I learned in law school," because they don't know what type of lawyer you're going to be or even if you're going to end up being a lawyer. So you're going to take all these different classes that are sort of detached from how you're going to be using things when you're actually a lawyer.

SCOTT: I have some mixed feelings about that. On the one hand, I think you're definitely right. School is a wasteful process if you consider the end outcome of when you use the knowledge. Most of us do relatively specialized jobs, so we're only using a very small subset of the academic knowledge that we may have learned that might be relevant to that job, including, again, you're a lawyer, but you're only this kind of lawyer. So there are all those law classes that you just don't need for your particular practice of law. And then even just in daily life, there are so many facts and pieces of knowledge that may not come up and be that useful to you. There is this general inefficiency with general training because if you're trying to prepare someone for a vague, amorphous set of future tasks, then you have to teach them lots and lots of things, and most of those things will not be useful. That's true. The reverse side of that, and I think the side that's worth pointing out, is that if you don't know what specialty someone's going to be, and then you train them in a particular specialty and it turns out not to be the one they want, then you've kind of really set them up for failure. I know there was this example of a bunch of people going into petroleum engineering studies when that was a booming business, and the people who had those training qualifications were all getting nice six-figure jobs out of college, and then there was a change in oil prices, and those jobs shrank. If you really studied petroleum engineering, you were kind of in a bad place because you invested in a skill that wasn't useful in that economy. I think there is this kind of trade-off between learning useless things because sometimes they might be useful, and you have to hedge your bets, versus if you spend all your time learning things that are not useful and not directed towards your particular goal, then that's a really inefficient way to reach a specific destination.

SPENCER: Of course, if you go to law school, you're kind of subject to whatever the law school teaches you, but what you're saying suggests that it's important to figure out what we really want to learn, or why we want to learn it. The earlier we can figure that out, the more efficient we can be in the learning process.

SCOTT: I think for a particular project, that's how I like to think about it. So when I am embarking on a new skill or a new subject, I've tried to be very mindful of what the end use case for this knowledge is. So even if I'm learning a subject, let's say I'm learning about nutrition, I want to learn more about healthy eating or something, and I'm thinking, "Well, I want to know this to make practical differences in my health advice, and that's going to affect how I study things, which resources I focus on," versus, "I'm training to become an epidemiologist, and I need to know how to run studies because I'm going to do a PhD." I mean, the subsets of knowledge overlap, but they're often different in many ways too. The things that I need to know in the first case maybe are idiosyncratic behavioral things, like, "Okay, this is how I can cook healthy food at home," whereas the person who's running some kind of scientific study, they don't even know anything about that; they just need to know about doing statistical tests and all this kind of stuff. So I think again, the more you can view knowledge and learning as all these fine-grain little details that have to accumulate to become general abilities and skills, the more you can often make wise choices for projects in the ways you study and which resources you use that really do get you to your results that you're after a lot quicker.

SPENCER: Does this connect to your principle of meta-learning?

SCOTT: If you know what a subject is and how it breaks down, then you have a lot better time choosing those resources and navigating. There is a bit of this bootstrapping problem that if you go to a completely new subject that you know nothing about, it's very hard to make wise choices about which resources to use. But I think that's one of the reasons why we can leverage so heavily people who actually know a field or know a subject. If you talk to someone who's already got the job that you want, for instance, they can tell you a lot about what kind of skills they're actually using on the job, and that if you got good at them, that would help you in the interview process. Or, for instance, if you're trying to learn about a particular subject, and then you talk to an expert in it and you're saying, "Okay, well, which books would you recommend, or which sorts of ideas should I look into?" You can get some starting points for research and help you avoid maybe all the garbage that's floating around that's going to deter you and sideline you from what's actually helpful.

SPENCER: So what is the principle of meta learning more generally?

SCOTT: Yeah, so the principle of meta-learning is this idea that you should form a map of what you're trying to learn before you actually go out and learn it. And the map doesn't have to be extremely detailed. But the idea is that most people don't do much of that at all; they just start doing whatever they're trying to learn. They start reading things, and that's not necessarily a bad idea. But if you're trying to be more precise or efficient about it, having some sense of, "Well, this is what's involved in getting good at this skill," or "this is what's involved in really understanding this subject," can help you make those smart decisions ahead of time, especially for things that might not be obvious at first. I think people who have a lot of expertise in a subject often will point out, "Oh, well, it's a beginner's mistake to focus a lot on this when you should really be learning this other thing," or "Make sure you get this right, or it'll create bad habits later." I know a lot of piano teachers, for instance; they prefer to teach students who have never played the piano than people who are self-taught, because the self-taught people have all these bad ways of playing the piano with their fingers that are going to lead to these bad habits in the long run. If you could talk to the piano teacher and be like, "What are the bad habits that people fall into when they're self-taught?" based on the position they're holding their hands or how they like to do things, then you can consciously avoid those things when you're trying to practice even on your own.

[promo]

SPENCER: Could you give us an example of a skill that you learn and kind of how you approach it sort of end to end?

SCOTT: Yeah. So one that I've approached a few times is language learning. I think languages are kind of a unique case because the knowledge in each language is totally different, but the structure of the knowledge is very similar. Usually when you learn math, for instance, and you learn one branch of math, it doesn't really mean anything about another branch of math; they're going to be kind of different from each other. But languages all have words, grammar, and pronunciation. The specifics are different, but the structure of it is very similar. One of the things that I often do when I'm thinking about a language is, "Okay, what are the thousand words that would give me that kind of bootstrap vocabulary so that I could have easy conversations?" That's going to be a mixture of the most common words and things that are relevant to my particular situation or the things that I want to talk about, like my profession, where I live, and topics of interest to me. Making flashcards of those and learning those, you can quickly get to a point where maybe you could have some conversations. If someone just dumped words on you, you're going to quickly run into these little situations where you want to say something, but you're not going to be able to do that. That's an area where I think meta learning can be helpful, where you sort of map out, "Okay, these are the thousand words I'm going to learn. These are the phrases that I want to be able to say," before you even start putting in the work of practicing and memorizing them.

SPENCER: Got it. So once you figure out the basic set of vocabulary, how do you approach it?

SCOTT: In this particular case, a lot of it is okay. Once I know what kind of knowledge it is, then that suggests some techniques you can use to memorize it. For vocabulary, there's a type of software called spaced repetition systems, which is particularly valuable. It's basically flashcards, but they queue them up for you so that you can deal with thousands of flashcards without it being just this chaotic mess of little pieces of paper. Also, mnemonics can be very useful. There's a number of mnemonics for learning vocabulary words, in particular the keyword mnemonic where you think of what the foreign word sounds like. You turn that into a picture, and then you link that picture with the English equivalent. It doesn't work for all languages. Some languages have phonologies that are so different that it can be tricky, but definitely for European languages, I've used this heavily, and it lets you memorize the words a bit more quickly. So I've mapped out the sort of, "These are a thousand words I want to learn. Then I figured out, okay, this is the way I'm going to memorize it." Then you're just off to the races. It doesn't actually take that long to memorize, let's say, thousand words, and that's the building blocks for conversations. There's more than that in learning a language, but that's just an example of a way that you can efficiently get from point A to point B, point B being having conversations with people.

SPENCER: For those interested in spaced repetition, we actually have a free app you can use called Thought Saver, where you can create daily routines, and they can include spaced repetition in them as well, if anyone wants to check that out. So regarding language learning, let's say you've memorized these thousand words. You're just on the first step to learning the language. So what's the next step to be really efficient about it?

SCOTT: Yeah. We can go back to this directness idea. If my goal with the language is primarily conversation, then I need to practice conversations. There are lots of services where you can find tutors online, and you can practice with people. I mean, it's still in that kind of shaky territory, but I wouldn't have even been able to suggest this two years ago. Things like ChatGPT, I know people who are doing conversation practice with ChatGPT now, because you can actually just talk to it in the language and it will respond back to you. It's still a little bit iffy. There's a bit of delay. Sometimes it has a hard time recognizing heavy foreign accents if you're learning a new language. But these are options for practicing, because one of the major problems I think with school-based language learning is that it's very easy for the teacher to stand at the stage and give you instructions, and it's sort of annoying, or maybe doesn't always fit the format to have a lot of unstructured practice time. School-based language lessons tend to cover a lot of material but not provide that much practice opportunity.

SPENCER: What about full immersion? People often recommend that as a way to learn languages. Do you think that's what people should do?

SCOTT: Yeah. I think full immersion is great. The challenge is that sometimes full immersion is hard to achieve, and so people use that as a reason not to do it. I'm a big fan of full immersion. I did a project about a decade ago with a friend where we went to four different countries, and the strategy was very much that every time we would go in the country, it was three months each, we would just speak in the language of the country with each other and everyone that we met, and it worked pretty well. I'm not going to say that we were perfect after three months, but you can watch videos of us; we made friends and did all the normal things that you would do in Spain and China and some of these places. The advantage of immersion is that it creates this near-constant practice opportunity. If you go full in on it and you're actually speaking the language all the time, you're getting an amount of practice that would take you a decade in a classroom in a matter of several months. That can make a huge difference. I don't think there's something special about immersion, per se. It's just that you are getting a lot of practice.

SPENCER: Presumably, though, it has at least two other advantages. One is that it's super direct; you're actually learning the things you need to learn. You want to talk about X, well, you got to learn how to talk about X. So it's very direct in that sense. Second, it's very motivating. Because you have a really good reason to do it. Every day you actually want to say something, and so you want to learn how to say it.

SCOTT: Absolutely. I think those are both really good reasons for it. The only reason I am somewhat cautious about suggesting it is that I think some people see it as a magic pill, particularly living in the country as a magic pill to learn their language. They underestimate that even living in the country, you're still you, and if you were finding it hard to practice in North America, then you go to some other country, you're not necessarily automatically going to be speaking the language all the time because you're still going to have the same social difficulties or reluctance to doing it. That's why I encourage, even if you can't do immersion, get that tutoring session, get that kind of practice in, because that behavior is going to be the same in both contexts. I know lots of people who have spent decades overseas, and their mastery of the language of the country they live in is quite poor. You can live there and not be in immersion. It's important to recognize that it's an opportunity, but you also have to be motivated and very serious about pursuing it as well.

SPENCER: A lot of people use learning apps, especially Duolingo, which is really popular. What do you think about that as we learn languages?

SCOTT: Duolingo is a bit of a moving target, so I'm often a little reluctant to continue to talk trash about Duolingo when I haven't actually opened the app in five years, because I think it's changed quite a bit since I first dismissed it in UltraLearning. My beef with Duolingo is essentially this indirectness problem that it gets you to do a lot of language-themed games, and to me, a lot of those games seem kind of only questionably related to language use. I don't think it's a total waste of time. You'll probably learn some words, you'll probably learn to recognize some language and to comprehend some language, but it seems to be kind of ill-equipped to deal with the sort of spontaneous speaking situations. When I was using the app, and again, this is admittedly years ago, one of the games that it got you to do was this drag-and-drop sort of sentence completion. I want to say the sentence was in English, and then you had to drag and drop from a word bank. There were like 15 words in the bank and seven words in the sentence, and you had to put them in the right order to make the sentence in, let's say, Italian. It's funny because while this is a language-themed game, it's pretty clear this is quite distinct from what you have to do when you are creating a sentence in your head in Italian. If you're still learning Italian, there's no word bank. You have to remember the word from memory. You have to say it in real time. You have to be able to do this recall fast enough to get it correct. You don't have the advantage of knowing it's either this or this. There could be an infinite number of things that it could be, including words that maybe you've never learned before. There are all of these dissimilarities between that particular task and what you would have to do in an equivalent situation in real life. I don't think it's particularly surprising when people tell me they've spent nine months on Duolingo and they can't have a basic conversation. I don't think the knowledge is necessarily a waste of time, but I think it's important to keep in mind because sometimes the promise is that you're going to do this thing for five minutes a day, and then if you just stick with it, after a year, you're going to be speaking French or what have you. I think sometimes that's wishful thinking on behalf of the app users and maybe the people at Duolingo, too.

SPENCER: I think with language learning, it's pretty common for people to use flashcards; that's a pretty well-accepted way to learn vocabulary. But there are also a lot of people, more broadly, in school, for instance, that the way they learn material and get it into memory is they just kind of review it. They'll reread that part of the textbook. What is known about reviewing material versus other ways of learning material?

SCOTT: I think one of the things that's important to point out is that when we're doing a flashcard, we're not just looking at it again. The whole idea of the flashcard is that you've got one side, and then on the other side, you've got the answer. This is a key distinction because what you're doing when you do a flashcard is trying to actively recall the word from memory. You can note how that's different from this word bank example I gave from Duolingo. If I have to remember the word for water in Spanish is agua, and I just see water, and I'm searching my brain for Spanish water, and then agua, that is a different mental activity than if I see a word bank, one of which says agua, and there are three other options that are obviously not water. I'm like, "Oh yeah, it was this one." It's a different activity. A lot of times when students are studying with review, they're kind of doing the version of that latter thing. They're looking through their notes, and they're like, "Oh yeah, I know this. Oh yeah, it's this thing, and oh yeah, it's that thing." Then they go to the test, and what they actually have to do on the test is not that. What they have to do on the test is go through their brain, find the correct answer, and produce it perfectly from memory. That turns out to be not only harder but not particularly well supported by that kind of practice. People who do retrieval practice, which is the generalized version of what you do when you do flashcards, have much better performance on tests, typically, than those who use this repeated review strategy. I think that's one of the interesting things from the research: a lot of students spend a lot of time studying in ways that are quite ineffective.

SPENCER: My understanding is that not only do people use these ineffective methods, but their intuition about what's effective is almost exactly backwards. Is that right?

SCOTT: Yeah. So one of these studies was by Karpicke and Blunt, and they had students study in four different ways. There were people who read the text once, people who read it repeatedly, people who did a concept map, which is kind of like a mind map, where you draw a circle and try to create a branching structure of the ideas, and people who did free recall, which means you shut the text and have a blank piece of paper, and you try to remember everything. It's an even more basic version than flashcards because there are no questions. You just try to remember everything you can. They did two things. First, after this part, they asked the students how well they thought they learned the information, and the people who did repeated review, meaning that they read the text multiple times in the allotted time, thought they learned it the best, whereas the people who did free recall, and if you've ever done free recall, it's extremely hard. You feel like you've learned it really badly; you can't remember anything. I don't remember anything from that text. But when they actually tested the students and saw how well they performed, the free recall group did the best, and the repeated review group did worse than the free recall group. I think this is a very interesting example because a lot of students gravitate toward this repeated review strategy. When you repeatedly see something, it feels more and more fluent to you. It feels easier to read it again and again. That's convincing you that you know it, but really, you're getting good at this other skill, which is just recognizing it, and you're not practicing as much this retrieval ability, which is so important for actually performing on the memory test.

SPENCER: Based on this idea, would you recommend that when people read an article, they try to reconstruct the main ideas themselves right after they've read it?

SCOTT: Yeah. It's hard to know exactly what to recommend because there's this sort of trade-off point. What we're comparing here is someone who is reading something again and again, which is not typical outside of studying situations. Usually, people only read things once. So there is a question about whether, if you're deciding to do a lot of study on a particular article, that's taking away time from reading other things, and that's sort of an interesting question. I don't have an exact answer to that, but I would say that if you really want to know something, either because it's for a test or because you've read a book and think it's particularly important or life-changing, inserting a little bit of this retrieval practice is probably a good idea. I remember when I was doing research for this UltraLearning book, and I knew I was going to have to write about this stuff later. After I would read one of these academic papers, I would do the free recall. I would insert a loose-leaf note in the binder that I was writing in and just try to remember everything I could from the paper. I think it helped quite a bit, not only in making the notes but in actually forming some of those memories. The amount to which you do this kind of retrieval practice depends on your situation, but it is something that I think is underused as a strategy for learning.

SPENCER: My suspicion is that the way we read is incredibly inefficient. If we're reading to learn, obviously you could read for fun or just to stay up to date on the newest thing, but if you're reading to learn, I suspect that if we spent just a little bit of time, five minutes after each article we read, just trying to summarize it in our own mind, reconstruct it in a notepad, that we would actually get huge gains in efficiency in our total learning. I'm wondering if you agree with that.

SCOTT: I think you're right. Again, it's this sort of trade-off point, because let's say you have ten articles to read, and they all have some overlapping information and some new information. You could either read all ten articles, or you could read five of them and do some kind of retrieval practice. Where's the balancing point? I don't actually know where the balancing point is, but I think it is an interesting question. I would say that for me, the idea of making reading more efficient is that if you are reading for a particular purpose, and one that's not really far away, it's one that you're continually engaging in, I do think that improves the value of your reading. This is a bit anecdotal, but the fact that I write about what I'm reading often, or I'm writing a book about what I'm reading, it just forces you to process the text differently. You're very purposeful. You pay attention to things you wouldn't have paid attention to, you reread sections that you realize your mind was wandering. I think this kind of purposeful reading, particularly when it's combined with some kind of activity, like, "Okay, I'm reading this book, and I have to write a book review on it later, or I'm going to write the book review as I'm going to make these notes." That can be very powerful. I think what I would prescribe more generally, even though it's not as quantitative, is just this purposeful reading. Having reading projects, having this sort of, "Okay, I'm going to read a bunch of books on this with the goal of doing why, and keeping that goal in mind, doing this kind of updating my knowledge activity." I think those readers are going to be a lot more successful than someone who's just, "Ah I got an audible credit, and I'm just listening. Sometimes I'm not even paying attention."

SPENCER: I'm a bit surprised that you're not sure about the trade-off between spending time reconstructing it versus reading. Because to me, it seems like you can actually do the reconstruction really quickly. You could spend even just two minutes trying to jot down what you thought the most important points were in something, and that can improve your retrieval in the future a lot and how much you get out of it. It doesn't have to be a time-consuming thing. It could just be 10% of the overall learning time and get a big benefit. Do you disagree with that?

SCOTT: No, I think you're probably right. I'm just expressing some agnosticism about where the trade-off point is, but yeah, 10% is probably not that much. One of the things I was doing — again, this is a practice I've started since researching my latest book — was every time I would read a book, I would begin this sort of book notes — I call it notes and quotes — I would just begin with my own remembered summary of the book. Then I would go through and look at all my highlights and choose which ones I want to save in this compressed Word document. That Word document became the thing that was easy to reference if, six months later, I had to be like, "Okay, there was some study, and I want to talk about it. I know it was in this book, but I don't want to look through 400 pages." I think this kind of compression activity was very valuable. The trade-off point is that you can go arbitrarily far in this direction; you can make flashcards for literally every sentence you encounter in the book. To me, that's probably going too far in the other direction. I don't know of any strong research on balancing reading more versus reading better the things that you've already read. But yeah, you're probably right, 10% more is more than likely in the direction of better knowledge.

SPENCER: What about when we're trying to learn a skill? Let's say, play the piano or martial arts or something like that. What do you think the best way to approach learning is if you're not just doing it for fun, but actually want to get good at it?

SCOTT: Yeah. Well, again, going back to this idea we talked about, meta-learning, the starting point for me learning the piano, because I don't play the piano, would be to talk to piano teachers. What are the approaches to teaching piano? What are the different ways that people learn the piano? What are the pitfalls? What are the problems that you see with students who have plateaued early? There's a lot of knowledge there. There's a lot of meta-knowledge about how piano playing works, how the skill is acquired, and who is good at it. This is true of piano playing, but it's true of any domain; there is incredible lore in any field about how it's acquired, what matters, and what doesn't matter. Not all of it is going to be 100% accurate, but I think that hard-won knowledge from teachers and people who are very proficient in a skill is very valuable for a beginner starting out. So that's where I would start. In terms of practice, I think I would figure out that there's a mixture of basics that you need to acquire correctly in order to learn the skill. Then there's the specific piano playing you want to accomplish, whether you want to play Chopin, do some jazz improvisation, or whatever you want to do. That's going to be a little bit more choose-your-own-adventure. There is a balancing act. A lot of teachers and schools have a kind of blanket approach that everyone needs to learn these basics, but the self-education approach I advocate for is a little more bespoke. Clearly, you need to have both. You're not going to be able to learn Spanish if you don't learn the basic conjugations. But maybe you don't have to go all the way to some schools where they devote one week to each conjugation and that's all they teach you. I think understanding those two elements is very important.

SPENCER: Everyone's heard this phrase, "practice makes perfect." Do you think there's a lot of truth in that, or would you dispute it?

SCOTT: I think, well, practice makes you more fluent, and fluent is not always better. This is sort of a half-truth, which is why it sticks around. As you practice something, the skill becomes more automatic and effortless. That's clearly an important component of proficiency for many skills. If you can't fluently decode the letters that are written in text, you can't read; that's an important part of learning to read. Similarly, if you have to figure out which pedal is for the brake and which one's for the gas whenever you're driving a car, you're not winning any NASCAR races with that. At the same time, if you get really fluent at a suboptimal method or a bad way of doing something, then practice can actually be negative. It can inhibit you from learning the actual skill because it becomes this easy alternative to doing it the proper way. That's why, when you're practicing, you need a lot of practice to learn skills, but it's important to have awareness of best practices and the right ways to do skills. It is easy to get caught in the trap of thinking, "Well, this works, but it's not the best way." My best example of that is learning to type on a keyboard. How many people hunt and peck with two fingers? That's just because you started doing that, and it got easier and easier. The correct approach of having all four fingers on each hand on the home row and typing each letter like that is much faster, but it's not obvious from the beginning that it works better. You typically have to go to typing school or deliberately practice that approach. Figuring out what the equivalent to typing with the home row is for the skill you're learning is a very important investment to make early on.

SPENCER: I think this comes up a lot in skills like martial arts as well, where, let's say, one day, for whatever reason, you throw your right hook in a kind of funny way. Well then you can get a groove of doing that, of throwing it the wrong way, and now your practice is actually practicing and doing it wrong, and then it can be really hard to get out of that groove once you've thrown that punch hundreds of times.

SCOTT: Yeah, and I think motor skills are particularly susceptible to this because they often take place at a level that's very distant from our conscious awareness. So the parts of the brain that determine how much certain muscle fibers contract and in which order are very removed from your conscious understanding of the skill that you're executing. That lack of conscious awareness can make it very hard to step back and do it properly. There's a lot of research interestingly showing that if you make people more aware of how their body is moving when they're doing a certain skill, you can actually make them learn it more poorly. Your own conscious understanding of, "Oh, I'm tightening this muscle," and then it ruins your performance. Sports teachers and coaches often have to use clever ways to get you to do something in not the way you automatically want to do it, without making you consciously aware of doing it a different way, sort of tricking you. I know one example of a friend who played squash quite a bit, and they really encouraged people to hit the ball in the center of the racket. The way they got the students to do that is they gave them rackets that had really small heads so you could only hit it in the center, because otherwise you wouldn't hit the ball at all. That's an example of using a constraint to force a particular motor program.

SPENCER: I like that example. That's cool.

[promo]

SPENCER: This connects to the idea of getting feedback, which is another one of your learning principles. So what's the right way to get feedback and why is it so essential for learning?

SCOTT: Feedback is obviously essential for learning. We very rarely get things perfectly right the first time. So we need some signal from the environment about how to adjust our progress. It's interesting because the research on feedback is really mixed. It's not the case that just more feedback is always better. In a classroom environment, I think it's something like over a third of the studies done in one of these meta-analyses found that feedback actually made things worse. Sometimes this can be because the feedback is not useful. When you tell someone,

Oh, you're so smart or you're no good at this," you're not really giving them any information they can use. Instead, you're just giving them this kind of motivational evaluative judgment that is often negative. If you tell someone, "Oh, you're so smart or you're so good at this," basically, I've told you to stop trying so hard. Similarly, if I just tell you, "Oh, you're bad at this, you should just give up." Unless giving up is what's in your best interest, it's not helpful for you in acquiring the skill. One of the things we know about giving people feedback and verbal feedback is that you want it to be task-specific and not evaluative of the person. I think this is true even when you're seeking out feedback. I know a lot of people, sometimes because I talk about feedback, they'll say to me, I started a blog and tell me what you think of this kind of thing. I'm always very reluctant to give these people feedback because either you are being pandering and not really being honest, or you're being overly critical when this person sought out your support. If you're seeking feedback from other people, you need to understand that social reality, that people are not going to give you honest feedback. It's better to avoid that evaluative part entirely and just be like, "I'm doing this right now. What do you think I could do to improve?" All of a sudden, the suggestions come out, and they're task-specific, and they're actually information you can use. That's an example of where getting the right kind of feedback really matters.

SPENCER: Some feedback, as you point out, is just sort of, "Oh, you're doing well or badly," but other feedback actually tells us what we're doing wrong, which seems a lot more useful. Would you agree?

SCOTT: If you can get corrective feedback, like, you're making this mistake, do this better, that's helpful. Unfortunately, in a lot of the situations that we encounter, getting that corrective feedback is hard or non-existent. I think a lot of us learn skills where that signal is weak or absent. You talked about learning programming when you were quite young. How many times did you try compiling something and then you just get some cryptic error, and then you spend a while trying to figure out what it means? This kind of thing is part of skill learning; sometimes having not the best feedback signal and still nonetheless trying to figure it out.

SPENCER: When I think about learning a skill that I don't know, like piano, I have no idea how to play the piano. What I imagine is trying to identify your areas of weakness and then coming up with drills to practice them, where during the drill you get rapid feedback. So let's say, for example, you struggle with chords. You could have a drill where you practice just doing chords, but in a way where you can tell right away whether you're hitting the chord right or not. Then you practice that to hone that component, and then you look for your next area of weakness to design a drill for that. What do you think of that approach?

SCOTT: Yeah, I think drills are very important, and I think they're clearly a part of learning complicated skills. There are lots of different ways you can approach drills. What you're talking about is kind of the classic view of drills, where you're working on chords or certain difficult combinations in a particular piece you're trying to play on the piano. I can think about layup drills or three-pointer drills, where you just do them repetitively. But drills, I think, have a more general sense; it's about trying to focus your attention on an aspect of your performance that might otherwise be neglected. Even as a writer, for instance, you can write a piece where the thing I'm focusing on right now is my storytelling or writing a good headline, or making the prose funny or readable. That's out of the universe of qualities that you're trying to improve in all of the writing you're doing. You're focusing on one of those, and that deliberate attention to a specific aspect, even without particular feedback for that, just the fact that you're attending to that aspect of your performance lets you use that mental bandwidth on that particular feature. I think that's a very important way to think about learning in general: what's one specific thing that I could try to improve this time around?

SPENCER: That's something I noticed in my very amateur martial arts practice. It really helps to have one thing in mind whenever I'm practicing, like, "Okay, right now I'm focused on my footwork and standing in the right position when I do my attack," or, "Right now I'm focused on making sure I'm protecting my face as I'm punching." I find that if I have more than one thing, it's actually very easy to not improve at anything. If I have nothing I'm thinking about, then I often feel like I'm not really making progress.

SCOTT: I think this is another piece of coaching lore that's pretty true. The advice that if you're the coach and you're seeing someone's performance and you think they need to correct something, giving them a complicated picture of their skill is really counterproductive. You have to give extremely simple commands to focus on. This isn't because the athlete is unintelligent; performing the skill itself requires quite a bit of bandwidth. Unless it's so over-learned that they can do it in their sleep, they're thinking about lots of other things just to execute the skill, so you have very little leftover room to work on these other components. That's why I think drills and these particular focuses for improvement need to be so concentrated. If you give people three things to focus on in their golf swing, they're doing none of them.

SPENCER: At that point, what's the right challenge level to practice a skill? We could practice at an easy level, where we can get it right almost 100% of the time. We could practice at a really hard level, where we screw it up most of the time.

SCOTT: My feeling is that for most beginners, when they're learning a skill, the difficulty is usually set too high. You would be more successful if you were closer to, let's say, 80 to 90% success on a particular thing, however success is defined. There's some interesting research on school children, where they found that teachers who get about 80% right when they ask for students' responses tend to have better educational outcomes than the more typical 60 to 70%. I found this interesting paper that was talking about this in the context of machine learning. They also found that about 80% correct decisions were sort of the optimal point for learning in this particular case. I don't want to say that 80% is the right rule for everything, but it does give a qualitative sense that in the beginning, difficulty is usually higher than that. A lot of skills we're learning start at 0% success, so finding ways to modify it, break it down, and make it simpler to get to that 80% or in that mostly successful range is probably beneficial in the long run. In the long run of the skill, you're often trying to improve a well-learned skill, which means you might have to fail more often at a task you're familiar with because you're pushing yourself in this deliberate practice zone. I don't know of any specific training success rate in that literature, but it does seem qualitatively that in the beginning, learning is usually too hard, and in the end, learning, counterintuitively, is often too easy, or the skill is too easy, rather.

SPENCER: People don't necessarily connect learning to fear and anxiety. But if you take someone who's, let's say, learning math, you can imagine that anxiety could be a big factor that prohibits learning or learning a language. If you're afraid to go out and actually have conversations with people in another country, that could really inhibit learning. So yeah, how do those two ideas connect?

SCOTT: Yeah. It's funny, because in talking about learning with people, so much of the research is about memory and cognition and reasoning, but if you talk to people's actual struggles with learning, they're all emotional. It's all about motivation. It's all about being afraid of this, or not feeling confident, or all these sorts of things. So there is this sign of disconnect sometimes when you talk about things from a perspective of flashcards, when what really stops them from speaking Spanish is that they're afraid of looking stupid or something like that. And so what's interesting from the research is that there is this powerful circuitry in our brain that's largely unconscious, that detects threats, and these threats are largely hardwired. So, we have fears of spiders and snakes, but also things like heights and public speaking. And these fears exert this powerful influence over our behavior. We often don't want to look foolish, say the wrong answer, do poorly on an exam, or engage in similar activities. And this fear can be crippling in a few ways. It obviously makes it harder to do the practice, which means that it's harder to get better at it. It makes you have these worried thoughts which crowd out the space that you need for those few little instructions that you need to improve. It tends to make you more agitated. It makes the whole experience of learning less pleasant. So there are all these negative effects with anxiety and learning, but one of the things that has really been found to help with anxiety is what's called exposure therapy. This is a therapy that was originally used to treat phobias, but the idea is that you just give people progressive exposure to increasing levels of the thing that they're afraid of, and the fear response just naturally subsides, because most of the time when you have this kind of fear, nothing really bad happens. So generally, the response of speaking Spanish to someone is not going to be this heightened sense of fear, but just this diminished sense of fear over time. And it's amazing how this fear axis is different from the skill axis. So I was talking about this language learning project that I did a decade ago with a friend. And what was very interesting is that if you imagine right now, let's say you have minimal Spanish, and you're going to go to Spain, you're only going to speak in Spanish for all your conversations for three months. The feeling might be that this is a bit daunting. This is something that's a little bit scary, and it certainly was for us when we started too. But if you keep doing it, and again, we're talking about exposure on a continuous basis, the fear goes down really quickly. So after about two weeks, you have zero fear, even though you're still making lots of mistakes. And so your fear of speaking can be very low, even though your proficiency is not at this elite level yet. Whereas people who, let's say, spent years studying in school, maybe they're actually quite proficient, they actually have quite a bit of knowledge of Spanish, but maybe they don't practice very much, and so they still have this kind of latent fear about it.

SPENCER: Personally, I found exposure therapy really powerful. I've mentioned this in previous episodes, but when I was younger, I had a lot of social anxiety, and I went to a therapist who gave me exposure therapy assignments. For instance, go to a bar and introduce yourself to five strangers. And okay, some of them might not want to talk to you, which is fine, then just leave them alone. But when I would do this, I would feel incredibly afraid, like my heart would be pounding in my chest, but once you do it a bunch of times, you kind of realize, "Oh, nothing really bad happens." Your brain starts to settle down. And now I have dramatically less social anxiety, I would say probably even below average social anxiety. So, yeah, just a really powerful method.

SCOTT: Yeah. And I think for people who are thinking about this, because I think exposure, it's not all or nothing either. So if your description of social anxiety, if you have high social anxiety, just above baseline, and then I tell you, you gotta go to a bar and talk to five people, that might be so terrifying that you don't even do that, but you can take baby steps. I think the thing that people underestimate is how much this circuitry adapts. So, I mean, there's this great example of people who worked on the Empire State Building, doing the construction, and you see those photos of them way high up on those girders in the air. It must just be terrifying, but in the beginning, they feel that way, but then they go again and again and again, and eventually the fear goes to a really low level, even though, objectively speaking, most people would be very afraid of heights. And so I think this kind of taking baby steps, doing little bits that are outside your comfort zone, I think we underestimate how powerful that can be in rewiring that part of your brain which feels so default and automatic.

SPENCER: You mentioned that a lot of the barriers to learning are emotional. How do we set ourselves up at the beginning of a new learning process so that we're more likely to persist and stick with it and get to our final learning goals?

SCOTT: Yeah, so I think a lot of it has to do with what we were talking about earlier, about the success rate. The success rate, probably for most skills, is too low in the beginning, that even with schools and even with patient teachers, we don't break down things enough, especially for the students who are likely to struggle. So I mean, there are always those students who breeze through the intro class and need it to be more difficult. But I would say that if it were dealing with 80% of the cases, it probably could have been made easier. And then for the people who are at the bottom 20% of the class, it could have been made a lot easier. They're unlikely to succeed in the format that it was given. And so that's one of the things that we can do, is how do we make it so that you start off with success? Start off with, you're getting it. You're feeling this confidence. There's this whole theory of self-efficacy, which I find very valuable when thinking about learning in general, which is that our motivation is not just, "Is this set of actions rewarding? But do I feel like I can achieve this set of actions?" So it's not just, "Would mastering programming be useful for my life?" It is, "Do I think if I attended the classes, I could actually learn programming?" If I don't believe in that quality of myself, I'm unlikely to pursue it, even if I know it would be beneficial for me. So there are so many beneficial behaviors, not just related to learning, that are tied to this construct of self-efficacy. I think the research really shows that if you want to improve self-efficacy, you have to show people succeeding at this skill, so they can see how it works, and they can feel like they can emulate it themselves. You need to have mastery experiences, so you have to experience success personally. Those two factors, I think, are very important, and they really point toward, "How can you break things down more and make them a bit easier in the beginning, so people have an on-ramp to success?"

SPENCER: I've seen so many people label themselves as bad at math because they had some bad math experiences. Maybe they got a bad grade, and then they apply this label to themselves, and it seems to me that this probably inhibits their future learning.

SCOTT: Yeah, I think identity is an important component too, because we form these stories about ourselves, and it's not that the stories are totally false. It's not like everyone who's bad at math is actually Einstein; if they just don't realize it, maybe they are a little worse at math than other people. But I think sometimes we take these stories as reality, rather than as just something that we're telling ourselves over and over again. It may be the case that, yeah, you've had some bad experiences with math, but maybe you could learn math, maybe you could learn a language, maybe you could do some of these things, and you don't know. I think the viewpoint that I try to hold, because I think it's often very difficult to persuade people that they're wrong about their self-assessments their whole life, because usually there's at least a grain of truth in them, is that we don't know enough about ourselves. We only have this limited set of life experiences, and we draw these really firm conclusions about who we are, what we could be, what our potential is, about our identity, that really, I think, are unwarranted. That level of confidence is unwarranted. I think it's better to have this kind of flexible view that, "You know, maybe I've had some bad math experiences, but maybe I'm not actually bad at math. Maybe I could learn math. Maybe I could do this." If you have that opening, then you at least give yourself the opportunity to start building self-efficacy, to start finding skills and talents that you didn't know existed.

SPENCER: I suspect at least some of the people who think they're bad at math really just had a bad math experience, a bad teacher who didn't explain things clearly, or who was teaching at a level that they weren't at yet, and so they felt dumb. Maybe some of the other students in the class had more prior experience. It can be hard to tell the difference between, "Yeah, you're actually not that good, you have less innate talent," versus, "Oh no, you just went through a bad experience, and now you've labeled yourself this way."

SCOTT: I actually have a story that I like to tell, which I think really encapsulates what you're talking about here. I have this friend, and she has a master's in civil engineering, and she works on big hydroelectric dam projects where you're doing fluid flow calculations and differential equations. This is a complicated technical skill that she possesses. I remember she told me this very casually, but very sincerely. This wasn't just false modesty. She told me that when she was in university, in her first year, she took an intro computer programming class, and it was too difficult for her. She said, "I wasn't smart enough to do that intro programming class." Here is someone who's done a master's that involves a lot of advanced calculus and physics, and they think that an intro programming class was too hard for them. What's going on? I think what's going on is that the class is populated with people, perhaps like you and me, who maybe did programming in high school, and we already knew the stuff in the class. When the assignments got handed out, we didn't struggle with them; we did them right away. She's looking around, and it's much of these nerdy guys who already knew programming, and they're breezing through the intro computer science assignments, while she's having to actually learn them, so they're a little bit difficult for her. She's making a social inference about her environment, saying, "It must be that I'm bad at this. It must be an intrinsic quality of me that I'm not good at this," rather than the equally true alternative, which is that all these other people just have more experience than she does. We have to be very careful because while I believe there are intrinsic differences in learning ability and strengths among people — I'm not trying to make that claim — but I think we often blend those together with people's actual life experiences, and we neglect that someone has quite a bit more training or prior experience that helps them learn a skill than we have. Then we take our unfavorable comparison to them and think, "Oh well, I'm either doing something wrong, or they're much smarter than me or much more talented than me, or I don't have what it takes to learn it." It might just be that you do have what it takes; it just takes you a little bit more time.

SPENCER: Before we wrap up, I'd be interested to hear about one of your own learning projects, maybe since you've written your book UltraLearning, and break down for us what techniques you used and how you approached it.

SCOTT: There are lots that we could talk about. I did a project called the MIT Challenge, where I was learning MIT's computer science curriculum. That was one of the first ones I did. We talked about the language learning project a little bit. I did another one that was learning portrait drawing. This was a skill that I've always had an interest in; I've always been interested in drawing, but drawing faces is really hard, especially if you want someone to actually look like the person you're drawing. It's very difficult, and a lot of drawings, when you try to do a portrait or a self-portrait, just look awkward. They look bad. This was a project I did over 30 days, and I feel like I got fairly competent by the end of it. It's hard to describe in audio, but if anyone wants to check out Scott H. Young's 30-day portrait experiment, you can see my self-portrait at the end and judge for yourself. The thing I found really helpful was getting a lot of rapid feedback. I did a lot of practice where, after I was done, I would take a picture of the drawing I did, put it over the computer photo of whatever I was trying to draw, scale it a little bit to make it overlap, and then reduce the transparency of one of those a little bit so you could see how it's lining up. You could very quickly tell, "Oh, I'm drawing the eyes too big," or "I misjudged this distance," or "the head is too wide or too narrow." That was really high-quality feedback, which is often absent in the skills you're learning. I also did the meta-learning. I looked online and found lots of different ways that people used to do portrait drawing. It turns out a lot of these techniques are not obvious. The one I ended up using that gave me a lot of success was taught by this fantastic artist, David Jamieson, at a website called Vitruvian Studios. He has this method where you find these landmark points and triangulate them. It takes a bit of time, so you can't do a quick caricature, but it allows you to get phenomenal accuracy when you're drawing someone's face, and it can be quite satisfying in the end. Finding the right method and practicing the right method really made the difference. I know that's just one example, but I hope it encapsulates some of the techniques we've been talking about in this conversation.

SPENCER: The experience people have when they're learning, and I think this happened to you as well in your porch drawing experiment, is you hit a plateau. What do you suggest to people when they're like, "I've been training, I'm training, and I was improving, but now I'm just stuck. I don't seem to be getting better anymore?"

SCOTT: Yeah. I think plateaus are very common, and they can happen for a few reasons. One reason is related to getting stuck in suboptimal habits. I don't even want to say bad habits, but not the best habits. Seeking a coach or more expert feedback can be very important for moving up a level. The other reason you can hit a plateau is simply that that's the nature of the skill. Language learning is a great example of that situation because of the distribution of word frequencies. If you think about how frequently the word the shows up in English, and then the word ball, and then the word skeleton, it keeps going down in frequency. The amount of work it takes to learn one word, especially if you're learning it through flashcards, is about constant, but the words really drop off in frequency. You have to learn more and more words in order to get the same proficiency boost because they show up so infrequently in language. That is an example where the nature of the skill itself is that you're going to have really rapid progress because those first 1,000 words are super useful, and you're going to practice them all the time. Then you get to the 10,000th word or the 20,000th word that maybe you use once every three months, and it's not nearly as beneficial. I think it's important to keep these factors in mind because while there are places where you can break through a plateau just by, "Oh I am learning a new technique or a new way to make progress," in some cases, you're just hitting the diminishing returns part of the learning curve. That can be a frustrating situation, but ultimately one that requires patience when you're trying to approach it.

SPENCER: Scott, thanks so much for coming on.

SCOTT: Oh, thanks so much. It was a great conversation.

[outro]

JOSH: A listener asks, "What's your favorite mathematical concept that you've applied to the field of psychology?"

SPENCER: I find that to make progress on questions of psychology, you often don't need super fancy mathematics. You can often make progress with relatively simple things like linear regression or logistic regression or even just correlations, right? So a lot of simple tools are actually really, really useful. That being said, I have gotten to applying more advanced tools. I have done a bunch of work applying neural nets in psychology, and I actually think it's very exciting, and there's a lot of really interesting things that can be done with neural nets. For example, we've created a system called Personality Map. You can actually use it at personalitymap.io, although as of the time of this recording, we haven't announced it to the public yet. Hopefully, we will soon. And it actually lets you see the results of applying a neural net to predict the relationships between many different traits related to human psychology. So we actually have over a million correlations that you can explore on that website.