The Feynman technique + flashcards: how to combine them for faster understanding

The Feynman technique is a four-step method for testing how well you actually understand something. You pick a concept, explain it in plain language as if teaching a beginner, notice where your explanation falls apart, and go back to the source to fix the holes. On its own, it builds understanding. Paired with flashcards, it turns that understanding into memory that survives the semester.

Most students treat Feynman as a one-off exercise and stop there. The cost shows up two weeks later, when the same gaps quietly return. Every honest Feynman session produces a small list of things you almost understood, and "almost" is the exact kind of material that needs spaced retrieval if you want it to still be there next month.

What is the Feynman technique?

Richard Feynman won the Nobel Prize in physics in 1965 and spent most of his teaching career arguing that the real test of understanding is whether you can explain something simply. The method now attached to his name was popularized later, partly through James Gleick's biography and partly through study-YouTube.

The version that gets passed around online has four steps:

Pick a concept you want to understand.
Explain it in plain language, as if your audience knows nothing about the topic.
Notice every point where you hand-wave, paraphrase incorrectly, or stop.
Go back to the source, fix those points, and try again.

You loop until the explanation runs from start to finish on its own. The catch is step three: most learners do not notice their own hand-waves until they hear themselves out loud, which is also why the technique works in the first place.

Why does the Feynman technique work?

Two mechanisms are doing the work, and they reinforce each other.

The first is retrieval practice. Pulling a concept out of memory is more effective than rereading it. Karpicke & Roediger (2008) reported roughly double the long-term retention from retrieval-based study compared with restudying. The Feynman technique is retrieval practice with an extra constraint: the answer has to be intelligible to someone who knows nothing.

The second is what cognitive scientists call the illusion of explanatory depth. Rozenblit & Keil (2002) showed that people consistently overestimate how well they understand ordinary objects — bicycles, zippers, toilets — until they are asked to walk through how the thing actually works. The act of explaining puts a number on what you know. You discover the gap not because someone tells you it is there, but because your own voice trails off mid-sentence.

Most of the value comes from that uncomfortable moment, which is also why people skip it. A halting, slightly embarrassing monologue about, say, how a transformer attends to tokens is far more useful than rereading the original paper for the fifth time. The monologue tells you exactly which sentence of the paper to reread.

What are the four steps in practice?

The textbook version is everywhere, so here is the working version with the parts that tend to get glossed over.

1. Pick a single concept

Not a whole chapter. One idea. "What is a derivative." "Why does insulin resistance develop." "How does the citric acid cycle produce ATP." The narrower the target, the easier it is to spot where you stop knowing things.

2. Explain it to a hypothetical beginner

Out loud is better than thinking it through silently. Writing it down is usually better than out loud, because you can read it back. Either way, the rule is no jargon unless you can also define the jargon in the same explanation. If you find yourself saying "and then the protein gets phosphorylated," stop and ask what phosphorylation actually is at the level of atoms moving around. If you can't answer, that's a gap.

3. Mark the gaps

Every time you stall, paraphrase incorrectly, or wave your hands, write it down. Be specific. "I don't really know how an enzyme lowers activation energy" is useful. "Bit shaky on enzymes" is not. The audit is the hardest step in the whole method, because it requires you to admit, on paper, what you do not know. Without it, you have done a slow rereading session dressed up as something else.

4. Go back to the source and patch

Not Wikipedia. Your actual textbook, lecture notes, paper, or whatever authoritative source the topic deserves. Address each gap, then re-explain from scratch with the patch in place. If a new gap surfaces, repeat. The technique fails for people who stay vague at this step or who patch the gap without re-explaining.

How does the Feynman technique pair with flashcards?

The technique finds gaps. Flashcards keep them closed.

The failure mode I see all the time goes like this. A student does a Feynman session, identifies five things they did not understand, looks them up, feels good about it, and moves on. Two weeks later the same five gaps are back, because understanding decays without retrieval practice. That is what the forgetting curve describes, and it is what spaced repetition is built to counter.

A flashcard, in this context, is a small commitment to your future self: I will check whether I still understand this in a few days, then again next week, then again next month. Each Feynman gap becomes a card. Each card becomes a scheduled review. The spacing algorithm handles the timing so you spend your effort on the material that is actually fading.

Neither method handles both jobs well on its own. Flashcards by themselves are good at memorization and bad at depth — the danger is rote cards you can answer without understanding. Feynman by itself is good at depth and bad at retention — without review, even the explanations you fought hard for slip out of reach within a couple of weeks. Stack them and you cover both.

A concrete workflow: from Feynman session to flashcard deck

This is the part almost nobody writes down. Sized to a 30 to 45 minute study block.

0 to 10 minutes. Pick a target and explain it cold. Open a blank document or a voice recorder. Say or write the explanation without consulting any source. Force yourself to keep going even when you know you are wrong. The wrong parts are exactly what you need.

10 to 20 minutes. Audit. Read back what you wrote (or transcribe roughly what you said). Mark every gap, every hand-wave, every undefined term, every sentence you could not defend if questioned. Use a highlighter, or a simple [gap] tag in brackets. Be picky. A common mistake is to mark only the obvious unknowns and miss the places where you used a word correctly without really knowing what it meant.

20 to 35 minutes. Patch from the source. Open your textbook, paper, or lecture notes. Work through each gap. Rewrite the relevant part of the explanation in place so it actually runs from start to finish.

35 to 45 minutes. Convert gaps into cards. This is the bridge step. For each gap you patched, write one or two cards. Keep them small — one fact or one mechanism per card. Some examples for biology:

"Why does an enzyme lower the activation energy of a reaction?"
"What happens to ATP yield if the electron transport chain is blocked, and why?"
"In one sentence, what does it mean for a protein to be phosphorylated?"

For math or physics, prefer cards that ask for a derivation step rather than a memorized end-formula. For languages, the card might be a worked-out grammar pattern from a sentence you had to break down during the session. The principle stays the same: the place where your explanation stalled becomes a question that forces the same retrieval next week.

Save the cards to a deck and stop thinking about them. The scheduler will surface them at the right time.

One more thing on volume. A good Feynman session usually generates somewhere between three and ten cards. If you are producing fifty, you picked too broad a concept. Narrow it and start again.

What kinds of cards work best for Feynman gaps?

Not every gap makes a good card. The ones that earn their place:

Mechanism cards. "How does X cause Y?" These are the highest-leverage cards, because they encode the causal step you missed in the first place.

Definition-in-context cards. Not "define mitochondria," but something like "what does the 'powerhouse of the cell' metaphor get right, and where does it mislead?" The second version forces retrieval of actual understanding rather than recognition of a phrase.

Worked-example cards. A short problem on the front, with the first step (or the full solution) on the back. Useful for math, physics, and anything procedural.

The cards to avoid are the ones you can answer by pattern matching alone. "What year was the Treaty of Westphalia signed?" is a perfectly fine card for a history exam, but if your Feynman gap was about why European sovereignty changed after 1648, that card will not help you re-explain it next week. Write a card that asks the harder question instead.

What are the common mistakes with this method?

A few patterns come up again and again.

Picking a concept that is too big. "Explain the immune system" gives you a vague paragraph. "Explain how a B cell selects which antibody to produce" gives you something you can actually audit.

Skipping the audit. The audit is uncomfortable because it forces you to write down what you don't know. If you skip it, you've slow-reread the chapter and called it something else.

Treating the cards as the goal. The cards are scaffolding. The goal is the understanding that the cards keep refreshing. If you find yourself optimizing for card count, you've inverted the method.

Not coming back. A Feynman session from three months ago, with cards you never review, is barely better than no session at all. The whole reason for the flashcard layer is that the scheduling keeps you honest when motivation drops, which it always does, usually mid-semester.

How does Memor More fit into this workflow?

Memor More is a spaced-repetition app, which means it handles the "review this card at the right interval" part of the loop. The app does not run the Feynman session for you. That part is on you, with a notebook or a voice memo. What the app handles is everything after: holding the cards you generated, scheduling them, and surfacing them on the days when retrieval is actually useful.

The practical pattern is straightforward. Keep one "Feynman" deck per subject. After each session, add the new cards to that deck. Review whenever the app surfaces it. Over a semester, the deck becomes a record of every concept you have actually had to fight for — which, not coincidentally, is the material most likely to show up on a hard exam.

I'd push back, while we're here, on the idea that flashcards and "real understanding" are in tension. They are in tension when the cards are bad. A card built on the back of a Feynman gap is not rote — it's the question that exposed the gap in the first place, asked again before you have time to forget the answer.