필사 모드: Working Memory and the Brain Circuits — How to Use Your Brain to Learn Efficiently

Opening: I Tried to Hold Everything in My Head and Lost All of It

As a junior developer, I wanted to be someone with a good brain. The decisions made in meetings, the system architecture a senior taught me, the clues I found while debugging — I tried to keep all of it inside my head alone. I think I worried that taking notes would somehow make me look incompetent.

The result was disastrous. By the time a meeting ended, half of it was gone from my memory. The debugging clue I had worked so hard to find had vanished by the time I came back from lunch. I even asked a senior the same question twice. I thought I simply had a bad brain, and I felt defeated.

But after studying a bit of cognitive psychology, I realized that it was not a problem with my brain but with the structure of the brain itself. Human working memory is far narrower than we think. Trying to cram everything into that narrow space was, in itself, the wrong strategy.

This piece is not about "how to become smarter" but about "how to use the brain you already have efficiently." Once you understand how the brain works, you can learn and remember far more with the same effort.

The Limits of Working Memory: A Narrow Workbench

What Is Working Memory

Working memory is the ability to briefly hold and manipulate information in your head at this very moment. In computer terms, it is less like a disk and more like a small cache or register. Its capacity is small, and when new information comes in, the existing information is easily pushed out.

The classic research by psychologist George Miller proposed that the number of chunks of information a human can handle at once is roughly seven, give or take. Later studies suggest the number may be even smaller. Either way, the core point is clear: our workbench is very narrow.

Cognitive Load Theory

The Cognitive Load Theory of educational psychologist John Sweller offers insight into how this narrow workbench should be used. Cognitive load is broadly divided into three kinds.

- Intrinsic load: the inherent difficulty of the material you are trying to learn

- Extraneous load: the unnecessary burden created by the way information is presented

- Germane load: the good load spent on turning new knowledge into a long-term memory structure (a schema)

When learning does not go well, it is usually because the extraneous load is too large. Disorganized information, a distracting environment, too many variables to recall at once — these fill the workbench and leave no room for the learning itself.

| Type of Load | Identity | Response |

| --- | --- | --- |

| Intrinsic load | The difficulty of the material itself | Break it into small pieces |

| Extraneous load | A poor mode of presentation | Organize and simplify |

| Germane load | The effort of building a schema | Focus here |

Externalization: Get It Out of Your Head

The Easiest Way to Widen a Narrow Workbench

If working memory is narrow, the answer is simple. Do not pile everything onto the workbench; instead, place one more large desk beside it. That large desk is exactly what externalization is. Notes, documents, diagrams, to-do lists, code comments — all of them lighten the burden on working memory.

I ended the era of trying to hold everything in my head and began writing everything down. During meetings I took notes on decisions, while debugging I recorded the hypotheses I tried and their results, and when understanding a system I drew diagrams myself. Surprisingly, it felt as if my memory had improved. In truth, all that had happened was that I needed to remember less.

Another Effect of Externalization

The very act of writing aids understanding. A thought that is hazy in your head reveals its gaps when you try to put it into words. When you try to draw a diagram, you can see where you truly did not understand. Externalization is not mere storage but a tool for making thinking clear.

Examples of externalization tools

- Decisions: meeting minutes, architecture decision records (ADR)

- In-progress thinking: debugging logs, hypothesis notes

- Structural understanding: diagrams, system sketches

- Tasks: task lists, kanban boards

- Long-term knowledge: a personal wiki, organized notes

Turning It Into a System

Writing something down once is not enough. Externalization is only complete when you can find what you wrote again. I built a searchable note system and developed the habit of recording in a fixed place and in a fixed way. If you spend working memory just trying to remember where you wrote something, you have put the cart before the horse.

Retrieval Strengthens the Circuit

The Power of Pulling It Back Out

Here lies an important balance. Externalization lightens the burden on working memory, but if you rely entirely on the outside for everything, then nothing is left in your own brain. Real learning happens during retrieval — the process of pulling information back out.

One of the well-established phenomena in cognitive psychology is the retrieval practice effect, also known as the testing effect. Rather than reading the same material several times, it is far more effective for long-term memory to read it once, then close the book and try to recall it on your own. The research of Henry Roediger and his colleagues, and the book 'Make It Stick' that popularized it, demonstrate this clearly.

The Metaphor of a Circuit

Comparing the brain's learning to a circuit makes it easy to understand. Each time you recall a piece of information, the neural path leading to that information is strengthened a little. The more often you pull it out, the wider the path becomes, and eventually it comes to mind without effort. Conversely, information you never pull out has a path that grows fainter and fainter.

That is why "reading and rereading" is inefficient. Reading is merely the act of putting information in; it is not the act of widening the path. To widen the path, you have to pull it out yourself.

How to Build Retrieval Into Daily Life

- After learning something, close the book and try to summarize the key points yourself

- Explain it to a colleague. Teaching is the best form of retrieval.

- Make a simple self-quiz and solve it a few days later

- Review with spacing (distributed learning). Pulling it out again right around the time you start to forget is effective.

The learn to externalize to retrieve cycle

1. Learn something new

2. Externalize the key points as notes

3. Close the notes and try to recall them yourself (retrieval)

4. Recheck only the parts where you got stuck

5. Try to recall it again a few days later (spacing)

The Brain Does Not Spend Energy Where It Is Not Needed

The Difference Made by Urgency and Context

For a long time I struggled to memorize English vocabulary, but it just would not stick. Yet once I actually started working with foreign colleagues — once I was in situations where failing to understand in a meeting caused real trouble — my vocabulary grew quickly. They are the same words, so why such a difference?

The brain is an organ that pursues efficiency. It conserves energy on anything not needed to survive and function. So information that carries no urgency and no context is poorly stored. Conversely, information that is needed right now and connected to the context of my own life stays strongly.

The Importance of Meaning and Connection

The more you connect new information to what you already know, the more solid the memory becomes. This is called elaboration. Rote memorization is like creating an isolated island; elaboration is like connecting that island with a bridge.

When I learn a new concept, I always ask, "What that I already know is this similar to?" The very way this piece compares working memory to a cache, externalization to an auxiliary desk, and the learning circuit to a path is itself that elaboration. A metaphor is a bridge that attaches new knowledge to a familiar context.

Manufacturing Urgency Artificially

True urgency is hard to control, but a weaker form of urgency can be created.

- Create a situation where you must soon use what you learned (a presentation, writing, a small project)

- Set a small deliverable with a deadline

- Apply what you learned directly to a real problem

- Share it publicly. The fact that someone is watching becomes a weak form of urgency.

Chunking: Bundling Small Pieces Into Large Lumps

Grouping by Units of Meaning

Another way to go beyond the limits of working memory is chunking. When you bundle scattered small pieces of information into a single meaningful lump, the space it occupies on the workbench shrinks to one.

It is easy to see if you think of a phone number. If you memorize an eleven-digit number one digit at a time, the workbench fills up; but if you bundle it into three lumps, it shrinks to three items. The reason a chess master can remember the whole board at a glance is that they see it not as individual pieces but as lumps of familiar patterns.

Expertise Means Having Larger Chunks

The difference between an expert and a novice is not memory capacity but the size of their chunks. A skilled developer recognizes "this code is the observer pattern" as a single lump, while a novice looks at it line by line. They use the same workbench, but because the expert bundles in larger units, they can handle more.

Learning is, in the end, the process of accumulating larger and more useful chunks. That is why solidly mastering basic patterns matters. The more good chunks you have, the faster you can learn new things.

The growth of chunks

Novice: a-t-o-m-i-c (6 letters)

Intermediate: atom + ic (2 lumps)

Skilled: atomic (1 lump, connected even to its meaning)

The Trap Called Multitasking

We Cannot Do It Simultaneously

The fact that working memory is narrow also shatters the illusion of multitasking. What we call multitasking is actually closer to rapid task switching. And switching comes with a cost.

Every time you change tasks, you have to clear the narrow workbench and fill it with a new context. If you are writing code, check a messenger, and then come back, you have to reload the variables and flow that were in your head moments ago. When this reloading cost accumulates, it feels as if you did several things at once, but in reality you did none of them deeply.

One Thing at a Time

The solution is simple but hard. Focus on one thing at a time. Before I start a task, I try to close other windows and notifications and give the workbench to a single thing alone. The value of deep work is exactly what Cal Newport emphasized in 'Deep Work'. A focused few hours produce far greater results than many scattered hours.

| Approach | Appearance | Actual Result |

| --- | --- | --- |

| Multitasking | Busy and efficient | Frequent reloading, shallow results |

| Single focus | Looks slow | Deep understanding, fewer mistakes |

Traps and Balance

Do Not Rely on Externalization Alone

As I said earlier, if you leave everything to your notes, no circuit gets built in your brain. Lighten the burden through externalization, but the core must be made your own through retrieval. Balance matters.

Do Not Obsess Over Tools

People often wander in search of the perfect note app or the perfect learning system, and the learning itself gets pushed to the back. The simpler the tool, the better. What matters is not the tool but the principles: externalization, retrieval, and chunking.

Efficiency Is Not the Goal

I have talked about how to use the brain efficiently, but not all learning has to be efficient. Sometimes aimless wandering exploration gives rise to unexpected connections. Efficiency is a good tool, but it cannot replace the driving force of curiosity and enjoyment.

70-20-10: Most Learning Comes From Experience

The Limits of Desk-Bound Learning

I once believed that if I just read enough good books and watched enough good lectures, my skills would grow. Yet no matter how much I read, until I actually put my hands on it, something felt as if it were floating in the air.

The 70-20-10 model, often cited by leadership research institutes and others, explains this experience. It is the observation that a person's growth comes roughly 70 percent from real experience and challenging tasks, 20 percent from interaction with others and feedback, and 10 percent from formal learning. There is debate over the exact ratio, but the core message is clear: most learning comes from grappling with things directly.

This also connects to the "urgency and context" discussed earlier. Real tasks carry urgency and context. That is why the brain spends energy. The reason desk-bound learning does not stick well is that it lacks urgency and context.

| Learning Mode | Share (approximate) | Characteristics |

| --- | --- | --- |

| Challenging experience | About 70 percent | Strong urgency and context |

| Social learning | About 20 percent | Feedback and observation |

| Formal learning | About 10 percent | Organizing concepts and fundamentals |

A Bridge to Using What You Learned Immediately

So whenever I learn something, I apply it to a small task as quickly as possible. If I watched a lecture, I build a short example myself; if I read a book, I solve one real problem with that concept. I intentionally lay a bridge that connects formal learning (the 10) to experience (the 70).

Active Learning: The Hands Teach the Head

The Trap of Passive Input

When we nod along while listening to a lecture, we often deceive ourselves into thinking "I understood it." But nodding is not retrieval. Information coming in and being able to pull it back out are entirely different things. This gap is precisely the illusion of fluency.

Active learning studies, well known from physics education research, repeatedly show that actively solving problems, discussing, and explaining clearly improves learning outcomes more than passively listening to a lecture. Learning that moves the hands and mouth sticks better in the head.

Small Devices for Creating Active Learning

- Pause the lecture and first predict what content will come next

- After reading a paragraph, summarize the key points aloud without looking

- Make your own questions using the concept you learned

- Explain it as if teaching a colleague

The passive to active shift

Passive: watch the lecture all the way through

Active: pause at each segment and summarize, predict, and question yourself

Effect: same time, a deeper circuit

Sleep and Spacing Harden the Circuit

It Takes Time to Harden

A circuit does not become solid the moment you make it. It needs time to set. As I covered in an earlier piece, sleep is an active process that organizes and consolidates what you learned during the day. This is the reason that pulling an all-nighter to learn everything at once is worse than learning over several days and sleeping deeply in between.

Distributed learning (spacing) works on the same principle. Cramming everything in one sitting seems effective right after the exam, but over time it disappears rapidly. Information that you pulled out several times with intervals stays far longer. The slight difficulty of pulling it out again right around the time you start to forget actually makes the circuit even more solid.

Desirable Difficulty

Psychologist Robert Bjork calls this "desirable difficulty." If it is too easy, the circuit is not strengthened; if it is too hard, you get frustrated. Learning happens best at the point where you pull it out with just the right amount of effort. This is the reason review that reads smoothly has little effect, and a slightly challenging self-quiz has a large effect.

Cramming vs distributed learning

Cramming: 4 hours in one sitting in a day, looks good in the short term, disappears fast

Distributed learning: 1 hour each over 4 days plus enough sleep, stays long

Learning by Teaching: Explaining Is the Best Retrieval

To Explain, You Have to Understand

When I want to know whether I have truly understood something, I try to explain it to another person. I felt I knew it in my head, but when I actually try to explain it aloud, the points where I keep getting stuck reveal themselves. Those points where I get stuck are exactly the parts I did not really understand.

There is a phenomenon known as the "protégé effect." It is the finding that when you study with the expectation that you will teach someone, you understand more deeply and remember longer than when you study merely for yourself. Teaching is the most powerful form of retrieval and the most honest form of self-check.

The Learning Version of Rubber Duck Debugging

Think of "rubber duck debugging," familiar to developers. It is the half-joking technique where, by explaining a problem to a rubber duck, you find the answer yourself. The same principle works in learning. It does not have to be a person; simply explaining a concept to a blank screen or a notebook organizes the circuit.

Checking through explanation

1. Choose a concept you just learned

2. Explain it aloud without looking at anything

3. Mark the points where you get stuck

4. Relearn only those parts

5. Try to explain it again

The Environment Determines Cognitive Load

An Environment That Clutters the Workbench

Even with the same brain, the slack in working memory varies greatly depending on the environment. A messy desk, constant notifications, and countless windows opened all at once increase extraneous load and steal the space meant for learning. When I cannot focus, the first thing I do is tidy my desk and screen. Physical tidiness often leads to cognitive slack.

The Conditions for a Good Environment

- Few distractions in your field of view

- Notifications that do not interrupt the task

- Only one task displayed on screen at a time

- Frequently used information kept in a fixed place

| Environment | Extraneous Load | Slack for Learning |

| --- | --- | --- |

| Cluttered and full of notifications | High | Narrow |

| Tidy and quiet | Low | Wide |

Organizing the environment is a device that works even on days when your willpower is weak. Rather than trying to defeat distraction by willpower every time, it is far more efficient to remove in advance the cracks through which distraction creeps in.

A Small Anecdote: The Same Lecture, Different Results

A Lecture Watched Twice

I once watched the same online lecture twice. The first time, I just hit play and watched it to the end. After finishing, I felt accomplished, but a few days later almost nothing was left. I had fallen squarely into the illusion of fluency.

The second time, I changed my approach. Each time a segment ended, I paused, and without looking at the screen I wrote the key points into a notebook (externalization). And the next day, I closed those notes and tried to recall them again (retrieval). A few days later, I built a small example myself with the concept I learned (experience). It was the same lecture, but what was left from the second approach was beyond comparison.

The difference was not the brain but the method. The first time I only put information in; the second time I externalized, pulled it out, and applied it. Same amount of time, an entirely different circuit was left behind.

Interleaving: Mixing It Up Makes Learning More Solid

The Trap of Digging a Single Well

Block learning — fully mastering one topic before moving to the next — seems intuitively right. Yet many studies report that interleaving, mixing and alternating between similar topics, can be more effective in the long run.

The reason connects to the retrieval and the desirable difficulty discussed earlier. When you mix topics, you have to recall each time, "Which method should I use this time?" This small difficulty makes the circuit more solid. Conversely, if you dig only a single well, you mechanically repeat the same method, and the very ability to choose "when to use what" does not grow.

How to Apply Interleaving

- Practice by mixing problems that are similar but of different types

- Before perfectly finishing one thing, alternate to look at others too

- When reviewing, weave in things you learned in the past

Block vs interleaving

Block: AAAA BBBB CCCC (looks easy but stays less)

Interleaving: ABC ABC ABC (looks hard but stays more)

That said, interleaving is more effective after the fundamentals are reasonably in place. When first grasping a concept, it is better to focus on one thing at a time, and once you are somewhat used to it, mixing is good. Like all tools, it must be used to fit the context.

Summary: Five Principles for Using the Brain

Let me bundle the story so far into five principles. These principles are not separate from one another; they flow as a single stream.

1. Externalization: lighten the burden of the narrow workbench by getting it out of your head.

2. Retrieval: widen the circuit through the act of pulling it out again.

3. Chunking: bundle small pieces into meaningful lumps.

4. Focus: give the workbench to one thing at a time.

5. Context and urgency: create a reason for the brain to spend energy.

| Principle | One-Line Summary | Representative Practice |

| --- | --- | --- |

| Externalization | Get it out of your head | Notes and diagrams |

| Retrieval | Pull it out again to widen the path | Self-quiz and explaining |

| Chunking | Bundle into larger lumps | Repeating basic patterns |

| Focus | One thing at a time | Turning off notifications |

| Context | Create a reason to use it | Immediate application and sharing |

These five require no special talent. They are principles that apply to anyone's brain, and you can start small this very day.

Frequently Asked Questions (FAQ)

If I take a lot of notes, doesn't my memory get worse?

The pure burden of memorization decreases, but if you use the workbench space you freed up for understanding and retrieval, you end up remembering even more deeply. The problem is only when you take notes and never pull them out again.

It feels like just reading is enough to study, isn't it?

When reading, it is easy to deceive yourself into thinking you know it because it feels familiar (the illusion of fluency). Closing the book and trying to recall on your own reveals what actually remained.

Aren't there people who are good at multitasking?

Most research shows a large gap between how good people believe they are at multitasking and their actual performance. Feeling good at it and the actual result are different.

How do I grow chunks quickly?

There is no shortcut. Chunks grow in the process of repeatedly mastering basic patterns and applying them yourself. Retrieval and distributed learning are the key here too.

Closing: Instead of Becoming Smarter, Learning to Use It Better

I once wanted to have a better brain. But now I think differently. What we need is not a larger workbench but a way to use a narrow workbench well.

Getting it out of your head to lighten the burden (externalization), pulling it out again to widen the path (retrieval), bundling small things into large lumps (chunking), and focusing on one thing at a time. And creating urgency and context so the brain spends energy. These ordinary principles were closer to the secret of the "smart person" I so envied in my junior days.

The brain does not wait for us to become smarter. It merely responds to how we use it. Closing the book and recalling once what you learned today — start from there.

References

- George A. Miller, The Magical Number Seven, Plus or Minus Two, Psychological Review, 1956

- John Sweller, research related to Cognitive Load Theory, https://www.ncbi.nlm.nih.gov/

- Peter C. Brown, Henry L. Roediger III, Mark A. McDaniel, 'Make It Stick', Belknap Press, 2014

- Cal Newport, 'Deep Work', Grand Central Publishing, 2016

- Henry L. Roediger III, Jeffrey D. Karpicke, Test-Enhanced Learning, https://pubmed.ncbi.nlm.nih.gov/16507066/

- James Clear, The Spacing Effect, https://jamesclear.com/spaced-repetition

- Harvard Business Review, A Better Way to Learn, https://hbr.org/2016/10/learning-is-a-learned-behavior-heres-how-to-get-better-at-it