The Feynman technique for MRCPCH revision: does it work?

The Feynman technique can support MRCPCH revision when you close the source, explain one bounded idea in plain language for an imagined beginner, expose what you cannot justify, correct the explanation and test it on a fresh question. Its value comes from retrieval, organisation, audience-aware explanation and feedback - not from sounding simple, recording a polished video or adopting a famous label.

This article covers non-clinical MRCPCH examination preparation. It contains no clinical teaching, diagnostic, treatment, prescribing, dosing, safeguarding or patient-management guidance. Research on preparing to teach and learning by teaching spans varied learners, subjects, formats and outcomes; it does not establish that the Feynman technique improves MRCPCH scores or pass rates.

The direct answer: teach from memory, then audit the lesson

The popular version of the Feynman technique is usually described as choosing a topic, explaining it simply, finding gaps and refining the explanation. That sequence is sensible, but "the Feynman technique" is not one standardised intervention tested directly in MRCPCH candidates.

The closer research tradition is learning by teaching: studying with the expectation of teaching, then producing an instructional explanation. Evidence suggests that preparing to teach and actually teaching can improve the teacher's own learning under some conditions. The benefit is not automatic. Reading a script, copying a source into slides or rehearsing an unchecked explanation can preserve the appearance of teaching while removing the learning work.

A useful teach-to-learn attempt begins without the answer in view and ends with the answer checked. Simplicity is valuable only when it preserves the important relationship rather than hiding uncertainty or deleting necessary conditions.

Use the TEACH framework: Target one question, Explain without the source, Audit every claim, Challenge the explanation and Handoff a precise next return.

Key points

• Meta-analyses generally report positive average effects for preparing to teach and teaching, but effects vary by expectation, interaction, mode, comparison task and outcome. There is no evidence that the Feynman technique improves MRCPCH pass rates.
• Explaining from memory matters. Research comparing teaching without notes, retrieval practice and script reading suggests that retrieval may account for an important part of the benefit.
• A fluent explanation can still be wrong or incomplete. Verify it against a current authoritative source, mark omissions and unsupported links, then check transfer with delayed retrieval and first-seen questions.

What learning-by-teaching research shows

Preparing to teach can change how you study

Kobayashi's 2019 meta-analysis synthesised 28 studies comparing preparing to teach, or preparing and then teaching, with ordinary study without teaching expectancy. It reported positive average effects for both preparation and teaching, including delayed and deeper-learning outcomes. Effects were larger in some interactive teaching conditions.

A 2024 meta-analysis extended this question. Across 39 studies, teaching after preparatory learning had a small positive overall effect compared with merely studying without teaching expectancy. The teaching benefit was larger when learners had expected to teach while studying; teaching without that expectation did not show the same average benefit. A separate synthesis also found an added benefit of actually teaching after preparation rather than preparation alone.

These results make teaching expectancy a useful design choice: before opening a source, decide what you will need to teach and to whom. They do not show that every candidate should produce a lecture, or that longer explanations are better.

Retrieval may do much of the work

Koh, Lee and Lim compared learners who taught without notes, taught by reading a script, completed retrieval practice or did an unrelated task. One week later, the teaching-without-notes and retrieval-practice groups performed better than the script-reading and control groups, while teaching without notes and retrieval practice did not differ substantially.

That experiment supports a retrieval-practice explanation: teaching can help because the learner must reconstruct the material. A later review by Kobayashi concluded that retrieval is a plausible mechanism while also identifying unresolved questions about how retrieval and explanation generation interact.

For MRCPCH revision, keep the source closed during the first explanation. If every sentence is paraphrased while visible, the session may improve the document without testing whether you can access the idea independently.

Explaining to another person is not always better than explaining to yourself

An audience can change the task. You may organise the material, define unfamiliar terms and anticipate questions more carefully. But audience effects are conditional.

Lachner and colleagues combined experiments with a cumulative meta-analysis comparing instructional explanations with other learning activities. They reported a small average effect on conceptual knowledge, with oral explanations tending to perform better than written ones in their synthesis. The average transfer effect was small and not statistically significant. In one field experiment, self-explanation outperformed explaining in writing to a fictitious student for transfer.

This matters because the self-explanation guide and the Feynman-style task are related but not identical. Self-explanation asks you to make reasoning explicit for yourself. Learning by teaching adds audience design: what must a beginner know first, which term needs defining and what question would expose ambiguity? Use the audience only when it improves those decisions.

Interaction can help, but another person is optional

Interactive teaching can create questions, requests for clarification and corrective feedback. The meta-analytic evidence suggests that expecting and completing interactive teaching may sometimes be more beneficial than non-interactive teaching.

However, a real partner also introduces scheduling, social and accuracy costs. The partner may accept a plausible error, interrupt useful retrieval too early or steer the session into material outside the declared task. An imagined learner, empty chair, audio recording kept private or blank page can still require a complete explanation.

The study-groups guide covers group accountability and discussion. This article addresses the teacher's learning process and can be used alone.

The Feynman technique is not a simplicity contest

Plain language is a diagnostic tool. Replacing an unexplained technical term with an accurate description can reveal whether you understand the relation behind the label. But three shortcuts create false clarity:

1. Deleting conditions: a short statement sounds clear because exceptions or boundaries have vanished.
2. Replacing explanation with analogy: the analogy is memorable, but no longer maps accurately to the source.
3. Using fluency as proof: the talk is smooth because it has been rehearsed, not because the knowledge can transfer.

Keep necessary technical terms. Define them once, connect them to the explanation and then use them accurately. The goal is not to speak as though the audience knows nothing; it is to make every important link inspectable.

The dual-coding guide can help when a sequence, hierarchy, comparison or spatial relation genuinely benefits from a visual. Do not add a diagram merely to make the lesson look complete.

Use the TEACH framework

TEACH is a PaedVault editorial framework, not an RCPCH requirement or a validated assessment instrument.

T - Target one question

Set a narrow teaching objective before studying. Better prompts describe an output:

• "Explain the difference between these two concepts using three discriminating dimensions."
• "Teach the order of this process and justify each transition."
• "Explain why this answer option was preferred and what evidence would change the decision."
• "Define this examination term and show how it affects the study plan."

"Teach the whole syllabus area" is too broad. Use the current RCPCH theory syllabus to define the official boundary, then reduce the session to one explainable relationship. The RCPCH structure-and-syllabi page was last updated on 27 August 2026 and links to version 3 of the theory examination syllabi.

Write three audience questions before opening the source:

1. What must the learner know first?
2. Which relationship is most likely to be confused?
3. What fresh question would show that the explanation worked?

E - Explain without the source

Study the source briefly, then close it. Give a two-to-five-minute explanation from memory. A recording is optional; a live audience is optional; retrieval is not.

Use a simple structure:

1. state the question;
2. give the shortest direct answer;
3. define necessary terms;
4. connect the steps or comparisons;
5. give one boundary or counterexample; and
6. ask one checking question.

Mark pauses rather than hiding them. "I know these two ideas differ, but I cannot state the deciding feature" is useful evidence. Open-book fluency would conceal that gap.

The open-book and closed-book guide explains how to alternate an unaided attempt with a focused source check. Teaching from memory is the closed-book phase, not the whole learning loop.

A - Audit every claim

Reopen the authoritative source and compare line by line. Use four labels:

Audit label	What it means	Next action
Accurate	Claim and relationship match the source	Keep concise
Omitted	Necessary element or boundary is absent	Add and re-explain
Distorted	Simplification changed the meaning	Replace with precise language
Unsupported	Claim cannot be traced to the source	Remove or verify elsewhere

Do not correct only the words. Identify the reasoning failure: forgotten fact, confused comparison, missing prerequisite, unjustified causal link or outdated source.

If a partner listens, ask them to challenge clarity rather than certify factual accuracy unless they are using the same authoritative source. A confident audience reaction is not evidence that the explanation is correct.

C - Challenge the explanation

After correction, change the task. Use one or more of these tests:

• answer the three audience questions without the lesson script;
• explain the same relation in a different order;
• identify what would make the explanation fail;
• distinguish a near-neighbour concept;
• answer a first-seen question that depends on the relation; or
• retrieve the explanation again after a delay.

The timed-or-untimed question guide provides a wider attempt - feedback - reattempt progression. Teaching should improve the reasoning that enters the loop, not replace first-seen practice.

Record confidence before feedback. If the explanation felt excellent but the fresh task fails, you have found a calibration problem rather than a reason to rehearse the same speech. The confidence-calibration guide explains how to prioritise confident errors and uncertain correct answers.

H - Handoff a precise next return

End with a short restart instruction, not a vague promise to "review later". Examples:

• "Tomorrow: explain the three discriminating dimensions without the table."
• "In four days: answer five first-seen questions, then reteach only the failed link."
• "Next week: teach the process in reverse and identify the two omitted conditions."

Retire the lesson when the relationship can be retrieved after a delay, explained accurately and applied to varied questions. Continuing to polish the delivery after the learning job is complete turns revision into production work.

A worked non-clinical example

Suppose the study question is: "How does the TEACH loop prevent a fluent but inaccurate explanation?"

A weak attempt might say: "Explain it simply and you will understand it." That is memorable but circular. It gives no test of accuracy.

A stronger lesson is:

1. Target one relationship so omissions can be seen.
2. Explain without the source so retrieval gaps appear.
3. Audit against the source so fluent errors are corrected.
4. Challenge with a changed task so script memory does not masquerade as transfer.
5. Handoff a delayed return so the correction is tested again.

Now close that list and answer two audience questions: "Why must the source be reopened?" and "Why is a fresh question different from repeating the speech?" If those answers are weak, the lesson has diagnosed its own next step.

Use another person without creating a bad teaching session

If you work with a partner, separate roles and evidence.

The teacher

• explains from memory before feedback;
• states uncertainty rather than bluffing;
• answers clarification questions; and
• records corrections and a later return.

The learner

• asks for definitions, links and boundaries;
• identifies where the explanation became hard to follow;
• does not supply the answer during the initial retrieval attempt; and
• checks factual claims against the agreed source.

Switch roles only after the first explanation is complete. Keep each lesson short enough that both participants retrieve rather than one person delivering a long tutorial.

Do not share proprietary question text, answer options, explanations or study notes. PaedVault's current Acceptable Use Policy permits personal revision but prohibits copying or sharing protected content. Teach the underlying, source-checked idea in your own words and direct each subscriber to their own account.

Common traps and better replacements

"If I can say it simply, I must understand it"

Simplicity can expose understanding, but it can also remove necessary detail. Audit every important claim and boundary against the source.

"Making a teaching video is active learning"

Only if the process requires useful generation or retrieval. Reading a finished script into a camera may add production time without adding a memory test. Record privately only when replay helps identify gaps or audience clarity.

"I need someone junior to teach"

No. An imagined learner or blank page can provide an audience constraint. A real learner adds value when their questions improve the explanation, not merely because they are present.

"I should teach everything I revise"

Teaching is time-intensive. Reserve it for unstable comparisons, complex relationships, repeated errors or ideas that remain hard to explain. Use briefer retrieval for secure material.

"A partner understood me, so the explanation is correct"

Comprehensibility and accuracy are different. Check the source and use a fresh question. Two learners can agree on the same plausible error.

"The Feynman technique replaces question practice"

It does not reproduce selection among options, pacing, sustained attention or first-seen transfer. Use teaching as a targeted repair or retrieval activity inside a wider revision plan.

A seven-day teaching experiment

Test the method rather than assuming it suits every topic.

1. Choose two syllabus-bounded relationships that have produced errors or weak explanations.
2. For one, use TEACH; for the other, use an equal-length closed-book retrieval and feedback session without an audience explanation.
3. Keep source quality, study time and delayed interval as similar as practical.
4. Record omissions, distortions, unsupported claims and production time.
5. After a delay, test both through a brief explanation and fresh questions.
6. Compare accuracy, transfer, confidence calibration and time cost.
7. Keep learning by teaching only where the audience-oriented explanation adds value beyond a simpler retrieval method.

This personal comparison cannot establish a general effect or predict an examination result. It can show whether the extra work of designing a lesson solves a real problem in your revision.

Conclusion

The Feynman technique is useful when it is treated as a structured learning-by-teaching task rather than a slogan. Research supports cautious optimism about preparing to teach and teaching, while showing that retrieval, expectation, audience, format and feedback can change the result.

Use TEACH: Target one question, Explain without the source, Audit every claim, Challenge the explanation and Handoff a precise next return. The finished product is not the lesson, video or performance. It is an explanation that remains accurate after checking and can be used on a fresh problem after a delay.

Frequently asked questions

Does the Feynman technique work for MRCPCH revision?

It can support revision when you explain a bounded idea from memory, check every claim against an authoritative source and apply the corrected explanation to fresh questions. Meta-analyses report positive average effects for preparing to teach and teaching, but the studies are not MRCPCH-specific and do not establish a pass-rate benefit.

Do I need to teach a real person?

No. A real learner can add clarification questions and social accountability, but an imagined beginner, private audio recording or blank page can still require an audience-aware explanation. Use another person only when the interaction improves retrieval, organisation or feedback without introducing distraction or unchecked corrections.

Should I use notes while explaining?

Keep the source closed during the first explanation so gaps become visible. Afterwards, reopen the authoritative source and audit omissions, distortions and unsupported claims. Reading or paraphrasing notes may help create a reference, but it does not test whether you can retrieve and organise the idea independently.

Is the Feynman technique better than active recall?

It is not a clean alternative to active recall. Explaining from memory is itself a retrieval activity, with added organisation and audience design. Research suggests retrieval may account for an important part of learning-by-teaching benefits. Use a simpler retrieval prompt when designing a lesson adds time without improving accuracy or transfer.

Can I teach PaedVault questions to a study partner?

Teach the underlying, source-checked principle in your own words, but do not copy or share PaedVault question text, options, explanations or study notes. PaedVault content is licensed for personal use under its Acceptable Use Policy. Each subscriber should access protected material through their own account.

Sources

1. Royal College of Paediatrics and Child Health. Theory exams: structure and syllabi. Updated 27 August 2026.
2. Kobayashi K. Interactive learning effects of preparing to teach and teaching: a meta-analytic approach. Educational Psychology Review. 2024;36:26.
3. Kobayashi K. Learning by preparing-to-teach and teaching: a meta-analysis. Japanese Psychological Research. 2019;61(3):192 - 203.
4. Kobayashi K. The retrieval practice hypothesis in research on learning by teaching: current status and challenges. Frontiers in Psychology. 2022;13:842668.
5. Koh AWL, Lee SC, Lim SWH. The learning benefits of teaching: a retrieval practice hypothesis. Applied Cognitive Psychology. 2018;32(3):401 - 410.
6. Lachner A, Hoogerheide V, van Gog T, Renkl A. Learning-by-teaching without audience presence or interaction: when and why does it work?. Educational Psychology Review. 2022;34:575 - 607.
7. Lachner A, Backfisch I, Hoogerheide V, van Gog T, Renkl A. Learning by writing explanations: is explaining to a fictitious student more effective than self-explaining?. Learning and Instruction. 2021;74:101438.
8. Agarwal PK, Nunes LD, Blunt JR. Retrieval practice consistently benefits student learning: a systematic review of applied research in schools and classrooms. Educational Psychology Review. 2021;33:1409 - 1453.
9. PaedVault. Acceptable Use Policy. Updated 28 July 2026.