Methods · Generative practice & retrieval · Short guide
B23
Retrieval-before-reading
Answer questions on a text before you read it, so the reading has gaps to fill instead of washing over you.
When to use it
- You are about to open a chapter and usually highlight on autopilot.
- You want the reading pass to answer questions you already failed.
- Pre-questions from a lecturer helped before, and you want that on demand.
What to ask AI
Type something like: “Before I read [chapter/section], ask me 5 questions I should be able to answer afterward. Wait for my pre-reading attempts, even guesses, before I open the text.”
Ask for questions that target structure, not trivia alone.
Only after pre-answers, go read, then re-answer the same questions.
How it helps
Questions before reading create hooks. Your brain notices answers when they appear.
Guessing wrong is allowed; the wrong guess still primes attention.
Re-answering after reading shows what the text actually taught you.
Skipping the pre-attempt turns this back into ordinary reading.
How to run it
- Name the text or section you are about to read.
- Get five pre-questions; answer from whatever you already know.
- Read with those questions in mind.
- Re-answer the same five without the book.
- Note which questions stayed wrong, that is the next study target.
Example exchange
Before I read the section on light-dependent reactions, can you give me a few pre-reading questions? I will guess cold first, then read.
First cold question only: what molecule is split to supply electrons in the light-dependent reactions?
Water? I am not totally sure.
Fine as priming. Second: what two energy carriers does that section say are produced?
ATP and something with an N. I cannot name it cleanly.
Good enough priming. Read the section, then re-answer those same questions without notes. I will wait.
Copyable prompt
I am about to read: [SECTION].
Ask 5 questions I should be able to answer afterward. Wait for my pre-reading
attempts (guesses OK). After I read, ask the SAME five again before any teaching.
The Tell
You will know the method has flipped when you glance at the questions and dive into the chapter without answering, planning to “think about them while reading.”
Without a pre-attempt, nothing is primed.
Write five rough answers first, wrong is fine, then open the text.
Principle evidence
Strength of the underlying learning idea, not a claim about AI products.
Prequestions and advance retrieval attempts can improve learning from subsequent text by directing attention and activating related knowledge. The pre-attempt is the load-bearing move.
AI delivery evidence
Whether an AI tutor delivers this method well is a separate question.
AI-chosen prequestions may miss your syllabus emphasis. Prefer lecturer outcomes when available; otherwise edit the list before you pre-answer.