How Does the Feynman Technique Work for Studying?

The Feynman technique is a four-step self-test: pick one concept, explain it in plain words as if teaching a beginner, mark where you stall, then go back to the source to fill those gaps and simplify. It works because trying to explain exposes holes that feeling you understand hides. It's a popular framing, not a method Feynman himself named.

Why Can You Nod Along and Still Not Be Able to Explain It?

Because following an explanation feels the same as understanding it, and reading only ever tests the feeling.

You finish the article. It was good. Every paragraph made sense as you went, you nodded in the right places, you may even have said "huh" out loud at the clever bit.

Twenty minutes later someone asks what it was about. You open your mouth and hear yourself say, "Basically it's about how, um... it's kind of complicated."

That "um" is the whole problem.

You weren't distracted. You didn't skim. You understood it, in the only way you had of checking at the time, which was that it felt understood. And that feeling is exactly the thing that's off.

It happens with books you spent a weekend on. It happens with the YouTube explainer that made everything click, right up until a question phrased slightly differently and the click was gone. It happens in lectures, where you follow every step on the board and then sit in front of the problem set with nothing. The worst version is the quiet one: you never get asked, so you never find out, and you carry the sense of knowing it around for months.

If you've read this far wondering whether you're just bad at studying, you can put that down. There's an experiment on exactly this, and the people in it weren't bad at studying either.

Here's a quick way to feel it rather than take my word for it. How well do you understand how a flush toilet works? Give yourself a number from 1 (no idea) to 7 (could take one apart and explain every part). Hold on to it.

Two psychologists at Yale, Leonid Rozenblit and Frank Keil, ran that experiment with a flush toilet among the devices. In their first study, sixteen graduate students rated how well they understood a list of 48 devices and phenomena, on a 1 to 7 scale. Then each was asked to write a detailed, step-by-step explanation of four of them: a speedometer, a zipper, a piano key and a flush toilet for some, a cylinder lock, a helicopter, a quartz watch and a sewing machine for the rest. Then they rated themselves again.

Nearly everyone's rating dropped.

Afterwards they were asked a pointed question, such as how you could pick a cylinder lock, step by step. Then they read an expert explanation. Their own paper notes that many participants reported "genuine surprise and new humility at how much less they knew than they originally thought."

It wasn't a general loss of nerve, either. After reading the expert explanation, the same people rated their new understanding higher. They'd lost confidence in the particular thing they'd just failed to explain, and nothing else.

Rozenblit and Keil called it the illusion of explanatory depth. The first line of their abstract says people "feel they understand complex phenomena with far greater precision, coherence, and depth than they really do." Two details further down matter if you study for a living, or just read a lot.

First, the illusion is "far stronger for explanatory knowledge" than for facts, procedures or narratives. You rarely fool yourself about a capital city, because you can produce London or you can't. A procedure has an end state: a cake, a tied bow tie. An explanation has no obvious end. There's never a moment where you bump into the wall, so you never learn where it is.

Second, they found the illusion is strongest where the environment shows you the mechanism in real time. The toilet's parts are right there in your head as a picture, and the picture props up the feeling.

Here's an extension of that, and it's mine, not theirs: their paper is about devices, not reading. But an article is about as supportive an environment as exists. Every claim is on the page in front of you. All you have to do, sentence by sentence, is recognise it. Recognising is easy. Producing it with the page gone is a different act, and that's the one you get asked to do on Monday.

So the nod-along-then-blank isn't a flaw in you or in your attention. It's a gauge problem. The only gauge you had, "does this feel clear?", reads high on exactly this kind of material, for Yale graduate students and everyone else. Same toilet. Same result.

Which means you need a different gauge. One that doesn't ask how it feels.

Go back to your number for the toilet. Keep it. We'll come back to it.

Where Does the Feynman Technique Actually Come From?

It comes from learning writers, not from Feynman. The physicist never published a four-step study method.

The learning writer Scott Young popularised the label, and his own 2024 post on the method says his version is "only loosely based on Richard Feynman's practices." That's a more honest account than most of the pages ranking for it give.

There are two real Feynman stories underneath, and they're worth having straight.

The first is from the physicist David Goodstein, who told it in a 1989 Physics Today piece and again in his book Feynman's Lost Lecture. Goodstein asked Feynman to explain why particles with spin one-half obey Fermi-Dirac statistics. Feynman said he'd prepare a freshman lecture on it. A few days later he came back and admitted he couldn't reduce it to freshman level, which, he concluded, meant physicists didn't really understand it.

The second is in James Gleick's biography Genius. Preparing for an oral exam as a young physicist, Feynman opened a fresh notebook and wrote on the title page: "Notebook Of Things I Don't Know About."

Neither of these is a study method with steps. Both are the same instinct: if you can't say it plainly, you don't have it yet. The four steps are a sensible packaging of that instinct, and I use them below. But they're a packaging, which matters when we get to how well the whole thing is tested.

What Are the Four Steps?

The four steps are: pick one concept, explain it plainly, mark and fix the gaps, then simplify and test it again.

Step	What you do	What it catches
1. Pick one concept	Write its name at the top of a blank page. One idea, not a chapter.	Scope that's too vague to test. "Understand inflation" can't fail. "Why do prices rise when..." can.
2. Explain it plainly	From memory, in short words, as if to a bright twelve-year-old. Full sentences, not bullet fragments.	Jargon doing the work that understanding should do.
3. Mark the gaps	Underline every spot where you stalled, said "it just", or skipped a step. Go back to the source for those spots only.	The illusion itself.
4. Simplify and test again	Rewrite using shorter words and one analogy. Then ask a "what if" question the explanation should be able to answer.	Explanations that work only for the exact case you studied.

Two details carry most of the weight.

The explaining happens without the source open. The moment you can see the page, you're back in the supportive environment that produced the illusion, and you're recognising again instead of producing.

And step 3 is the one that does the work. Step 2 alone is just writing down what you already feel you know. The value is in the underlines.

What Does a Worked Example Look Like?

Here's the technique run on the flush toilet, the same concept as the experiment.

Step 1. Concept: how a flush toilet works.

Step 2. A first draft, the kind you'd write in twenty seconds, for a twelve-year-old:

When you press the handle, a plug at the bottom of the tank lifts. The water in the tank rushes down into the bowl and pushes everything down the pipe. Then the tank refills and the plug drops back.

It reads fine. It has a beginning, a middle and an end. That's what makes it dangerous.

Step 3. Now push on each sentence with "what exactly?" and underline where it gives way.

What exactly keeps water in the bowl before you flush? Nothing in the draft says. The bowl is open at the bottom, so why doesn't it just drain? The answer, per Wikipedia's article on the flush toilet, is that the pipe leaving the bowl is a bend that rises behind the bowl before it descends. The top of the bend limits how high the water in the bowl can stand. The water left sitting there also blocks sewer gas from coming back up.

What exactly does "pushes" mean? This is the big one. In a siphonic toilet (washdown bowls skip this and rely on force alone), the surge from the tank fills the bend. Once the bend is full of water, it behaves like a hose you've filled and dropped over a ledge: water leaving the far end drags more in behind it. According to Wikipedia, it pulls nearly all of the bowl's contents down the drain in roughly four to seven seconds. The tank water isn't the broom. It's the starter motor.

What exactly stops it? The draft never ends the flush, it just stops talking. In a siphon, the flow carries on until the bowl runs low enough that air gets into the bend and breaks the column of water. That air is the gurgle at the end.

What exactly tells the tank to stop refilling? A float on the fill valve. As the water rises, so does the float, until its arm closes the valve.

One more, and it's a bit awkward. If you're in the UK there's a decent chance your cistern has no plug at all. Wikipedia describes the British siphon-type cistern: lifting the lever raises a disc that pushes water over the top of a siphon inside the cistern, which starts the flow. The first sentence of the draft, the one that felt the most obvious, is the North American picture.

Five underlines in a three-sentence draft.

Step 4. Rewrite shorter, with an analogy, and ask a "what if":

A toilet doesn't push waste away. It gets a bend in the pipe to pull it. Pressing the handle dumps tank water into the bowl faster than the bowl can drain, which fills the bend. A full bend works like a primed hose: water leaving one end drags more in behind it. That keeps going until the bowl gets low and air slips into the bend. Gurgle, and it's over. A float shuts off the tank's water when it's full again.

The "what if" test: why does a toilet flush badly if you pour the water in slowly? If the explanation is right, the answer falls out of it. Slow water never fills the bend, so no siphon starts. If it doesn't fall out, there's another gap.

Now back to your number. Whatever it was, the draft above is roughly what a 6 feels like from the inside: a complete story with no visible holes. The holes only show up when something pushes on them. Rate yourself again. The second number is the one worth having.

Why Does Explaining Work Better Than Rereading?

Because it makes you generate the explanation, and rereading lets you skip that. Three separate lines of research point the same way.

Explaining exposes gaps you can't feel. That's Rozenblit and Keil, above. Writing out the steps is where their ratings first dropped.

Explaining to yourself builds understanding. In 1989, Michelene Chi and colleagues had students study worked physics examples while thinking aloud. The students who learned best produced more explanations as they went and tracked what they did and didn't understand more accurately. The weaker ones explained less and judged their own learning less accurately. In 1994 Chi's group ran a cleaner test: they asked 14 eighth-graders to explain a passage on the circulatory system to themselves after each line, while 10 others just read it. The prompted group gained more, and the ones who produced the most explanations understood the most. Dunlosky and colleagues' 2013 review of ten study techniques rated self-explanation as "moderate utility."

Explaining from memory is retrieval practice in disguise. A 2018 study by Koh, Lee and Lim split learners into groups. Those who taught the material without notes did about as well on a test a week later as those who simply practised retrieval, and both beat those who taught with notes in hand. Their reading: the benefit of teaching comes largely from having to pull it out of memory. It fits the rule above about keeping the source closed.

If you read my post on active recall vs passive reading, this is the same mechanism wearing a different coat. The Feynman technique is active recall for understanding, rather than for facts.

Where Does the Feynman Technique Fail?

It fails in a handful of predictable places, and the evidence for it is thinner than its fan base suggests.

You're marking your own homework. The gaps you can see are the ones you know to look for. A fluent, confident, wrong explanation passes every internal check. That's why step 3 sends you back to the source, and why Chi's weaker students, who judged their own learning less accurately, are worth remembering. An explanation you've never checked is just a more elaborate feeling of knowing.

The package is under-tested. The parts have decent support, as above. The Feynman technique under that name is another matter: I couldn't find a large randomised trial or meta-analysis of the packaged method. That's a gap in what I found, not proof that it doesn't work. I'd call it a good use of ten minutes, not a proven method.

It isn't top of the evidence tree. In Dunlosky and colleagues' review, practice testing and distributed practice earned "high utility." Self-explanation got "moderate," partly because it hasn't been tested as widely in real classrooms. Use it next to retrieval and spacing, not instead of them.

It's the wrong tool for some material. Rozenblit and Keil found much better calibration for facts and procedures, which means there's less hidden to expose. This table is my judgement, built on that finding:

Kind of material	Does explaining help?	What to reach for
How something works (a mechanism, a cause)	Yes, this is the best fit	The four steps
Why an argument holds (an essay, a chapter)	Yes	The four steps, one claim at a time
Definitions, dates, vocabulary	Little; you know or you don't	Flashcards with spaced repetition
A skill you can do (a proof, a recipe, a chord)	Partly	Doing it, with feedback

A good analogy is a loan, not a gift. Step 4 asks for one, and it will hold up until the point where it breaks. The technique doesn't tell you where that is. Treat the analogy as a handle for the idea, and keep the mechanism underneath it.

It needs something to explain. You can't explain what you haven't met. It's a tool for consolidating after a first pass, not a replacement for one.

How Do You Use It on Things You've Read?

Close the tab, take a blank page, and explain the one claim you'd defend if someone challenged it. Then find out where you can't.

Here's a version that takes about ten minutes and uses the same four steps on an article or a chapter:

Pick the claim, not the topic. Not "this article about studying." The specific argument: "Spaced practice beats cramming, because..." If you can't state the claim in a sentence, that's your first gap.
Write it from memory, with "because" after every step. Scott Young's advice is to go tight and debug a specific confusing point instead of explaining the whole subject. The word "because" is a cheap detector. Every time you can't finish that sentence, underline it.
Reopen the article for the underlines only. Not a full reread. You're looking for the missing step, and finding it is much more memorable than being handed it.
Rewrite, then repeat tomorrow from a blank page. The next-day repeat is where the Feynman technique stops being a one-off and becomes retrieval spread over time. The forgetting curve says a lot of it will be gone by then. What survives is what you actually understood. Our guide to how to remember what you read covers the wider habit.

This is the loop Alexandria's web app is built for: save a source, pull out the ideas, review them later. The explaining step is the one thing no tool can do on your behalf, and it's the step that makes the rest worthwhile.

The test I'd run first is on the last article that felt completely clear to you. Not the confusing one. The clear one. That's where the illusion lives, and the ten minutes will tell you whether you've been carrying something real or just the feeling of it.

Frequently Asked Questions

What is the Feynman technique for studying?

The Feynman technique is a self-test with four steps: pick one concept, explain it in plain words as if teaching a beginner, mark where you get stuck and fix those gaps from the source, then simplify and test again. It exposes the difference between feeling you understand something and being able to explain it.

Did Richard Feynman invent the Feynman technique?

No. Feynman never published a four-step study method. Scott Young popularised the label and says his version is "only loosely based on Richard Feynman's practices." Real Feynman stories sit underneath it, but the steps are a later packaging.

Does the Feynman technique actually work?

Its parts are well supported. Self-explanation improves understanding in several studies, and trying to explain exposes gaps people don't notice otherwise. The packaged method is under-tested: I couldn't find a large randomised trial or meta-analysis of it. It's a sensible use of ten minutes, not a proven method.

What is the illusion of explanatory depth?

It's the finding, from Rozenblit and Keil (2002), that people feel they understand how things work far better than they do. Ratings of understanding dropped once people had to write a step-by-step explanation. The illusion is much stronger for explanations than for facts, procedures or narratives.

When does the Feynman technique not work?

It works poorly for arbitrary facts, dates and vocabulary, where you already know whether you know it. It also fails when your explanation is fluent but wrong and you never check it against the source, and it can't replace practice problems or spaced retrieval, which have stronger evidence.

How do you use the Feynman technique on an article or book?

Close the article, take a blank page and explain its one main claim and why it's true, in plain words. Mark every point where you stall. Reopen the article only for those points, then rewrite the explanation. Repeat from a blank page the next day to check it stuck.

Do you need another person to use the Feynman technique?

No. The popular version uses a blank sheet of paper. What matters is producing the explanation without the source in front of you. A listener can help by asking questions you didn't think of, but the gap-finding works alone, provided you don't peek.