Tests aren't just a way for teachers to torture their students, according to a new study that finds the brain encodes better mental hints during test-taking than during studying alone.

Although many people view tests as a way to mark and grade students' progress, research has found that the act of retrieving information from memory actually makes remembering it easier. In other words, tests improve learning.

The mechanisms behind this improvement are still unknown. In the new paper, Pyc and her co-author, Kent State's Katherine Rawson, investigated pieces of information called "mediators." Mediators are concepts, ideas or phrases that connect one piece of information to another. To be a good mediator, the idea has to be both easy to remember and easy to link to the information you're trying to retrieve.

To investigate the effect of testing on mediators, Pyc and Rawson had 118 English-speaking participants learn 48 Swahili words. In this case, the mediators were English words that would help the learner link the Swahili word to its English meaning. For example, someone trying to learn the word "wingu," which means "cloud" in Swahili, might pick the word "wing" as a mediator. If the participant remembered wing, he or she could think of birds flying in the clouds, which would lead them to the English translation.

Some of the learners took tests on the information and then got to restudy the material before being tested again. Others only studied and restudied without tests. A week after learning the words, each group took a final test. Some tests required them to give the English translation from the Swahlii word alone. Another group got a test that gave them the Swahili word plus the mediator they'd used when learning the words. A third group not only had to translate the word, but they also had to remember and write down their mediator word.

Overall, the group that took practice tests did three times better than the study-only group, mimicking earlier studies that found benefits to test-taking. But the three types of tests also teased out whether people who pretested recalled their mediators, and, if so, whether the mediators helped them remember the correct translation.

As it turned out, more test practice made for better mediators. Those who had to recall both their mediator and the translation got scores averaging 51 percent if they'd been in the pretesting group and just 34 percent if they'd only studied. Those who didn't have to remember their mediator, just link it to the translation, also did much better if they'd completed practice tests.