Learning, Habits and Automation

Table of Contents


Learning a skill, which you’ll use many times, has two basic steps. First, there’s figuring out how to do it at all. Second, there’s practicing it until it’s an automatic habit. When you’re done, you can autopilot the skill without really thinking about it. That lets you do other things at the same time. For example, I’m autopiloting my touch typing while thinking about how learning works and what words to write.

Practice also lowers your error rate. Once you are able to do something once, you’ll still often screw it up a lot, e.g. half the time. But after you practice, you can get your screw up rate way down, e.g. to 10%. For a skill you use a lot, you generally want to get to the point that you screw up less than 1% of the time. My success rate at pressing the correct keyboard keys is over 99%.

You can’t practice until after you know the right way to do something. You don’t want to practice doing it wrong. If you’re trying stuff out to see what might work, that’s not practice, it’s experimenting. There’s a conceptual difference between trying out multiple ways of doing something (experimenting, exploring, testing or trying out stuff you brainstormed) and repeatedly doing it one way in order to get better at doing it that way (practice, repetition, gaining experience). The lines are blurry though because, while practicing, people will try out small adjustments and try to optimize. While practicing something, people will simultaneously experiment with some details that don’t change the overall thing.

Methods like these mean learning tends to follow certain patterns instead of being fully unpredictable. A common pattern is going slow at first and speeding up later. Why? It’s harder to correct your errors when going fast. You need to figure out what you’re doing wrong and fix it, which is easier if you go slowly. After you’re satisfied with how well you’re doing something, then you can repeat it a bunch to make it more intuitive/automatic/habitual, and you can speed up while trying to maintain the same quality level.

Another pattern with learning is paying high attention to skills at first, while learning, and then paying less attention later. It takes a lot of attention to learn something initially because you need to figure out how it works, correct your mistakes, figure out how to go faster, etc. But the final goal is often that you no longer need much attention to do the skill.

So you practice a skill and get it to be fast, low attention, and low error rate. Then what? Now you can build on it. You can now learn a more complicated skill that uses the first skill as a component. Once something is easy, cheap and reliable, it can be used within other more complicated things.

For example, carrying something while walking is a more difficult skill than walking alone. First you need to get good at walking. And you need to get good at holding objects, too. Then, after you’re good at those skills (can do them with high success rate and low attention), you can do the more difficult task of walking while carrying something, which uses both the earlier skills. Complex skills come from combining multiple simpler skills, but you can only do that after you master the simpler skills so that they don’t need much attention – that way you can focus your attention primarily on the complex skill. Once you get good at carrying something while walking, and also good at reading, then you can combine them to create the more complex skill of reading while walking.

Philosophy Learning

In philosophy discussions, you need skills like looking at things from a neutral perspective (and from many perspectives) instead of being biased. You need to recognize and avoid logical errors. You need to understand grammar and what words modify what other words. You need to be able to create a mental model of a person you’re talking to and use it to help you guess what their point is, what they do and don’t know, etc. And you need to understand your own viewpoint enough to compare it with what the other guy understands to see what he’s missing, so you can tell him. And much more.

For a good, productive philosophy discussion, all of those things (and many more) need to be small/easy skills that you already practiced in the past and can do quickly, with a low error rate, and without much attention. That lets you focus your attention on the ideas being discussed. Each skill I mentioned, and dozens more, should be happening pretty much automatically and habitually. That lets you use your conscious attention to monitor what’s going on, to look for problems, to overrule your habits in some cases, and to think about and try to learn about the topic being discussed.

What do you do if that’s hard? What if you’re not ready for all that yet? Then have simpler discussions, for practice and learning, and don’t expect them to make significant philosophy progress. You should aim primarily to work on your skills rather than to figure out correct answers in philosophy. A good approach is to try to do the skills you already have practiced and automated while also developing one more skill. Keep in mind one specific thing that you’re practicing and working on, and make sure to keep focusing conscious attention on it. Learning is generally easiest in a one-thing-at-a-time way. Learning a few things at once is possible but harder and unnecessary.

Learning one thing at a time can easily be faster than learning two at once. In other words, you can learn two things in a row, sequentially, and finish before you would have learned them both together. The biggest exception is when you find stuff easy. If it’s easy enough for you, you can learn a few things at once. It’s the things that take more attention and effort, and that you struggle with some, that are usually good to go one at a time with. If you’re having difficulty, try focusing on one thing at a time. But if some things are so small or easy that you barely even notice them as things to be learned, then there’s no problem because they don’t need much attention. In other words, learn one thing at a time for the things that require you to focus on them.


Learning, and particularly practicing, creates habits. If you can do something without a bunch of effort – if it’s automated and intuitive – then it’s at least partly like a habit. It’s something you might do without really thinking about it, on autopilot.

Be careful with what habits you create for yourself because changing them can be hard. Recently created habits can be pretty easy to change, so you might think you’re flexible and it’s no problem. But once a habit gets older it’s often much harder to change because you forget some details of how it works and what the alternatives are. To change an old habit, you may have to relearn the subject again, similar to how you learned it originally, which can be a bunch of work. And it can be hard to approach something like a beginner when you already have old habits.

With old habits, you know how to do them automatically but you may not have much conscious knowledge of what you do, let alone why. When you create a habit/skill, you’re giving conscious thought to what to do and what you’re doing. The longer ago that was, the less you may consciously know how it works. But it’s possible to change old habits, and sometimes it’s not that hard. I recently changed a very old, ingrained typing habit (which finger presses ‘p’) with only around an hour of effort. (It’s hard to measure the effort. I spent less than an hour, over several sessions, practicing typing the new way while paying full conscious attention to what I was doing. But then, for a while after, I paid partial attention – not much but some – while typing other stuff.)

It’s best to make a serious effort to decide what’s correct before you put in the effort to practice until it’s a habit. You don’t want to be constantly changing your habits. That’d waste effort. But don’t be scared to learn things, either. You’ve got to do your best, use your judgment, and develop some skills. You can change them later if there is a problem. Just don’t do it carelessly.

If you’re a child, you should mostly start with skills that lots of people think are good. Learn those. Many of them are useful, and if you decide that one is mistaken it’ll still help you understand other people. And there are lots of resources to help you learn our society’s standard knowledge. There are a lot more ways to be mistaken than correct, and smart people have put a lot of work into correcting mistakes over the generations, so start with what people already know instead of trying to go your own way immediately. You can’t judge that common knowledge is wrong before you understand it.

If you have a hard time learning philosophy skills, you can practice learning by learning something else easier. Then after you get the hang of how learning works, and you’ve practiced the steps (figuring stuff out, then practicing to make it a habit you can autopilot), then you will have an easier time learning difficult skills. Get the hang of learning with something easier before trying to learn something ambitious.

Testing Your Knowledge

The hardest things to learn are ones where you don’t know if you’re doing it right or not. It’s easier to learn things where you have a clear goal and it’s easy to tell what is success or not. E.g. with shooting a basketball, you want it to go through the hoop. It’s easy to see if you missed or succeeded. But if you write a philosophy essay, there’s no simple way to know if your arguments make sense or if your writing is clear or unclear.

Grammar, math and speedrunning are pretty good areas to learn and practice. Grammar has “right answers” to a reasonable extent but not completely, so you can somewhat look up the right answers or ask an expert. Math is easier than grammar since you can check many answers with a calculator and there’s less debate over right answers among experts. (I think math experts mostly debate advanced stuff, but not the math that’s taught before university, whereas grammar experts still debate some pretty basic issues.) Speedrunning has a timer to tell you how good a strategy is – the timer is like an answer key that makes it easy to see what’s better and what’s a mistake.

To learn you have to test your knowledge, meaning check if it’s correct. If you have no methods to check if you’re correct, you’ll end up learning lots of errors. There has to be an error correcting aspect to learning.

Error correcting is a larger issue in the first stage when you’re figuring out what the correct way to do something is. When practicing, you mainly try to correct detail errors in how you do the skill, so the error correction is more limited. You aren’t reconsidering the whole thing. (Though it’s reasonably common that people think they know how a skill should work, start practicing, then run into problems and have to go back to stage one and revise their view of the correct approach.)

Things like answer keys or seeing if the ball goes through the hoop are good tests – pretty reliable and pretty easy to use. If you try to read and understand a philosophy book, one of the best tests is writing down your understanding of the concepts and then asking for criticism from an expert (from someone who already understands it). That’s harder though. How do you judge who is an expert? What if they say something and you’re pretty sure they’re wrong? What if multiple experts disagree with each other? What if they say you’re wrong but you don’t understand their reasoning? What if you can’t find an expert who will read your stuff and comment? You can end up debating what’s true, which is not very similar to consulting an answer key or measuring success with a timer.

But do you know what’s worse than a hard-to-use method of testing the correctness of your knowledge? No method! If you read a book and just assume you understood it correctly, and do nothing to correct errors or test your knowledge, that’s way worse. A somewhat effective method of error correction beats no method, by far. You can try to spot your own errors, and that is a method – it’s not really good enough but it’s way better than nothing. It’s too vulnerable to your biases, blindnesses, weaknesses, irrationalities. Other people differ from you and don’t make all the same mistakes as you (even if someone is worse than you, there are some mistakes you make that he doesn’t), so it’s valuable to involve other people in error correction.