You finish a chapter on a topic you genuinely care about. You highlight a few lines, nod along with the author, and close the book feeling sharper. Then a friend asks what it was about, and you produce a vague paraphrase that feels embarrassingly thin. The gap between "I read it" and "I could explain it" is where most learning quietly fails.
Richard Feynman, the physicist who shared the 1965 Nobel Prize for his work on quantum electrodynamics, was famous among his colleagues for refusing to accept jargon as understanding. When a student handed him a complicated answer, he often pushed back until the explanation became simple enough for a curious teenager. The technique that bears his name is a distillation of that habit. It is not something Feynman wrote up as a method; it is reverse-engineered from how he learned and taught. The mechanics are plain, and the payoff is large.
The four steps
The technique has four moves. Done in order, they convert a fuzzy sense of understanding into something you can defend.
Step 1: Choose a concept and write down what you already know
Pick one specific idea. Not "machine learning" but "how gradient descent updates weights." Not "evolution" but "why sickle-cell anemia persists in malaria-prone regions." Take a blank page and write down everything you currently believe to be true about it, without checking any source.
This step matters because it forces you to commit. The act of writing turns a vague mental fog into specific claims that can be right or wrong. Cognitive scientists call the comfortable haze you feel before this step the illusion of competence — the sensation of knowing something you cannot actually reproduce. Reading and re-reading inflates that illusion. Writing pops it.
Step 2: Teach it as if to a 12-year-old
Now explain the concept out loud or on paper, as if your audience is an intelligent 12-year-old who has never encountered it. Use plain words. Build from concrete pieces to abstract ones. Where possible, do this with an actual person — a partner, a sibling, a patient friend.
This is the hardest step and the most useful one. Teaching is a form of retrieval practice: pulling information out of your head, rather than letting it pass through your eyes again. Decades of memory research show retrieval strengthens recall more than any amount of re-reading. The 12-year-old constraint is doing extra work. Jargon lets you skim over your own confusion. Plain language doesn't. The moment you reach for "essentially, the system optimizes itself," you have flagged a place where your understanding ends.
Step 3: Identify the gaps and return to the source
While teaching, you will hit moments where your sentence trails off, or where you reach for a phrase you don't fully mean. Mark these. They are the entire point of the exercise. Then go back to the textbook, the paper, the lecture, the documentation — whatever you originally learned from — and look up only those gaps. Not the whole chapter. Just the parts you couldn't reproduce.
This targeted re-reading works because your brain is now primed. You have a specific question, which means the answer lands in a labeled slot rather than swimming around with everything else. The general principle: confusion that you have located is a hundred times more useful than confusion you haven't.
Step 4: Simplify with analogies
Once the gaps are filled, rewrite your explanation. Strip the jargon you smuggled back in. Where an idea is genuinely complex, build an analogy that connects it to something the reader (or the 12-year-old) already understands. A good analogy is not decoration; it is a bridge between an unfamiliar concept and a chunk of knowledge already wired into long-term memory. That bridge is what makes the new concept stick.
A worked example: compound interest
Pretend you have read about compound interest several times but want to actually understand it. Here is one full pass.
Step 1, brain dump: "Compound interest is interest that earns interest. The formula has something to do with multiplying. It grows faster than simple interest. Einstein supposedly called it the eighth wonder of the world. The longer you leave money invested, the more dramatic the effect."
Step 2, teach a 12-year-old: "Imagine you put $100 in an account that pays 10% a year. After year one, you have $110. Here is the trick: in year two, you don't earn 10% on the original $100. You earn 10% on the new total, $110. So you get $11 instead of $10. In year three, you earn 10% on $121. The interest itself starts earning interest, and the amount you gain each year keeps growing on its own."
The gap surfaces: The 12-year-old asks, "So how much would $100 be worth in 30 years?" You realize you don't actually know how to compute it without a calculator, and you are not sure why the formula uses an exponent. You have been pattern-matching on the word "exponential" without understanding the mechanism.
Step 3, return to source: You look up the formula: A = P(1 + r)t. The exponent is there because each year multiplies the previous total by (1 + r). Thirty years means thirty multiplications, which is what an exponent encodes. At 10% over 30 years, $100 becomes about $1,745.
Step 4, simplify with an analogy: "Compound interest is a snowball rolling downhill. After each rotation, the snowball is bigger, so the next rotation picks up more snow than the last. Simple interest would be a snowball that only ever picks up the same amount per rotation. The longer the hill, the more dramatic the difference — not because the snow is special, but because each rotation builds on the one before it."
You now understand compound interest in a way that survives a question you didn't anticipate. That is the test.
Where this usually goes wrong
Two failure modes account for most botched attempts.
- Skipping the brain dump. People jump straight to step 2 without first writing down what they think they know. They end up half-teaching, half-recalling, and the gaps blur together. Step 1 takes five minutes and changes everything.
- Stopping at "I get the gist." The technique only works if you push past the moment your inner voice says "yeah, basically, it's like…" That phrase is the sound of the illusion of competence reasserting itself. Keep going until you can say it cleanly, in full sentences, without hedging.
A third, subtler trap: aiming for a polished explanation on the first pass. Don't. Step 2 is reconnaissance. Its job is to find what you can't explain, not to produce something publishable. Treat the messy first attempt as data.
Understanding is step one. Keeping it is step two.
The Feynman Technique solves the comprehension problem. It does not solve the forgetting problem. A concept you understood deeply on Tuesday will start to fade by Friday and feel foreign in a month, unless you do something deliberate to retain it.
The complementary tool is spaced repetition: reviewing material at expanding intervals, with active recall, just before you would otherwise forget. The same retrieval practice that powers step 2 of Feynman, applied over time, is what converts hard-won understanding into durable knowledge. The natural workflow is to use Feynman to get a concept clean, then turn the cleaned-up explanation into a few high-quality recall prompts and review them on a schedule.
This is what tools like Neuraknow are designed for — turning the concepts you have already understood into a review schedule that keeps them accessible months and years later. Comprehension and retention are different problems. Feynman is the right tool for the first. Spaced repetition is the right tool for the second.
Pick one concept you have been meaning to actually understand. Spend twenty minutes with a blank page and the four steps. The next time someone asks what you have been reading about, you will notice the difference in your own voice.