When you want to remember something later, trying to recall it without looking can be more useful than another uninterrupted reread. This is retrieval practice, often called active recall. It means using memory to produce an answer, explanation or account of what you learned, rather than keeping the answer in view.
That does not make reading unnecessary. You need something to learn in the first place, and returning to the source helps you check omissions and mistakes. The research question is what happens when some additional reading is replaced with an attempt to remember.
The experiment where the answer changed with the delay
In Roediger and Karpicke's 2006 experiment, 120 university students studied short passages about the Sun and sea otters. Each student reread one passage and practised recalling the other. Researchers counterbalanced the passage and activity order.
During recall, students wrote what they remembered without seeing the passage or receiving feedback. Later tests scored the ideas they recalled, not exact wording. Different groups took that final test after five minutes, two days or one week.
The result depended on the delay:
- Five minutes: rereading produced 81% recall, compared with 75% after retrieval practice.
- Two days: retrieval practice produced 68%, compared with 54% after rereading.
- One week: retrieval practice produced 56%, compared with 42% after rereading.
These are group averages for the proportion of passage ideas recalled, not predicted exam marks. The experiment supports a benefit of retrieval over another reading for delayed retention under these conditions. It does not establish that every subject, learner or kind of assessment will show the same difference.
The practical implication is to distinguish performance immediately after studying from what remains available later. They need not favour the same study method.
Does this extend beyond a laboratory passage?
A 2021 systematic review by Agarwal, Nunes and Blunt examined 50 experiments in schools and classrooms, involving 5,374 students. Of the 49 effect sizes the authors could calculate, 57% indicated medium or large benefits. That percentage describes study effects, not the proportion of students who improved.
The comparisons included rereading and other classroom activities. Benefits appeared across different ages, subjects and question formats, but the review did not establish one optimal timetable or format.
Its scope matters. The search ended in January 2018, despite the 2021 publication date. It excluded unpublished studies, so publication bias remains possible. Only three of the 50 experiments took place outside the United States and Western Europe. The findings are encouraging classroom evidence, without representing every educational setting or every learner.
Many experiments provided feedback. The review did not establish the best timing for feedback, and differences between studies should not be treated as a direct experiment comparing immediate with delayed correction.
Remembering an answer and using an idea are different goals
Suppose you can recite a definition but cannot recognise an example. You have demonstrated something useful, but not everything the lesson requires.
Pan and Rickard's 2018 meta-analysis asked whether benefits of testing transfer to changed questions or contexts. It combined 192 effect sizes from 122 experiments, involving 10,382 participants. Compared with further exposure without testing, retrieval produced a positive average transfer effect.
The benefit was uneven. Transfer was stronger in some categories, including application and inference questions, and weaker in others, including previously studied information that had not been practised. Initial retrieval success and whether practice required the same kind of response as the later test also mattered. Analyses addressing publication bias further cautioned against assuming transfer in every condition.
That supports a more careful aim than memorising the answer on one card: practise the kind of explanation or application you need, then check whether you can handle a different example. It does not mean a recall quiz establishes expertise or guarantees success on an unfamiliar problem.
A short demonstration: recognition is not the whole task
Imagine a fictional workshop introduces this invented classification rule:
A “luma” shape is blue and has exactly three sides. Its size does not matter.
First, read the rule. Then look away and try to answer before returning to it:
- What are the two required features?
- Would a large blue triangle qualify?
- Would a small blue square qualify? Explain why.
- Would a red triangle qualify? Explain why.
Now check your answers: blue and exactly three sides; yes; no, because it has four sides; no, because it is not blue.
Simply recognising the word “luma” would not answer these questions. The first asks for the rule; the others ask you to apply it and explain exclusions. If you remembered “blue” but missed “three sides,” rereading has a clear purpose: correcting that particular omission.
This is an original illustration, not a validated learning exercise or a reported experiment. The deliberately simple rule makes the distinction visible; it cannot tell you how well you will learn complex material or handle less obvious changes in context.
For your own topic, the equivalent might be explaining an argument without the article open, drawing a labelled process from memory, or answering a question before checking the worked solution. Choose an output that matches what you need to understand.
Make rereading serve a purpose
A practical way to apply the evidence is to read a manageable section, put it out of view, attempt a relevant explanation and then compare it with the source. Correct errors explicitly. Revisit the material later rather than judging learning only by the answer you can give immediately.
This sequence is an editorial suggestion based on the research, not a tested prescription with an ideal number of minutes. If you cannot retrieve anything, return to an explanation, use a narrower question or seek clarification. The goal is useful practice, not proving that you can struggle without support.
The learning method is also separate from the conditions that let you study. Our attention guide discusses practical adjustments and accommodations. If you have chosen a study activity but struggle to start, turning intentions into action addresses making the first action concrete.
When reviewing a study session, ask what you could explain with the material out of sight, what needed correction and whether you could use it again later. Those questions give rereading a specific job and make room for practising recall.
About this article
Published by PsychPod. Sources checked 20 October 2026. Prepared with AI-assisted writing and editorial source checking. This article explains learning research; it is not a cognitive assessment or treatment guide. The “luma” demonstration is fictional and does not represent a study participant or testimonial.