Feynman Technique Review

The Feynman Technique is a learning method where you test your understanding by explaining a concept as if teaching it to someone else. If you can't explain it simply, you don't really understand it. FeynmanLM's AI-powered review loop puts this into practice for every card you study.

How the Feynman Loop Works

Each review session is a Socratic dialogue between you and an AI tutor:

1. You see the question — rephrased by the AI to prevent pattern-matching on wording.
2. You explain the concept — in your own words, as if teaching it. Typed or spoken.
3. The AI evaluates your explanation against a rubric covering:
   • Correct identification of the core mechanism
   • Causal reasoning ("because" and "therefore", not just "what")
   • Coverage of key concepts and their relationships
   • Absence of misconceptions
4. If your explanation is incomplete, the AI doesn't just tell you "you're wrong." It identifies the specific missing piece — the exact mechanism, causal link, or concept you glossed over — and asks a targeted follow-up question with a hint.
5. You answer the follow-up, and the loop continues for up to 3 iterations.
6. If you still haven't nailed it after 3 tries, the correct answer is revealed. You repeat it aloud to confirm you've absorbed it, then move on.

What Good Feynman Answers Look Like

The AI rewards explanations that:

• Name the mechanism, not just the outcome: "Insulin binds to the insulin receptor, triggering autophosphorylation of the receptor's tyrosine kinase domain, which activates the PI3K/Akt pathway..." is better than "insulin signals cells to take up glucose."
• Explain causality: why does A lead to B? What would happen if A were different?
• Use precise language without necessarily using jargon — clarity of reasoning matters more than vocabulary.
• Acknowledge scope — if a concept has exceptions or boundary conditions, noting them shows genuine understanding.

The AI will not accept vague or incomplete explanations, even if they're directionally correct.

The Follow-Up Question

When your answer is incomplete, the follow-up probe:

• Names the specific concept or mechanism you missed (e.g., "You described the outcome correctly, but what is the molecular mechanism that triggers GLUT4 translocation to the cell membrane?")
• Gives a hint to guide your thinking without giving away the answer
• Focuses on the single most important missing point — it doesn't pile on everything you got wrong at once

This is intentional: fixing one gap at a time is more effective than being overwhelmed with multiple corrections.

Voice-First Feynman Review

The Feynman Technique works especially well when you speak your explanations out loud — it forces you to organize your thoughts into complete sentences rather than vague bullet points in your head.

Enable Hands-Free Mode to run the full Feynman loop without touching your device:

1. The question is read aloud.
2. Speak your explanation — real-time transcription shows your words as you speak.
3. Your answer is submitted automatically when you stop speaking.
4. Feedback is read aloud.
5. If a follow-up question is asked, you answer it verbally.
6. Say got it or again when prompted for your SRS rating.

This makes it possible to do a proper Feynman review session while walking, commuting, or doing something else with your hands.

The Discussion Screen for Deeper Exploration

After any card, tap Discuss to go beyond the card's content:

• Ask "why does this happen at a deeper level?" to explore underlying mechanisms
• Ask "what would change if X were different?" to test causal understanding
• Ask "what are the exceptions?" to find the boundaries of the concept
• Say "explain it differently" to get an alternative framing that might click better
• Create a new card from the discussion — if you discover a concept worth remembering during the discussion, tap the Create Flashcard option in the chat menu

Setting the Understanding Threshold

In Settings → Understanding Required, you can configure:

• Force understanding before proceeding — toggle this on (default) to require mastery before moving on. Turn it off if you want a more relaxed review mode.
• Understanding required — the percentage score required to pass (default: 80%). Raise it to 90% for topics where you need deep mastery; lower it for topics where directional understanding is enough.

Why This Beats Standard Flashcards

Traditional spaced-repetition flashcards test recognition: you see a question, recall an answer, rate yourself, and move on. The problem is that you can rate yourself "correct" even when your understanding is shallow, because you're essentially matching your memory to the card you wrote.

FeynmanLM tests explanation: you have to generate a complete, correct account of the concept from scratch, and an AI evaluator assesses whether that account is actually right. There's no way to fool the system by recognizing the "correct" answer — you have to know it.

The result is that cards you mark "Got it" represent genuine understanding, not just familiarity.