Many Japanese students use a red sheet to hide answers in their word books. Then they try to remember the answers. This is called retrieval practice, and it is one of the best ways to learn.

A 2013 study found that answering in your head works as well as saying the answer aloud. But be careful. Do not look at the answer too quickly. And do not stop practicing a word after one correct answer.

The red sheet, a favorite tool of Japanese exam takers, works in the same way as retrieval practice, which is trying to recall an answer before checking it. A major 2013 review of study methods gave practice testing its highest rating. In the same year, Smith and his colleagues found in four experiments that recalling answers silently improved memory just as much as saying or writing them, and more than rereading.

However, two habits can weaken the effect. One is checking the answer too soon. A 2009 study showed that trying to answer first, even unsuccessfully, helps learning. The other is dropping words too early. When students were allowed to remove cards they thought they knew, their scores were slightly but consistently lower.

The red sheet, ubiquitous among Japanese exam candidates, embodies retrieval practice, one of the best-supported techniques in learning science. In a 2013 review of study methods, Dunlosky and colleagues gave practice testing their highest rating, while rereading was rated low. Smith, Roediger and Karpicke showed across four experiments that covert retrieval, answering only in one's head, enhanced retention as much as overt retrieval, with no difference on a test two days later, and that covert retrieval outperformed rereading. Two common habits, however, can undermine these benefits.

First, peeking too soon: Kornell and colleagues found that attempting an answer before seeing it improved later memory even when the attempt failed. Second, retiring items prematurely. When Kornell and Bjork let students drop flashcards they believed they had mastered, performance was slightly but consistently worse, apparently because students overestimated their own knowledge. These findings come with caveats.

The covert condition was carefully designed to ensure that recall actually happened, whereas real learners may glance at the answer without genuinely trying, and the benefit of failed attempts may vary with the material. A practical rule is to retrieve fully before checking and to retire a word only after several successful recalls on different days.