Does the Feynman Technique Work? The Evidence, Step by Step

Does the Feynman technique work? Partly. No study has tested its four steps as one package, but research on each step points one way: expecting to teach improves how you study, and explaining exposes gaps. Explaining aloud has more support than writing, and the method finds more gaps in how things work than in facts.

Where the Feynman technique comes from

The four steps don't come from Richard Feynman's own writing, as far as anyone has traced them. The earliest version we could find is Scott Young's page Learn Faster with The Feynman Technique, first saved by the Internet Archive in September 2011. Its steps:

1. Write the concept at the top of a blank page.
2. Write an explanation as if teaching someone who has never heard of it.
3. Wherever you get stuck, go back to the book or the teacher.
4. Simplify wordy parts, or replace them with an analogy.

Young wrote that he used it to learn MIT's physics class in under five days. The page cites nothing by Feynman.

The spirit is Feynman's, though. His Caltech colleague David Goodstein once asked him why spin-1/2 particles obey Fermi-Dirac statistics. Feynman promised a freshman lecture on it, then came back: "I couldn't reduce it to the freshman level. That means we really don't understand it" (Caltech Magazine). In his 1974 commencement address, he told graduates that "you must not fool yourself", "and you are the easiest person to fool" (Cargo Cult Science).

How well does the Feynman technique work at each step?

Each step has its own research, mostly lab experiments with university students:

1. Studying to teach changes how you study

In two experiments by John Nestojko, Dung Bui, Nate Kornell and Elizabeth Bjork, people read passages expecting either a test or to teach the passage to another student. In fact, everyone took the test and nobody taught. Those expecting to teach recalled more, in better order, and answered more questions, especially on the main points.

2. Say it, don't just write it

Vincent Hoogerheide's team ran two experiments with university students. In the first, with 123 students, writing an explanation for imagined fellow students did no better than rereading. In the second, 129 students restudied a text, explained it in writing, or explained it to a webcam:

Only the video group reliably beat restudying, though the gap between video and writing wasn't significant either. The authors' guess is social presence: a camera feels like an audience.

Two later experiments by Andreas Lachner, Leonie Jacob and Hoogerheide, with 147 and 51 students, found no gain from writing for an imagined student. Explaining to yourself did better. Their meta-analysis of the studies so far found a small benefit from explaining to others, larger when spoken than written.

3. The gaps you find are real

Leonid Rozenblit and Frank Keil asked Yale students how well they understood devices such as a zipper, a flush toilet and a cylinder lock. The students rated themselves from 1 to 7, wrote step-by-step explanations and rated themselves again. In one study, with 33 undergraduates, the average rating fell from 3.89 to 3.10 (Cognitive Science). The authors called it the illusion of explanatory depth.

The same gap shows up with objects you use every day. Rebecca Lawson tested how well people understood something as familiar as a bicycle. They made frequent and serious mistakes, such as believing the chain goes around the front wheel as well as the back one. Bicycle experts made fewer errors, and so did people shown a real bicycle during the test, but neither group got rid of them (Memory & Cognition, 2006). Her conclusion: most people's understanding of this everyday object is "sketchy and shallow". If you can't sketch how something works, you've found a gap worth going back for.

4. An analogy works best as a comparison

Louis Alfieri, Timothy Nokes-Malach and Christian Schunn pooled 57 experiments in a meta-analysis. Comparing cases side by side taught more than studying one case at a time (d = 0.50). It helped most when learners looked for what the cases shared and got the principle after comparing. So don't stop at "it's like a pump": list where your concept matches the pump and where it doesn't.

Where it works less well

The same paper is the clearest guide to where the answer to "does the Feynman technique work" depends on what you study. Rozenblit and Keil repeated the rate, explain, rate again test with other kinds of knowledge:

People overrate their grasp of how things work far more than their grasp of facts or step lists (Cognitive Science). Explaining finds the most gaps in how things work and why things happen. For vocabulary, dates or capitals, a gap shows the moment you can't recall the answer, so flashcards do that job.

The Feynman technique for math needs the same split. Explain why a method works, but learn the method by solving problems. Vincent Aleven and Kenneth Koedinger ran two classroom experiments with a computer tutor for problem solving. Students who explained each step while practicing understood more, and did better on new kinds of problems, than students who only solved. The explaining happened during the practice, not instead of it.

One caution comes from a 2021 review by Lachner, Hoogerheide, Tamara van Gog and Alexander Renkl. Teaching an audience you can't interact with can help, they note, but several recent studies did not replicate the effect.

How to do the Feynman technique by yourself

1. Pick one concept, not a chapter. "How a vaccine trains the immune system" fits on a page; "immunology" doesn't.
2. Study it as if you'll teach it tomorrow. Look for the main points and the order they go in.
3. Close the book and explain it out loud. Talk to your phone's voice recorder or a camera, as if to a classmate who missed the lecture.
4. Mark every stall. Listen back and note each pause, each term you used without defining it, each "basically".
5. Answer one specific question. Pick something your explanation should cover, such as "what changes if the dose is halved?", and answer it without notes.
6. Go back only for the marked spots, then explain it again, shorter.
7. Add one comparison. Say what your concept shares with something familiar and where the likeness breaks.
8. Write the final version for yourself, in a few lines, as a review card.

Alone, with a partner or in a group

A camera gives you part of the effect; a real listener can also ask "why?". Does the Feynman technique work in a group? It can, if one person explains, the others ask questions, and the role rotates. With three people, give each one a concept and ten minutes: five to explain with books closed, five to answer the others' questions before anyone checks the notes. For a pair, the teach-from-memory method splits the material in half and has each partner teach their half with the book closed.

On StudyDuo, a 1:1 session puts you on video with one other person, where you can talk with the mic on. Post a time on the board for now or later in the week, and whoever joins can be your listener. To find someone in your subject, see the study partner page or our guide to finding a study partner.