The Complete Guide to the Feynman Technique (and How to Use It on Edirae)
The Learning Method Physicists Use to Actually Understand Hard Things
THE PROBLEM WITH HOW YOU LEARN NOW
You watched a 45-minute YouTube video on a hard topic. You understood every sentence while watching. You closed the tab. Three days later, you tried to explain it to a friend, and nothing came out.
That's not a memory problem. That's a fake understanding problem.
The gap between "I understood while I was reading" and "I can produce this from scratch" is the gap where most adult learning dies. And the tool that closes that gap has existed for 60 years. It just isn't taught in schools.
It's called the Feynman technique. And after using it consistently for the past few months on quantum computing, Mandarin, and inference engineering, I can tell you it's the single highest-leverage change I've made to how I learn.
Here's the complete guide.
WHO FEYNMAN WAS AND WHY THIS IS NAMED AFTER HIM
Richard Feynman won the Nobel Prize in physics in 1965. He was famous for two things beyond his research: teaching the hardest concepts in physics in the simplest possible language, and refusing to accept anything as understood until he could explain it plainly.
His method was disarmingly simple. When he wanted to learn something, he'd take a blank notebook, write the topic at the top, and start explaining it as if to a child. When he hit a spot where he had to use jargon or copy from a textbook, he'd stop, go back to the source, and figure out what he was missing.
He wasn't inventing a study system. He was documenting how his own mind actually built understanding. Decades later, cognitive scientists formalized it, gave it his name, and confirmed with data what he already knew intuitively.
If you can't explain something in simple terms, you don't understand it. That's the whole insight.
THE 4-STEP METHOD (EXACT)
There are dozens of blog posts about the Feynman technique. Most of them add fluff. Here's the actual canonical version in 4 steps.
Step 1: Pick a concept
Not a topic. A concept.
"Quantum mechanics" is a topic. "Superposition" is a concept.
"Machine learning" is a topic. "Gradient descent" is a concept.
"French" is a topic. "The subjunctive mood" is a concept.
The Feynman technique works on discrete testable ideas. Try it on a whole topic and you'll drown.
Step 2: Explain it as if teaching a 12-year-old
Open a blank page or a voice recording app. Explain the concept in the simplest possible language. No jargon. No copying from the source. From your own head, in your own words.
Two rules:
- Use everyday words, not textbook words
- If you have to use a technical term, define it in plain language first
Set a timer for 5-10 minutes. Explain until you're either done or stuck.
Step 3: Identify the gaps
When you hit a spot where you can't explain, or where you catch yourself defaulting to jargon, or where your explanation feels shaky, mark it.
Those marks are the map of what you don't actually understand.
This is uncomfortable. You'll usually find you don't understand 30-60% of what you thought you understood. That discomfort is the entire point of the technique.
Step 4: Return to source and simplify
Go back to the source material (paper, textbook, video, teacher). Read only the sections that map to your gaps. Come back to your explanation. Rewrite the weak parts in plain language.
Repeat until your explanation flows without stopping.
That's it. Four steps. The whole method.
WHY IT WORKS (THE ACTUAL SCIENCE)
The Feynman technique isn't a productivity hack. It exploits five well-validated cognitive mechanisms simultaneously.
1. The generation effect (Slamecka & Graf, 1978). Information you generate yourself is retained 30-50% better than information you consume. When you write your own explanation, your brain does the work of encoding it, which is what makes it stick.
2. The testing effect (Roediger & Karpicke, 2006). Retrieving information from memory is one of the strongest predictors of long-term retention. Explaining without notes is pure retrieval.
3. Illusion of competence disruption (Dunning-Kruger literature). Reading a passage makes you feel like you understand it. Explaining that passage back forces you to face what you don't. The Feynman step is the fastest known method for busting the illusion.
4. Elaborative rehearsal. Connecting new information to existing knowledge (via analogies and simpler language) creates more neural pathways to the concept. More pathways means faster future retrieval.
5. Cognitive load reduction (Sweller, 1988). Complex ideas overload working memory. Translating them into simpler language forces you to chunk them into forms your brain can actually hold and manipulate.
Combined, these mechanisms produce something like a 2-4x retention improvement compared to passive study, across dozens of studies. There is no other single technique with a comparable evidence base.
THE 5 COMMON MISTAKES
Most people who try the Feynman technique quit within a week. Almost always for one of these reasons.
Mistake 1: Explaining in your head instead of out loud or on paper. Silent thinking feels productive but doesn't force full retrieval. Your brain lets you skip the parts you don't know. Speaking or writing forces you to commit.
Mistake 2: Picking too big a topic. "Explain machine learning" is not a Feynman prompt. "Explain gradient descent" is. Aim for concepts you could reasonably explain in 5-10 minutes.
Mistake 3: Peeking at the source mid-explanation. The whole point is to see where you fail. If you peek every time you get stuck, you never see the map of your gaps.
Mistake 4: Being kind to yourself about weak explanations. "That was pretty good" is the enemy. If a random 12-year-old would ask "wait, what?" at any point in your explanation, you're not done. Be your own harshest critic.
Mistake 5: Doing it once and stopping. The Feynman technique is not a one-shot activity. Concepts you Feynman-check today need to be re-Feynman-checked in 7 days, 30 days, 90 days. Otherwise the understanding decays like any other memory.
CONCRETE EXAMPLES (SEE IT IN ACTION)
Reading about the Feynman technique is not the Feynman technique. Here are three real examples across different subjects.
Example 1: Compound interest (finance)
First attempt (weak): "Compound interest is when interest gets added on top of interest so your money grows faster. It's exponential."
Gap check:
- What does "exponential" mean here?
- Why does interest getting added on top matter?
- What's the actual mechanism?
Rewrite: "Compound interest means the interest you earn each year gets added to your original money, and next year you earn interest on the bigger total. It's like a snowball rolling down a hill. It picks up more snow, so the next section it rolls through picks up more snow than the last, and it grows faster and faster. If you had $100 at 10% interest, year one you'd have $110. Year two you'd earn 10% on $110, not $100, so you'd have $121. Year ten, you'd have $259, not $200. The extra $59 is compound interest doing its work."
That second version is what actual understanding sounds like.
Example 2: Superposition (quantum mechanics)
First attempt (weak): "Superposition is when a quantum particle is in multiple states at once until you measure it."
Gap check:
- What does "in multiple states at once" actually mean?
- Why does measurement matter?
- Is this different from just not knowing which state something is in?
Rewrite: "Imagine flipping a coin, but instead of landing on heads OR tails, while it's still spinning in the air, the coin is genuinely both heads and tails at the same time. Not that you don't know which one. It's actually both. Only when it lands, meaning when you 'measure' it, does it become one or the other. Quantum particles do this with things like spin or position. They exist in a blur of possibilities until someone looks. Once looked at, the blur collapses into one specific answer. That collapse is what makes quantum computing different from regular computing, because you can use the blur to do many calculations at once."
Notice the coin analogy. That's not decoration. That's the elaborative rehearsal step where your brain links the new concept to existing physical intuition.
Example 3: A subjunctive verb (French)
First attempt (weak): "You use the subjunctive when there's doubt or emotion."
Gap check:
- Why doubt or emotion?
- When exactly do I use it?
- What's an example that shows me I understand?
Rewrite: "The subjunctive is a special form of the verb that French uses when what you're saying isn't a straightforward fact. If you say 'I know he's coming,' that's a fact, so you use the normal verb form. But if you say 'I want him to come,' or 'I doubt he's coming,' or 'It's necessary that he come,' the coming isn't a fact yet. It's a wish, a doubt, or a requirement. French marks that hypothetical quality by changing the verb ending. 'Il vient' (he comes, indicative) becomes 'qu'il vienne' (that he come, subjunctive). English kind of does this too with 'that he come' vs 'that he comes,' but we've mostly dropped it. French kept it."
The rewrite reveals what you actually understand. The first attempt would have failed on an exam question. The rewrite would have passed.
HOW TO USE THE FEYNMAN TECHNIQUE ON EDIRAE
We built Edirae to run the Feynman loop for you, at scale, with better feedback than a friend can give.
Here's how it works in practice.
The Explain-It-Back feature
After any learning session on Edirae (studying material you fed in, completing a course chapter, reviewing retention cards), the app can prompt you to Explain-It-Back.
You get:
- A specific concept from what you just learned
- A blank text or voice input
- 3-5 minutes to explain
You record your answer, out loud or typed. Edirae's AI grades it on four dimensions:
- Coverage: did you hit the key parts of the concept?
- Simplification: are you using plain language, or are you leaning on jargon?
- Analogy quality: did you generate an original analogy, or borrow the source's?
- Accuracy: did you say anything factually wrong?
Each dimension gets a 1-10 score with specific feedback. Not "good job." Specific pointers like: "Your definition of X was accurate, but you used the term Y without explaining it. A 12-year-old would be confused there."
The retention layer connection
Here's where Edirae differs from doing the Feynman technique manually.
When your explanation reveals a gap (you scored 4/10 on "coverage" for a specific sub-concept), that gap becomes a new priority card in your retention deck. Edirae schedules it for retrieval within 24-48 hours. Then within 7 days. Then 30. Then 90.
The gaps you couldn't explain today become the exact material your brain is forced to retrieve at the exact moment it's about to forget.
That's the compound. Feynman finds the gap. Retention closes it. Together, they produce durable understanding in a fraction of the time either does alone.
The typical Edirae Feynman workflow
Here's what a 20-minute Edirae session using this loop looks like:
- Minutes 0-10: Review retention cards from prior study (spaced repetition)
- Minutes 10-15: Learn something new (read a section, watch a video, work through a problem)
- Minutes 15-20: Explain-It-Back on the new concept you just learned
That's it. Daily. Compound.
Users who use this workflow consistently show 90%+ retention at 30 days on their explained concepts, vs the 20% baseline for pure passive study.
ADVANCED VARIATIONS (WHEN YOU'RE READY)
Once you've done the basic Feynman loop 20-30 times and it feels natural, layer these variations on for another dimension of learning.
Variation 1: The Feynman-to-child test
Explain the concept to an actual 12-year-old (or a real non-expert). Their questions will reveal gaps your own self-critique missed. Kids are exceptional at spotting jargon.
If you don't have a 12-year-old handy, Edirae's AI is trained to roleplay one when you request it.
Variation 2: The Feynman-in-writing test
Instead of speaking, write a 500-word explanation as if for a blog post. The constraint of writing publishable prose reveals different weaknesses than speaking does. Speaking hides logical gaps; writing exposes them.
Variation 3: The Feynman-with-analogy test
Explicitly require yourself to generate a NEW analogy for the concept. Not the standard textbook analogy. A fresh one, from a domain unrelated to the source material. This forces the deepest possible elaborative rehearsal.
Compare: "Neurons in the brain fire like electrical circuits." (Standard, weak.) vs. "A neuron firing is like a person deciding to speak in a crowded room. It waits until enough other people around it start murmuring the same thing, then it commits." (Novel, deep.)
Variation 4: The Feynman-under-time-pressure test
Give yourself 60 seconds instead of 5 minutes. Compression forces you to identify the absolute core of what you understand, and reveals anything peripheral you were relying on to fill space.
Variation 5: The reverse Feynman
Instead of explaining a concept, explain WHY a common misunderstanding of the concept is wrong. This tests understanding from a completely different angle and often reveals gaps the standard method misses.
FREQUENTLY ASKED QUESTIONS
What is the Feynman technique?
The Feynman technique is a learning method where you explain a concept in the simplest possible language, as if to a 12-year-old. Whenever you hit a spot where you struggle or default to jargon, you've found a gap in your understanding. You return to the source material to close the gap, then re-explain until your version flows in plain language.
Who invented the Feynman technique?
The technique is named after physicist Richard Feynman (Nobel Prize 1965), who was famous for teaching complex physics in the simplest possible language. He didn't formally invent it as a technique; he documented and demonstrated it in his own work. Later cognitive scientists formalized the steps and confirmed the underlying mechanisms with data.
How long does the Feynman technique take?
A single Feynman check on one concept takes 10-20 minutes: 5-10 minutes to explain, 5 minutes to identify gaps, 5 minutes to rewrite. Doing it daily for one concept per session is enough to see dramatic improvement in a few weeks.
Does the Feynman technique work for math and coding?
Yes, especially. Math and coding are the domains where "I thought I understood this" fails most spectacularly on exams or in production. Explaining a math proof or code snippet in plain language forces you to face what you actually understand vs what you'd just memorized.
Is the Feynman technique the same as active recall?
They overlap but aren't identical. Active recall is retrieving information from memory. The Feynman technique is a specific type of active recall focused on explanation and simplification. All Feynman is active recall; not all active recall is Feynman.
How is the Feynman technique different from just teaching?
Teaching is the outward-facing application of the Feynman technique. If you're teaching someone else, you're already running some version of the loop. The Feynman technique is what makes teaching effective as a learning tool: you're not just performing, you're forcing yourself to face what you don't understand.
Can AI grade my Feynman explanations?
Yes. Modern LLMs (including the AI inside Edirae) can grade explanations on coverage, simplification, jargon use, and analogy quality. AI feedback is less nuanced than a human expert but faster, cheaper, and available whenever you want to practice.
Should I do the Feynman technique on every concept I learn?
For most casual learning, no. Reserve it for concepts that are: important to remember long-term, difficult to grasp, or foundational to something else you're learning. For those concepts, running the Feynman loop once (plus scheduled retrieval afterward) is the highest ROI use of your study time.
THE ONE THING TO REMEMBER
If you can't explain it in simple language, you don't understand it. That's the whole technique in one sentence. Everything else in this guide is scaffolding around that truth.
The reason most adults never actually master hard concepts is that they never stress-test their understanding this way. They read, they nod, they move on. Then, when they need the concept, it isn't there.
The Feynman technique is the stress test. Do it daily on one concept. Feed the gaps into a retention system. In 90 days you'll operate at a level of understanding that most of your peers will never reach.
Written by Patrick Obafemi, founder of Edirae. Edirae runs the Feynman loop for you with AI grading and spaced-repetition follow-up on every gap. Free 5-day Pro trial at edirae.com.
Related reading:
Related Articles
I Watched Four People Try the 60-Day GMAT Plan. Only One Worked. Here's What She Did Differently.
Four people in my extended circle sat the GMAT inside an eighteen month window. All four gave themselves roughly sixty days. All four bought the same official guide and downloaded a similar looking spreadsheet off Reddit.
How to Introduce Yourself in an Interview (The 90-Second Formula That Actually Works)
The first 90 seconds of a job interview decide more than most candidates realize. Research from the University of Toledo showed that interviewers form a "hire or don't hire" judgment within the first 3-5 minutes of an interview. Everything after that is confirmation bias. Your qualifications, your project stories, your questions at the end, they all get filtered through the impression you made at the start.
How to Train Your Brain to Do Hard Things
The Neuroscience, The Protocol, and Why Most People Give Up on Themselves in Week 3