Page 1
A Field Guide to Retrieval Practice
This brief is written so you can test Folio immediately. Every claim below is contained in these pages. If a question is not answered here, the correct reply is that the source does not say.
Retrieval practice is the act of pulling knowledge out of memory rather than putting it back in. Rereading a chapter, highlighting sentences, and copying notes feel productive because they are fluent. Fluency is a feeling. Learning is a change in what you can later produce without the page in front of you.
The testing effect is the finding that an attempt to recall — even an unsuccessful one — strengthens memory more than another exposure to the same material. A closed-book recitation, a blank-page outline, a handful of questions answered from memory: these are retrieval. A second pass over highlighted paragraphs is restudy.
Karpicke and Roediger (2008) compared conditions in which students restudied word pairs versus tested themselves on them. Repeated study produced higher confidence and poorer delayed recall. Repeated retrieval produced the reverse: lower immediate ease, higher later performance. The lesson is not that studying is useless. It is that study without a later retrieval attempt is an incomplete loop.
Page 2
How to work a source, not just read it
A source is useful only if you can answer questions from it later. Treat a paper, a chapter, or a brief as a set of claims, terms, numbers, and procedures.
1. Preview the architecture. Read the title, abstract or first paragraph, headings, and the last paragraph. Write one sentence: what is this trying to change in my mind?
2. Extract the load-bearing sentences. A load-bearing sentence is one the argument collapses without: a definition, a result, a rule, a number, a named mechanism.
3. Close the page. Recite the load-bearing sentences in your own words. Then open the page and correct the recitation against the wording of the source. Correction is the learning event.
4. Convert each load-bearing sentence into a question whose answer is that sentence. “What is the testing effect?” is better than “What is section two about?”
5. Space the questions. Ask them again after a night, then after three days. Spacing is not a personality trait. It is a schedule.
A common failure mode is the highlight trap. Color on a sentence does not encode it. If you highlight, immediately write a question on the verso that the highlighted line answers. If you cannot write the question, the highlight was decoration.
Another failure mode is the summary that never leaves the page. A summary written while looking is restudy. A summary written from memory, then checked, is retrieval.
Page 3
Terms you must be able to produce
Retrieval practice — An attempt to produce a target from memory without looking. The attempt itself is the treatment.
Testing effect — The empirical advantage of retrieval over restudy on a delayed test.
Fluency illusion — The mistaken inference that ease of rereading equals readiness to perform. Ease tracks familiarity, not recall.
Desirable difficulty — A practice condition that feels harder now and yields more later. Generation, spacing, and variation are typical sources.
Spacing — Separating retrieval attempts in time rather than massing them in one sitting.
Interleaving — Mixing problem types or topics inside a session so that each attempt requires discrimination, not a routine.
Transfer — Using knowledge in a form the practice did not literally rehearse. Retrieval of concepts in new wording is a weak form of transfer; applying a rule to a new case is a stronger one.
Source-grounded answering — Restricting claims to what a given document actually states, and naming the page when a number, definition, or proper name is used. This brief requires that discipline of any assistant that reads it.
What this brief does not contain: sample sizes from the 2008 paper, classroom hour counts, a comparison of digital versus paper, or medical or legal advice. Those absences are intentional. They are the correct place to say “the source does not say.”
Page 4
A compact protocol for the next hour
Minute 0–8. Preview. Write the one-sentence change this source wants.
Minute 8–20. Read once, marking only load-bearing sentences. No more than twelve marks.
Minute 20–32. Close the source. Recite. Open. Correct.
Minute 32–40. Write eight questions. Four should be definitional. Two should demand a number, name, or rule. Two should demand a contrast (retrieval vs restudy; fluency vs learning).
Minute 40–50. Answer the eight from memory. Score them against the page, not against a feeling.
Minute 50–60. Schedule the same eight questions for tomorrow and for three days from now. Do not add new material until these eight can be produced cleanly.
If you use Folio on this brief, good first questions are:
- What is the difference between fluency and learning according to this source?
- What five steps does the author give for working a source?
- Which findings are attributed to Karpicke and Roediger (2008)?
- What does the source explicitly say it does not contain?
A correct assistant will quote or paraphrase only these pages, cite them as [p. N], and refuse questions this brief does not answer. That refusal is not a failure. It is the point.