Recently I read a lot of books and material on the science of learning and wanted to sort out some of my impressions and thoughts. But a structured article would take the length of a book, and I am far from ready for that. My initial plan for studying this field is five years; this piece can count as a decent start.

1. Why care about learning?

Two personal reasons. First, I ran into a lot of difficulty learning complex technologies such as AI. Second, I want to give my child an example lived out in practice. From a larger view, the human learning mechanism is far more complex than today’s machine learning. Thanks to brain science, cognitive science and psychology, the science of learning has produced some results, but most of it is still unknown, and even these limited results have not been effectively spread. As in medicine, personal experience matters a lot in learning, so custom and folk remedies often have more life in them than scientific conclusions. Programmers are among the professions that feel the importance of learning most keenly: new technologies keep arriving, and behind every line of code there may be a person doing repetitive work being replaced. As the internet levels the information gap, more and more of the rich list are top students. That is a picture of cognitive advantage replacing information advantage. Whether for the utilitarian goal of high scores and big money, or the joy of pursuing truth, learning is a discipline you have to practice.

2. Is learning only for students?

Equating study with being educated in school, getting high scores and a diploma, is the biggest misunderstanding of learning. If you have ever built a “stupid” artificial intelligence, you know every child is a genius at learning. Yet the failures of institutional education very likely have their original enthusiasm for learning put out, while its successes may fail in society and be held up as proof that study is useless. Many people stop progressing after their first few years of work; their last large-scale learning was forced on them while adapting to the job. When they become parents, their own knowledge has stalled, yet they expect their children to excel, and get anxious over advertorials from tutoring companies about “fifty thousand a month and you still can’t raise a child”. But learning is in essence just an instinct of being human, and what people generally call studying is only one of the paths to it. Confucius said: among any three people walking, one can be my teacher. He also said: to learn and to practice what you have learned, is that not a pleasure? I cannot put it better than he did.

3. Among all ideas about learning, what matters most?

The most important thing, I think, is to abandon the “entity theory” of intelligence and ability and embrace the “incremental theory”. Put simply: do not believe that intelligence and ability are innate and unchangeable, but believe that they can be raised through learning. Splitting people into genius and ordinary, high IQ and low IQ, smart and dull does the greatest harm to human progress. No one denies that people differ by nature, but at the level of effort most people actually put in, they are usually not yet in a position to be comparing talent. If everyone’s ability were laid out on a spectrum, each person would have a place on it; it is neither binary nor fixed. Everyone hits the ceiling of his own ability — even genius scientists, who may spend a lifetime failing to understand one problem. If a person is born with less ability and believes that cannot change, then clearly his ability will not improve. If a person is born with more ability but believes that cannot change, then when he hits the ceiling, meets difficulty and setback, he will doubt his own intelligence and ability, and inevitably fail or run away. That is what people call clever but not wise. Holding different beliefs for a lifetime produces results worlds apart. And it is not only a belief: brain science long ago proved that human nerve cells can keep being born throughout life.

4. Does effort matter more, or method?

This is of course a false question; the answer is obviously both. The deeper question is what kind of effort is the right effort. In traditional thinking, learning is a serious matter: sit upright, tie your hair to the beam and jab your thigh with an awl — that is the proper way. For those who get good grades but do not look like they work hard, we praise their talent and urge them not to waste it: they should work harder. For those who look hard-working but get mediocre grades, we are generous in praising their effort; compared with the poor students who get bad grades and don’t try, they are always much better off. Rereading the material, highlighting the book, taking beautiful class notes — these are common study methods that research has shown to be ineffective. What they share is that they cost the learner a lot of time on effort that stays on the page. However clear and pretty the notes, if they cannot be moved into your own head they are useless. Every useful method rests on the principle of active engagement. If you want to actually understand what you are studying, there can be no self-deception in it. Being hit by your own ignorance is unpleasant; hiding inside self-satisfied effort is comfortable. But there has never been an easy way to learn.

5. Why are exams important?

The most common and most effective study method is the exam. Treating the exam as nothing but a tool to check results, making the score the only goal — either score-worship, or loud talk of “quality education” that avoids the word exam — is our biggest misunderstanding and waste of this tool. First, an exam is the ultimate weapon against the illusion that “I already know this”, an illusion that rereading the textbook or your notes always produces. Second, an exam forces you to recall what you learned, which is an excellent way to consolidate memory and understanding. A good teacher keeps testing students by every means — pre-class quizzes, in-class quizzes, post-class quizzes. The aim is not to rank students but to make them retrieve their own memory, deepen the impression, and find the gaps. Even outside formal study — reading a book unrelated to your field, say — you can make full use of the exam. The simplest version: after a chapter, close the book and ask yourself a few questions. What was that material mainly about? Do I agree with these views? What connection do they have to what I already know or have experienced? If I agree, how would I apply them in life and work?

6. If study matters so much, why are there bookworms?

Why is it that we understand so many truths and still cannot live this life well? The problem is not the truths; it is ourselves. If you treat study as entertainment, like playing mahjong, then being a bookworm is no bad thing. But if you are after progress, study is only a beginning, not an end. The bookworm is keen on how many books are on the shelf, on how fast he reads, on how many books a year, keen on quoting from books, keen on how much “solid stuff” he has seen — but never on how much knowledge and how many ideas have taken root in his own head, whether those ideas form a complete, coherent, evolving system, and how many of them have been put into practice. Zeng Guofan was a scholar who could lead troops into battle. He wrote a diary every day to examine himself, and what he examined was how his practice should carry out his ideas. What you get from paper is always shallow; to know a thing truly you must act on it. I cannot put it better than the ancients. Confucius and Zeng Guofan needed daily self-examination. Why would we not?

7. Does more practice mean better learning?

The ten-thousand-hour rule is a widely circulated idea, and the “deliberate practice” theory behind it is also much misunderstood. “Effort always pays off” is a comforting thought, but effective practice is not simple repetition. Both scientific research and the personal experience of geniuses show that effective practice has to be highly active. First, you give ground in order to advance: divide the skill into fine enough modules, and practice module by module, hitting your weak spots. Second, the practice cannot be simple repetition; it has to vary and interleave. These findings are counterintuitive. People always feel that repeated drilling gives better mastery and that varied, interleaved practice works worse — an illusion that is hard to overcome. Truly effective practice burns a great deal of mental energy, and cannot be sustained for long stretches.

8. Why can’t I remember or understand what I read?

The latest memory theory holds that retrieving a memory matters more than forming it in the first place. Everyone is capable of total recall; the problem is that the fragments lie scattered in the brain and cannot be reorganized and pulled back out. The only way is to build more connections among them — like a village that was backward until concrete roads ran out in all directions, and then was no longer backward. For knowledge that is not yet systematic there are plenty of little tricks, such as making up a jingle: “for the long months there’s a rule that holds — odd ones before July, even ones after.” For systematic knowledge, the most important thing is to build your own knowledge system: every new piece of knowledge should be connected as much as possible to what you already know and have experienced. Obviously, this is not a quick or easy process. Many people have crammed before an exam, sometimes with decent results, but that too is an illusion. Knowledge that was never, through hard work, brought into your own system will be handed back to the book.

9. Can I sleep less to learn more?

One of the myths about successful people is that they sleep only three or four hours. But they can catch up on planes, they can put a bed in the office, their mental work as decision-makers is low-intensity — and they may simply be lying. If you do high-intensity mental work — a grad student, a programmer writing code — then sacrificing sleep, apart from possibly killing you earlier, will seriously lower your learning efficiency. Brain science shows that building synapses requires adequate sleep. Any learning that builds no new synapses is a waste of time. Modern habits lean more and more toward late nights, and even children are affected by this cultural shift. That children are pressed by homework into too little sleep is a mockery of brain science.

10. How is a poor student made?

In the Japanese film Flying Colors, the heroine prepares for the college entrance exam but her math is far too weak. Her tutor’s method is to start from elementary school math. This is not exaggeration; it has a real prototype. The flaw of institutional education is that it has to set one standard for the majority: what language a child should command at 8, how much math he should know at 9. Someone will always fall behind, and those who fall behind are likely to be stuck with negative labels. Maybe he only matures enough for math at 12, but by then the chance is gone; often these children no longer feel any involvement in learning at all. Europe and America do have mixed-age classrooms that follow the child’s actual level, but if a 12-year-old went to study third-grade math, neither the parents nor the child could keep face. In such a case the cram school may be the best way out. Learning anything starts slow and then speeds up: lay the foundation first. Learning is like building a house — a tower ten thousand feet high starts from the ground; do not build a high terrace on shifting sand. I cannot put it better than the ancients.

11. Why does what you never forget always come back?

When an expectant mother is pregnant, the expectant father notices more pregnant women while out shopping. When you are struggling over the structure of the benzene ring, at night you may dream of a snake biting its own tail. When you are thinking about a problem with everything you have, even after you stop, something like a background process keeps running in your head. Your senses collect related information without your willing it; your dreams try out various solutions. Breakthroughs and inspiration arise naturally, but not out of nothing: they belong only to a brain that has gone all out and then relaxed. Research shows that geniuses usually stand out only in their own field; their flashes of insight and strokes of brilliance are nothing but the natural gift that follows countless hours of hard work. If you want to get rich but spend less than ten minutes a day thinking about it, how could pie fall from the sky? The echo that follows never forgetting is a matter of probability, but we can raise that probability.

Main references (all links go to the Chinese editions on Douban):

• 有效学习
• 如何学习
• 所谓学习好，就是方法好
• 认知天性 : 让学习轻而易举的心理学规律
• 好好学习 : 个人知识管理精进指南
• 科学学习 : 斯坦福黄金学习法则
• 学习的艺术
• 学习之道
• 如何高效学习 : 1年完成麻省理工4年33门课程的整体性学习法

Thinking about it, the biggest reason is probably that university material carries far more information than high school material while the total hours are shorter, and on top of that I had less clear a goal than in high school, so without direction I worked less hard, and never went through the material more than a couple of times. Whereas that little bit of high school knowledge had been practiced over and over.

I do think that intensity of drilling in high school is excessive for learning itself; it’s an arms race whose whole purpose is to sort people into tiers for university admission. But things tend to swing from one extreme to the other: sickened by repetitive practice, people arrive at university and loathe all practice. Practice with immediate feedback, and a suitable amount of repetition, is an effective way to build permanent concepts; practice has shown it works, and it agrees with what neuroscience finds.

We can study all sorts of learning methods and look for time-saving shortcuts, but method is usually not the most important thing. What matters most are the truths that have been repeated to death — that you have to keep at it, that you have to practice repeatedly. Only on top of those does it make sense to talk about method.

In fact understanding grows as you age, and if you dig deep in one field so that a knowledge system is already built, new knowledge needs fewer repetitions. But all of that rests on the earlier knowledge being solid.

So there’s no room for self-deception in learning; whatever you skipped will be repaid in this life. Of course the sooner you know this the better — though when you have no goal in life, knowing it does you little good.

The basic goal of reading should be the unity of knowing and doing, so reading for entertainment is outside the scope of this piece. It applies to books, and also to reading medium-length and short articles.

Step One: Skim

This step exists specifically for impatient people like me. Get hold of a good book and you always want to tear through it. Though on reflection, what counts as having finished a book is itself a big question worth thinking about. So if you really can’t restrain yourself, cut yourself some slack: read it quickly once, and at least you’ll have a rough impression. The other case is that after a quick pass you find the book isn’t worth a careful read — that saves you time too. From this angle, speed-reading isn’t even necessary; a skim, a sense of the general shape, is enough — treat it as part of choosing the book.

Step Two: Enact-reading

I don’t know whether I coined this word. What I mean is reading in detail — in such detail that the author’s line of reasoning can be performed inside your head. If you can’t reach that level, there may be a few reasons:

1. Your comprehension isn’t up to it, your background knowledge is insufficient — this book has come to you too early;
2. You need to read it a few more times;
3. It’s a bad book.

Telling these three apart is genuinely difficult and takes accumulated experience; generally, the more you know about a field, the easier the distinction becomes.

The purpose of enact-reading is to truly digest the book’s content and understand the author’s thinking.

Note: during this process it’s best not to take notes, and certainly not to highlight anything. Don’t use busy hands to cover for a lazy mind, putting on a show of diligent reading to fool yourself.

Step Three: Extract

Close the book and start taking notes. Depending on how much information the book holds, whether you do this after one part or after the whole thing is up to you.

Taking notes with the book closed is a simple and effective self-check: it requires you to feed what you learned back onto paper. The notes can be a bare outline, as long as you can reconstruct the remembered and understood content from those points. If you have the time, the more detailed the better. If there’s really no time or no conditions, at minimum close the book and quiz yourself.

When the notes are done, open the book again and compare, to see whether there are errors or omissions.

Step Four: Network

To hold onto what you read firmly, the key is converting it into part of your own knowledge system, linking it into the neural network you already have. The basic methods are as follows.

1. Compare: compare it with knowledge you already have — where do they agree and differ, is this another face of something you knew, or an important supplement to it;
2. Reflect: is anything in the book inaccurate, or not detailed enough, or could it be put in a way that’s easier to understand.

Step Five: Practice

This is the last step, and the most important one, the one that cannot be skipped — especially for books that offer methodology. If one important practical technique in a book is understood and accepted by you, then the best thing is to go do it. That is the best destination for reading; a book with even one such point is worth a hundred times its price.

Of course, not every book has direct methodology you can practice. Then there are fallbacks:

1. Reading notes: ideally an improved version of the step-three notes, reorganized with your own understanding, or with your own understanding added, or using the ideas you learned to explain a new problem.
2. Conversation: introduce the content to someone who hasn’t read the book. Choose the appropriate approach and content for different listeners, but convey the same core idea.

At this point, you’ve roughly finished a book.

One last note: these five steps are not strictly separate or sequential. In practice you can trim and interleave them freely, but the last step cannot be skipped — otherwise it’s just entertainment reading, even if the book you’re reading is a very serious one.

Go looking for all those inspirational chicken-soup stories and you’ll find plenty of analysis, usually starting from Ma’s childhood, as if he were destined to be a business titan.

But if you analyze China’s historical trajectory over the past thirty-plus years, the appearance of a figure like Jack Ma was almost inevitable. China has the world’s largest market, and add to that the wave of the information revolution: the emergence of new-economy billionaires was inevitable. Hence: the wind rises and the pig takes flight.

So the question becomes: why is this person Jack Ma? Seen historically, it’s just a random event — if not Ma Yun, then Li Yun or Zhang Yun. All the chicken-soup stories are hindsight: they link every factor that can possibly be linked to the final success, because the soup needs to give the reader a momentary illusion — I can succeed too, as long as I do what the book says.

On the matter of success, I think the ancients said it all: do your best and leave the rest to heaven. I’d like to read it again through probability.

A person’s success is always a probabilistic event. There are too many uncontrollable factors in fate; no one can guarantee control over their own destiny. Send Jack Ma back in time for another run in a parallel universe, and perhaps one tiny change would be enough to keep him out of the ranks of the richest. But even so, I believe he would still live a fuller life than most people, because he did after all think harder than most, and acted on his thinking.

Success is a probabilistic event, and hard work can raise that probability — sustained hard work especially. Something with a 50% success rate, tried twice, raises the probability of success to 75%, because life is not gambling: often seizing one opportunity is enough to launch you into the sky. Take the Vietnamese guy who made a small fortune with Flappy Bird — before that he’d made a dozen-odd apps and not one of them took off. He merely, through his own persistence, waited for one success.

The truly useful principles may not excite you directly when you hear them. “Do your best and leave the rest to heaven” is that kind of principle: it asks you to exert your utmost, yet not to harbor excessive expectations of success. How many of the restless can actually do that?

If I work hard and still might not succeed, why should I work hard at all?

Hard work raises the probability of success. Even a pie falling from the sky requires you to go outside to pick it up.

What if I keep working hard and never succeed?

First, success should be seen as a return on effort. The return after effort can be large or small; it is a probabilistic event. So you should pay more attention to growth — in essence there is no clear boundary between success and failure. Take a step back: even if you are never “successful,” a person who can strive for a lifetime has not lived this life in vain, and has already come close to the meaning of life. That should be regarded as a gift from heaven.

1 Roaming

Roaming here means, to some degree, wandering the internet with no clear purpose. It is the internet’s biggest disruption to how we get information, and also its biggest double-edged sword. Since hyperlinks appeared, information has become easy to connect and infinitely extensible. Traditional information organization was mostly linear or tree-shaped; the internet’s is networked. That brings many benefits:

1. A network can model any complex structure, so it offers a truer way to represent the structure of information — Wikipedia’s structural advantage over a traditional encyclopedia, for example;
2. The boundary between the inside and outside of a piece of information blurs; “references” are easier to reach, and building a topic page on the internet is far easier than in traditional media;
3. It offers unlimited possibility of running into surprises: while roaming you may find things you never knew, strike sparks of ideas, and break out of information silos and your own cognitive barriers.

But the flaws of this mode are obvious too:

1. Too much time gets eaten by junk information; many people who produce information have ulterior motives, and clickbait of every kind is peeking at your attention at every hour;
2. What you get is scattered and unsystematic; internet information is still mostly free and, compared with books, still relatively low in quality;
3. Roaming easily comes with restlessness and so-called information overload: inspirational articles replace resolve, knowing about replaces understanding, awareness replaces knowledge.

For roaming to do its job well, what you most need is strength of your own — values, critical thinking — so that you can keep absorbing nutrients from the ocean of information instead of being drowned by it.

2 Subscribing

Subscription was the main way traditional media interacted with people before the internet, and as the internet developed it became one of the internet’s ways of getting information too. Most information providers want users to subscribe and build a long-term relationship, because that is how they profit; the form may be following, push notifications, and so on. WeChat official accounts, even Moments, are a retrograde reaction against information roaming — WeChat official account articles even deliberately keep share links unstable, trying to build their own little ecosystem. Subscriptions have many variants; the most avant-garde is the recommendation mechanism brought by machine learning, which is essentially a subscription based on individual interest. Retrograde or avant-garde, subscription has its advantages:

1. Subscription is a filtering mechanism: it concentrates attention on the people and things you actually care about, or lets you filter information by following people you trust;
2. Subscription takes readily to paid mechanisms, enabling high-quality knowledge services that keep information quality high while iterating fast.

But subscription brings risks too:

1. Cliquishness: whether you filter for yourself or a machine filters for you, in the end you hear and see only what you are willing to hear and see. Mistaking what you like for the truth about the world is a big risk;
2. Behind apparently active subscription there is a great deal of passivity. People tend to defend their own choices; once you are someone’s fan you may gloss over his mistakes, and if you have put money in, you will hold to your original choice even harder.

3 Searching

Searching is a required skill for academic research, and I think it is also a required skill for any modern person living in the internet age. The search engine is the greatest invention on the internet after the hyperlink. Through searching, a person getting information can truly treat the internet as a tool:

1. Convenient search makes holding isolated facts nearly worthless — questions like “how wide are China’s railway tracks?” on a television quiz show — so entering a new field becomes easy;
2. Searching is strongly purposeful, which gives “information for my use” a good starting point; you no longer browse for two hours and then cannot recall what you read;
3. Searching usually solves the problem right away, as long as you are not the first unlucky person to hit it.

For searching to work, beyond improving your basic search skills, what matters most is your own level of scholarship. Otherwise you fall into the trap of substituting search for thought, satisfied with the information this external brain hands you, unable to synthesize it and come up with ideas of your own.

4 Putting them together

I split out three modes in order to clarify the concepts; real situations are usually a combination of all three. Say I follow a programmer on Sina Weibo (subscribing); he mentions an interesting new programming language and gives a link, so I click over to take a look (roaming), find it is genuinely good, then Google the language’s homepage and a lot of other material (searching), and study it for an hour straight. But it could also be: I follow a programmer on Sina Weibo (subscribing); he shares a funny video, I click over to take a look (roaming), and one video leads to another, and an hour later I have no idea what I have been doing.

Each of these ways of getting information has its advantages; the key is to use them “consciously,” keeping your brain aware of its own state at all times. Even in the freest roaming state, you should set a time limit, otherwise countless attention black holes will kill your time and leave you with no idea how.

The internet is a powerful tool, and it will even steamroll the weak. If you do not keep improving your judgment and analysis, you will only have your attention harvested over and over, be brainwashed over and over, and feel that the world is full of perfectly sensible sayings — and still not live this life well.