The Feynman Technique: 4 Steps, the Evidence, and a Worked Example

What the Feynman technique is

The Feynman technique is a way to learn something by explaining it simply, as if to a beginner, without looking at your notes. It is named after the physicist Richard Feynman, who was known for explaining hard ideas in plain language. The four-step recipe you see online is a modern summary attributed to him, not a study protocol he published.

The four steps are:

1. Pick one concept and write its name at the top of a blank page.
2. Explain it from memory in plain words, as if teaching someone new to the field. No jargon you could not define.
3. Find the gaps. Wherever you got stuck, waved your hands, or fell back on a term you cannot unpack, mark it. Go back to the source and fix exactly that part.
4. Simplify and test again. Rewrite the explanation shorter and clearer, add an example or analogy, then try it again later from memory.

It feels slow. That is part of why it works, and part of why most people skip it.

Is the Feynman technique backed by research?

Honest answer: not as a named method. We found no controlled study that tests "the Feynman technique" as such. What has been tested, many times, are the ingredients it combines. That is a weaker claim than "proven", and you should know the difference.

Ingredient 1: retrieval practice

Step 2 asks you to explain without notes, which is a form of retrieval practice: pulling information out of memory instead of putting it in again. A review by Roediger & Butler (2011) concluded that retrieval practice often produces large gains in long-term retention compared with repeated studying, and that feedback makes it work better. Step 3, checking your explanation against the source, is that feedback.

Ingredient 2: self-explanation

Explaining material to yourself in your own words is called self-explanation. A meta-analysis by Bisra et al. (2018) pooled 69 effect sizes from 64 reports and found that prompting learners to self-explain had an average effect of g = .55.

What does g = .55 mean? An effect size like g or d tells you how far apart two groups are, measured in standard deviations (a standard deviation is roughly the typical spread of scores). An effect of about 0.5 means the average learner who self-explained scored about half a standard deviation higher than the average learner who did not. If scores are roughly bell-shaped, that moves someone from the middle of the class to around the 70th percentile. It is a solid, useful effect, not a miracle.

The classic review by Dunlosky et al. (2013) rated self-explanation as moderate utility: promising, but with less evidence from real courses than practice testing or spacing.

Ingredient 3: preparing to teach

The "explain it to a beginner" framing has some support of its own. In Nestojko et al. (2014), people who studied a passage expecting to teach it produced more complete and better organised recall, and answered more questions about main points, than people expecting a test. Nobody actually taught; the expectation alone changed how they studied.

Ingredient 4: exposing the illusion of understanding

The best argument for step 3 comes from Rozenblit & Keil (2002). They showed that people feel they understand how things work with far more depth than they really do, an "illusion of explanatory depth", and that the illusion is stronger for explanations than for facts or procedures. Trying to write the explanation out is what breaks the illusion.

When it works, and when it does not

Self-explanation has limits, and they apply to the Feynman technique too. Rittle-Johnson & Loehr (2017) describe four: it suits principle-based material (why something works) better than arbitrary facts; explaining your own choices can sometimes reduce learning; the effort can pull attention away from other content; and in some cases other activities work better. A related meta-analysis in mathematics, Rittle-Johnson, Loehr & Durkin (2017), found small-to-moderate immediate benefits, with more limited evidence for classroom settings and delayed tests.

In practice:

• Use it for concepts and mechanisms: how TLS works, why subnetting divides a network, how critical path analysis finds slack, why least privilege reduces risk.
• Do not use it for raw lists: port numbers, acronyms, the order of the OSI layers. Those need flashcards and spacing, see our guide on how to memorize things fast.
• Check against a source. A fluent explanation can still be wrong. Without step 3, you may be rehearsing a misconception.

Worked example: the CIA triad in four steps

Here is the Feynman technique applied to a concept that appears on Security+, CISSP and most entry-level security exams. The table below summarises each step; this section shows the actual writing.

Step 1: Pick the concept

"The CIA triad: confidentiality, integrity, availability."

Step 2: First explanation, from memory

"The CIA triad is the three goals of security. Confidentiality means data is encrypted. Integrity means data is correct. Availability means the system is up. You use controls to protect each one."

Step 3: Find the gaps

Reading it back, three weak spots show up:

• "Confidentiality means data is encrypted" confuses the goal with one control. Access control, classification and physical security also protect confidentiality.
• "Integrity means data is correct" is vague. Correct according to whom? The source says integrity means data is not changed without authorisation, and that changes can be detected (hashes, digital signatures).
• No examples of what attacks each goal, and no mention that the goals can conflict.

Step 4: Simplified explanation, second attempt

"Security tries to keep three promises about information. Confidentiality: only the right people can read it, so you use access control and encryption, and a data leak breaks it. Integrity: nobody can change it without permission and get away unnoticed, so you use hashes, signatures and change control, and a tampered invoice breaks it. Availability: the right people can reach it when they need it, so you use redundancy and backups, and a DDoS attack or a dead server breaks it. The promises can pull against each other: locking data down hard can make it less available."

Now test it the way the exam will. Take a scenario question such as "a user claims she never sent an approval email": which promise, and which control? If your explanation does not help you answer, go back to step 3. (That example is non-repudiation, which builds on integrity, and it is a real question on our free Security+ practice test.)

Common mistakes

• Explaining with the book open. That is copying, not retrieval. Close everything first.
• Doing it only in your head. Write or say it out loud. Vague thoughts feel complete; written sentences show the holes.
• Stopping after one pass. The second explanation should come a day or more later, which adds spacing. See spaced repetition.
• Picking topics that are too big. "Networking" is too big. "Why a /26 gives 62 usable hosts" is right.
• Treating it as your only method. It is slow. Use it for the concepts you keep getting wrong, and use practice questions for breadth.

How to use this for a certification exam

A practical protocol that puts the Feynman technique where it pays off most: on the concepts your practice tests show you do not really understand.

1. Take a baseline test. Do one of our free practice tests cold, for example the Security+ practice test or the CCNA practice test. Answer every question before you read any explanation.
2. Sort your misses. For each wrong or guessed answer, ask: did I forget a fact, or did I not understand the idea? Facts go to flashcards. Ideas go to a Feynman page.
3. Write one Feynman page per idea. Steps 1 and 2 from memory, about 10 minutes each. Then step 3 against the official exam objectives and your study guide.
4. Re-explain after a gap. Two or three days later, cover the page and do step 2 again. Compare. Anything you still miss goes back into step 3.
5. Retest. A week later, retake the practice test. If you get the question right and can say why every wrong option is wrong, the concept is done.

For the bigger picture of how this fits with spacing and practice testing, read our guide to studying effectively and the closely related elaborative interrogation, which asks "why is this true?" about individual facts.