“I don’t remember a damn thing.”
The book I held in my hands was full of highlights. It seemed like I’ve got all colors of the rainbow on a page. Apparently, this didn’t help. When I tried recalling ideas from the book, I didn’t hear a thing. Just. Silence.
Terrified, I started questioning how much I really know. If I forget everything I read, I can’t apply my knowledge to the problem at hand. I can’t transfer it. And without transfer, knowledge is very much like music for deaf ears.
I quickly did the math. I was planning to invest in learning a few hours a day for the next ~75 years of my life. Staring at the number of potentially wasted hours, I knew exactly what I had to do.
In the past six months, I’ve devoured dozens of books, research papers, and studies on how people learn. As a result, I’ve designed a learning process that works for me. It’s not perfect, but an order of magnitude better than what I had before.
In this work, I outline my workflow so that you can try it out. It applies to any subject or discipline, from programming to economics. If you stumble upon something where it doesn’t work, let me know.
Make it time-based, take regular breaks, and learn what you’re curious about.
The most important thing is that my learning is time-based, not goal-based. Setting learning goals such as “read X pages today” is a way to fail because you set up the wrong incentives. When you plan to read X pages by lunch, you can’t help but begin optimizing for the goal, which leads to focusing on speed instead of understanding. And when you don’t have those “aha” moments, it is hard to remember what you learn.
It’s also important to not overload yourself and take breaks. I do 3h learning sessions every day split into 30 min intervals with 5 min breaks. Breaks help to fall back into the diffuse mode of thinking and get access to a broader set of neural networks in my head. They also warm up my body, and I feel better after moving around for a few minutes.
As for material, I learn what I’m interested in. First, because life is too short to do things that you don’t love. Second, I’ve found that studying stuff I genuinely like awakens my curiosity. And curiosity is essential to develop mastery because mastery is about depth and breadth of knowledge.
How my learning session works
Clean up working memory, apply metacognition, and "siege" the thing with questions to improve understanding.
When I learn, I always have two devices on my desk. I have my laptop with the study material (ie, a book, a video, an article) on the right, and I have my iPad with a text editor open on the left.
This is how my current setup looks like:
When I begin learning, I set a timer for 30 minutes and create three files in Drafts:
- A file with a timestamp where my random thoughts go.
- A file with a timestamp where I think about the subject.
- A file with questions.
The first file is a mind dump. When I start learning, I immediately begin thinking about things. It’s almost as if my brain wakes up and starts throwing ideas, tasks, and memories at me. I suspect this comes from the associative memory because I present myself with many triggers when I’m learning; words and sentences that bear special meaning to me and invoke these ideas.
But here’s the problem. If I don’t write thoughts down, I can’t focus. My working memory is overloaded with todos, ideas, and emotions. You’ve probably experienced this for yourself – your mind is running too fast, and you can’t really concentrate on what you’re learning. Having this “dump” file is immensely useful to a) free up my working memory to focus on my learning instead of thinking about these things, and b) store these thoughts somewhere safe to go back to them later and take action.
The second file is where I write about what I’m learning. Folks in the kitchen call it metacognition, which means thinking about thinking. Metacognition is the single best trick I’ve found to improve understanding, and I will write more about it in the future. Whenever I don’t understand something or see that my understanding is shallow, I begin writing in the first person. It looks like this: “So Peter explains that there are four characteristics of a monopoly, but I don’t really understand why branding is one of them; why so?”
It’s also important to note that I don’t write in a usual sentence-paragraph manner. Instead, I write every thought on a new line. I don’t even put dots at the end of the sentences. This helps me to focus on understanding instead of nitty-gritty styling and typos. The “enter” key on a keyboard serves as the “end of thought” symbol and helps formulate ideas more clearly.
Another important idea is that my editor is plain text. I’ve found it incredibly liberating to operate in a plain text environment where you don’t have incentives to color, underline, bold, italicize, or do some other weird things with the text you’re writing. Instead of choosing the right font for my heading, I can focus on meaning instead. Also, my plain text app is way faster than all feature-rich text editors, and I’ve found it essential for a thought input environment to be fast. Otherwise, I can't think.
Here’s a fragment from my learning of React yesterday:
You’re probably thinking that it’s quite a bit of writing. It is. For an hour-long learning session, I usually do about 500-1000 words in this file. But it’s worth every character, and here’s why:
- When I apply metacognition, I understand things way better than when I don’t. I’ve tested many different learning modes and found metacognition to perform at least 2x better based on my later ability to recall and transfer knowledge. Also, there’s some research on metacognition as well.
- Having a file with my thinking about the subject keeps my working memory clean. I don’t feel overloaded as I usually feel after reading many articles at one go. You’ve probably experienced this yourself; your brain is almost melting after an hour of scrolling through the web. That’s because you present yourself with too much information without really making sense of it. After a few months of applying metacognitive practices, I realized that I can’t go back. It just feels so strange to experience that cognitive load again.
- Metacognition improves remembering through elaboration and interleaving. When I’m writing my thoughts in the file, I can’t help but begin connecting them with other ideas on that topic because of associative memory. And interleaving leads to mastery.
- (Speculation) Training metacognition improves my ability to transform vague notions and thoughts that I have during the day into specific words that I can write down for later analysis. This one is particularly interesting to me, but there’s no evidence besides my own experiments. And I might be biased because I’ve come up with this method.
Moreover, I type 2-3x faster than most people because I use shortcuts. So it’s not that bad.
The third file is questions. Whenever I stumble upon something that I don’t understand, I try to break it down into a set of simple questions. Each question in the group takes on a small part of the problem. If the concept is particularly challenging, I try to “siege” it with questions from many many different angles and break it down even further.
When I’m beginning a new session, I always start from the previous one’s questions file. I only look at questions and answer them before I’m beginning new learning. This doesn’t sound like very much fun, but it’s actually pretty interesting to explain stuff to yourself if you do it out loud. Answering questions improves my understanding and helps to connect ideas together. And yes, answering questions counts as learning – probably the most efficient learning you could be doing.
I'm not going into much detail on questions because Michael Nielsen has done a phenomenal job describing it here.
What happens after the session
Write a dense summary, provoke elaboration, interleaving, and transfer, and choose what to never forget.
After the session is done, and my three files are full of information, I begin the recap process.
First, I write a three to five sentence-long summary of what I’ve just studied. Here I try to distill the material’s core idea and compress the whole thing into a maximally dense chunk. When I’m summarizing, my laptop is closed. Not looking at the text helps to “compress” the idea to its core and make a small “hook” to my memory to later see what the whole book was about.
Here’s how my summary note looks like:
Very often, what I’m writing in the summary section is not what the text was about, but what it means to me. In other words, if both of us read this text and wrote a summary of it, mine would be very different than yours.
After I’m done with the summary, I write down the answers to three questions:
- What are the key ideas?
- How can I apply this knowledge that I learned?
- How do these ideas relate to what I already know?
The first question speaks for itself. I try to remember what I just read and write down as many ideas as I can bring back. When I began applying the metacognition trick that I mentioned earlier, I noticed a 3x increase in the number of concepts I could recall. And as I speculate that long-term memory recall is influenced by initial interleaving and recall, this might actually help to improve your long-term memory.
The second question is about transfer. The sole purpose of learning is to apply the knowledge that we learn. Without application, knowledge is useless. But to use my knowledge, I must remember it exactly when I need it. The transfer question helps to discover how to apply the knowledge I’ve just learned. Usually, it also uncovers some todos that I add to my inbox. For example, I was recently learning about monopolies, and the answer to this question was, “I can think what characteristics of a monopoly our idea has.”
The third question is a fallback for interleaving. It helps to proactively think about how this piece of knowledge that I just consumed connects to the broader picture of what I already know. The trick here is that when you answer this question, you not only make connections but improve your long-term memory because you recall that other thing that you connect the freshly learned knowledge to. And as creativity is just connecting things, having a habit of proactively making these connections helps think new thoughts.
Lastly, I take my file with questions and plug it into the Anki app. Anki is a spaced repetition system that I use for remembering things. It’s dead simple; you add flashcards and then test yourself. If you remember, it pushes the next test time further into the future.
The most exciting idea here is that using Anki changed my relationship with memory. I don’t fear forgetting anymore because I know that I will remember forever everything I add to Anki. It feels like a superpower. What’s more interesting, having this superpower provokes an unusual question, “What do I want to never forget?”
You can read more about Anki here.
What to do between learning sessions
Talk to friends about what you learn, apply spaced repetition to prevent forgetting, and build the daily recall habit.
Talking to friends
Whenever I learn some new cool thing, I feel an urge to explain it to my friends. This helps to quickly realize if I really know it or I just know the words. Sometimes I just message them, and sometimes I talk to them about an idea in a conversation.
Another trick I use to learn better is testing. Many people have this mistaken image of testing as something unpleasant. Like something you had to do at school. In reality, testing is very useful and fun, especially if done right. In some way, I'm actually learning through testing; I ask questions about what I'm reading even before I've read the answer and try to guess what the answer would be.
Between sessions, I use Anki whenever I have a minute to test myself. For example, yesterday, I did a few Anki questions when standing in line for groceries.
I also suspect that I'm utilizing a lot of spaced repetition during the workday because I always have my iPad with Drafts opened on the left, and apply a lot of metacognition. But that's a hypothesis.
Daily active recall
One more habit that I’ve found extremely useful is the daily recall. It’s the first thing I do every morning after I wake up and the last thing before I go to sleep.
Here’s how it works.
In the evening, I open a new file in Drafts and ask myself: “What do I remember from today?” In the first thirty seconds, it seems unbearably hard to recall even a few ideas. The mind is gone. But after a few minutes, I begin bringing back more and more. After a few months of deliberate practice, I can recall about 90-95% of ideas that I discover during the day, whether that’s how something in JS works, characteristics of a monopoly business, or why mixed practice improves remembering. And I have audacious speculation that doing this recall process twice a day actually trains my mental machinery to perceive and remember ideas better.
If you’re asking yourself, “why to bother so much?” here’s why.
So much of what we call creativity and intelligence is just memory. If you think you can look it up on Google, you’re wrong. The thought process is way faster than looking stuff up, and when you’re thinking about something or solving an important problem, you have to have your toolkit ready. Also, the most interesting ideas come when you’re not at your desk but showering, glazing over stars or wandering around in the city center. When you can’t Google. And this tiny 10-minute habit of recalling ideas is totally worth it if you think of the implications on a fifty-year timeframe.
Apply mixed practice and develop mastery by pattern-matching to complex environments.
When it comes to practice, I mix it up. Mixed practice means putting exercises at the end of the book, not at the end of the chapter. When we apply blocked practice (i.e., a chapter about subject X and exercises after the chapter about the same subject X), we trick ourselves into using way less cognitive power than if the problems are mixed. And the results are fascinating – medical students who had examples mixed up outperformed their classmates by 50%. Just to be clear on that, not some 5 or 10 percent improvement. 50%.
Example: mixed practice in learning how to code
Unfortunately, most coding schools and platforms haven’t bothered to look up cognitive science much. They all commit the same mimetic mistake of putting the subject on the left and exercises on the right. And this is even worse than blocked practice. When you do blocked practice, some forgetting occurs, and you recall things from the short-term memory. But on most coding sites, you just retype stuff from the left to the tiny window on the right. Monkey see monkey do. And it takes tremendous willpower to not do that when the example is on the left!
In other words, if you’re going through a gangster neighborhood with a giant pile of cash in your hands, the dumbfuck is probably you, not the gangsters who’ll gladly kick your ass and take the money.
I try to apply mixed practice by interleaving different ideas or ways of solving the problem. For example, if the task is to write a simple JS array sort of numbers, the expected solution is, well, to just write the sort. Instead, I create a function that takes in an array and returns a sorted one. Then I go ahead and make another one with, say, descended sort. I create an array of arrays and write a function that sorts all nested arrays. I apply all ways of sorting that I remember. Or I define a new array with words and letters and do alphabet sort. You got the idea.
Yes, it takes time. But I believe this is the time that matters most, not watching tutorials. At the end of the day, mastery is about pattern-matching that changes your perception to include things that had to be reasoned about before, so if you just keep practicing for decades, you’ll master anything you want.
To identify the best learning material, skim through one Amazon book sample every day.
How you learn matters. But another thing that matters is the source of knowledge, the artifact, that you learn from. And as I believe most people screw it up badly, I want to explain my approach here.
The idea is dead simple: if you want to find better learning material, what you have to do is, well, sample a ton of stuff.
You can’t trust the NYT bestsellers for two reasons. First, if they’re bestsellers, this means everybody reads them. And if you read what everybody reads, you will think what everybody thinks. Second, because the value of books is subjective. The same text can mean the world to me and nothing to you, and vice versa.
That’s why I’ve built a habit of reading one book a day. Here’s what I do:
- I go to Amazon and find a bunch of books I’m interested in. Usually, the books come from recommendations of thoughtful people who I trust. It’s usually a sufficient criteria to give a book a try. If many respectful people recommend a book, then it’s unlikely to be complete crap.
- I download samples for all these books. I believe book samples are the single most underrated thing in the world. They’re free and available for literally every book out there. Another thing I do is downloading books from Project Gutenberg; they’ve got a phenomenal free library of almost all old books worth reading.
- I have a folder in my Kindle app called “Want to Read.” All book samples that I download go into this folder.
- At 3 pm every day, I take an hour-long break from work. I go to Costa coffee, which is about a 10m walk away from where I live, and get myself a black Americano. I don’t take an iPad with me, only my phone.
- I open the first book sample from the Want to Read folder and begin reading. In my Drafts app on my phone, I create a new note with the book’s name that I’m reviewing. I usually start from the book description to understand what it’s about, and write down a couple sentences, trying to guess what’s coming up. Then I go straight to the table of contents and see if anything catches my attention there. If something does, then I jump ahead and begin reading the thing. If not, I start glancing through the book. I don’t speed read but glance at the first sentence of the paragraph and then skim through the rest of it. You might be wondering that when I’m reading like that, I don’t understand much. You’re right. But the purpose of this “picture walk” reading is different; it’s to catch my curiosity. And curiosity is different from understanding because it’s either happens or not. It’s like you’re seeing a pretty girl; you don’t really sit and reason, “well, do I really like her?”
- It takes me about 10-15 minutes to understand if the book worth reading in its entirety. Note that it’s tempting to list all books as “worth” reading. Don’t do that. If you’re not sure, delete the sample. As with girls, there will always be another one tomorrow and you will know when it’s really interesting.
- If the book is worth reading, I go ahead and buy it right now. If I’m not sure, then I delete it from the sample folder and forget about it.
- Repeat the same process daily.
By taking just 10 minutes a day to quickly scroll through the book instead of my Twitter feed, I end up having 365 books scrolled in a year. From these 365, about two end up being life-changing, ten – deeply interesting, and ten – sort of interesting. Everything else is noise. 94%.
Now, do the math. If I didn’t have this discovery routine and aimed to read books from cover to cover, I’d randomly sample about twenty books from 365 with meager chances of all of them being good. And I was doing that for years. So don’t repeat my mistake and invest 5-10 minutes a day to discover what you learn from, because learning is like eating. You become what you consume.
One thing you might have noticed is that the habit is tied up to a specific behavior. That’s the best thing you could do when building a new one because that’s how you guarantee you won’t forget it. You wouldn’t believe me if I said how many times I wanted to do something but simply forgot about it. Don’t rely on your shallow memory and tie your habit to some activity you’re going to do anyway. Taking a shower, eating a meal, brushing your teeth – all are good examples because they’re the things you have to do no matter what.
Also, you can set up a reward after you do the habit. For example, get yourself some coffee or go for a walk. For some reason, we don’t forget about our rewards as often as our habits, so by having a reward after you create even more opportunities for yourself to remember about doing the thing.
If you’re interested in reading up on habits, here’s a good book to start with.
To develop the learning habit, make it time-based, start really small, and do it every day.
When I started building my learning habit, I began at ten minutes a day. This doesn't seem like much, but that's precisely the point. If you hook it up to some must-do habit like taking a shower after work, then it won't seem like a big deal.
What's even more interesting, when I artificially lowered the time that I had to put in a newly born habit, I wanted it more because of the scarcity effect. My brain was like: "Well, we will only be learning for ten minutes. But I want to learn things because I want to be smart. So I better make the most out of these 10 minutes and not fuck around because otherwise, I won't have an opportunity to learn today."
But never postpone. If you fail to perform the habit in the morning, don't make up for it by cramming it into the evening. Because if you do, you set up a good excuse for your mind to trick you into snoozing more and more. It will begin seeming like it's better to postpone because "hey, we can always do it later!" Never snooze.
- One trick I’ve found extremely useful is deleting all potential sources of distraction from the device I use to learn. I went ahead and killed all messaging apps, all social networks, all email apps, and everything else that could be a potential source of distraction from my iPad. This made my life so much easier, and I’m still wondering why I hadn’t done this earlier.
- Another trick is environment design. When I begin learning, I leave my phone in a different room. That way, I’m not tempted to use it just because it gets into my perception. This trick saved me a ton of time and energy, so I welcome you to try it out as well. As you may have guessed correctly, this extends way beyond learning; I apply the same approach to my work. When I sit down to write, I leave all potential distractions (including my wife) in a different room. Immensely helpful.
- Metacognition works not just for learning. When I’m writing code or designing a process, I find it incredibly liberating to have a place for explaining what I really want to do here instead of just bashing the keyboard.
- When I'm facing an idea that I don't understand, I draw things. Sketching takes full advantage of spatial cognition and helps to break down the most challenging concepts.
- When I built a process for learning something new, I stopped being afraid. I know it’s only a matter of time until I master the thing, and this encourages me to take on ambitious projects that I’d not dare to think of before.
Does this apply to reading articles or watching videos?
The learning process applies to everything that you learn. If you read an article to understand it and use the knowledge from it later, yes. If it’s a fiction story or a newspaper article, then it’s probably overkill.
How long does it take to master the process?
Less longer than you think to apply and begin getting benefits, and way longer to master. I’m still learning!
Does the process apply to learning how to code?
Of course! This process was designed to figure out how to teach me to code. A big deal of intelligence is just knowledge, and knowledge comes from changes in long-term memory. If you’re learning how to code, ping me at email@example.com, and I’ll share additional tips.
Can I use this process if I don’t have a second device, read from my phone, or use physical books?
Yes. I’ve experimented with a single device study mode and found it to be quite productive. It’s less superior than the double-device process in terms of metacognition but better in terms of focus. If you minimize your kindle app on the phone and open the notes app, you have to bring those things back from your mind instead of reciting them from the book. In other words, this mode has better incentives for immediate recall.
Here are the most interesting works I’ve found in the past six months in no specific order:
- A Cognitive Perspective on How People Learn: Implications for Teaching, by Geoff Norman, Ph.D.
- How People Learn, by John D. Bransford, Ann Brown, and Rodney R. Cocking
- Learning How to Learn, by Barbara Oakley and Terrence Sejnowski
- The Art of Doing Science and Engineering: Learning to Learn, by Richard Hamming
- Make It Stick, by Mark A. McDaniel and Peter C Brown
Thanks for reading.