On this page
"You only really understand something when you can teach it" is old advice. The research mostly agrees, with one important detail: the learning comes from specific parts of teaching, and you can use those parts on your own, at your desk, tonight.
What are the parts of teaching?
In a 2023 chapter, Logan Fiorella of the University of Georgia describes three stages of learning by teaching, following earlier work by Bargh and Schul: preparing to teach, explaining to others, and interacting with them through questions. Each one can add something.
Does just expecting to teach help?
Often, yes. In experiments by Nestojko, Bui, Kornell and Bjork, students read passages either expecting a test or expecting to teach; in fact, everyone took the test. As Nestojko put it in the Washington University news release, "learners expecting to teach recalled more material correctly, they organized their recall more effectively and they had better memory for especially important information." The study appeared in Memory & Cognition.
Does actually teaching add more?
In Fiorella and Mayer's 2013 study, students learned about the Doppler effect expecting to teach it or to be tested. Some then recorded a video lecture for a student who did not know the topic. Fiorella's chapter summarises the results: on an immediate test, students who prepared to teach did about equally well whether or not they went on to teach, but on a delayed test, the students who had actually taught did best.
A 2019 meta-analysis by Kobayashi pooled 28 studies. Preparing to teach had an average effect of g = 0.35; preparing and then teaching, g = 0.56. Both helped deep learning as well as surface learning, and effects were larger when the teaching was interactive.
Why does it work?
A big part is retrieval. Koh, Lee and Lim had students teach with or without their notes. A week later, those who taught from memory, like those who simply practised recalling, beat those who taught while reading from notes. Teaching with your notes in front of you loses much of the benefit.
Fiorella's chapter adds two cautions. Some students explain by restating the material without adding anything, which researchers call a "knowledge-telling bias". And some studies found no benefit from expecting to teach; in one, students reported high anxiety at the prospect of teaching.
A 25-minute teach-back routine
Choose one topic, such as electric potential or Le Chatelier's principle.
Prepare (10 minutes). Study it as if you must teach a younger student next week. Pick the two or three main ideas and one example.
Explain from memory (5 minutes). Book closed. Record yourself on your phone, or explain to a sibling or friend. Talk to a listener: "you'll notice that..." Draw as you speak; in a study described in Fiorella's chapter, students who explained while drawing did better a week later than those who only explained.
Check (5 minutes). Open the book. List what you left out or got wrong.
Take questions (5 minutes). Have a friend ask two questions, or write two "why" and "what if" questions yourself and answer them.
Do this with a study partner and swap roles. Our guide to study groups that work shows how to run it in a group.
Making it fit coaching
Use it for the one topic per day you understood least in class.
Keep the recordings. Before a test, listen to one and pause to predict what comes next.
If your explanation keeps stalling at the same step, that step is the gap. Go back to the source, then teach it again.
When you can't find the gap yourself, Fermi's AI tutor holds a line of reasoning across a session, decides whether to explain, ask or show, and checks understanding as it goes. For a related method, see the Feynman technique.
Pick tomorrow's hardest topic and teach it back, from memory, before you sleep.
