91直播

Most textbooks present learning material in the same sequence: A chapter followed by a practice test, then another chapter, and so on. But what if practice tests were given weeks later? Would students retain more of what they had learned?

Test of Time

Students given study sessions in lieu of quizzes at first fared better on tests than those who were quizzed, but that pattern was reversed when both groups were tested on the material a week later.

*Click image to see the full chart.

SOURCE: Psychological Science

Probably, according to scientists taking part in a 4-year-old research program on cognition and learning that is being financed by the U.S. Department of Education鈥檚 Institute of Education Sciences. The program鈥檚 aim is to translate basic research in the cognitive sciences into practical applications for education.

鈥淥ver the past 30 years, we have learned an enormous amount through the cognitive sciences about how people think, or process information, and how we learn new information,鈥� said Lynn Okagaki, the commissioner of the institute鈥檚 National Center for Education Research, which oversees the grant program. 鈥淔or the most part,鈥� she added, 鈥渢he knowledge that has been gained has not made it into classroom practice.鈥�

Since its 2002 inception, the department鈥檚 Cognition and Student Learning, or CASL, program has given $41 million in grants to support 45 such 鈥渢ranslational research鈥� projects. Results from the first wave of those grants are trickling out now, and they shed light on a wide range of nuts-and-bolts learning questions.

Testing Helps

Studies coming out of the CASL program raise new questions, for example, about whether some of the 鈥渕anipulatives鈥� that teachers use to teach arithmetic skills may actually distract children from the learning at hand. Others are studying how particular academic skills develop in young children and involve designing artificial-intelligence-based tutoring systems and other interventions for teaching those skills more efficiently.

But a common thread among several of the programs is their focus on the role of memory in learning, noted Harold E. Pashler, a psychology professor at the University of California, San Diego.

The questions such studies address include: Is cramming a good study strategy? Does it hurt or help to guess at answers? How critical is it to give students feedback on wrong answers? Can just taking a test improve learning?

鈥淎 lot of educational failure is really forgetting,鈥� Mr. Pashler said. Though some widely reported studies have suggested that one-third of young Americans cannot locate the Pacific Ocean on a map, he said, 鈥淚 personally am completely convinced there was a time when they could.鈥�

At Washington University in St. Louis, one group of IES-funded researchers has learned that taking a test, in and of itself, can bolster learning.

Psychology professor Henry L. Roediger III and his research partners had college students read prose passages covering general scientific topics. Afterward, some students were given from one to three tests with open-ended questions in which they were asked to recall what they had read. Other groups restudied the material, but were not given tests.

Five minutes after the study or test sessions ended, all the students were tested. The study-only group did better, recalling more of what they had read than the quiz-only group. A week later, though, the test-taking group had the better grasp of the material鈥攅ven though they had had fewer opportunities to restudy the original material.

鈥淚 usually think of my tests as being for assessment purposes, but giving someone a memory test and having them retrieve information from memory actually changes memory,鈥� said Mr. Roediger, who is planning to take experiments into precollegiate schools this fall.

Mr. Roediger鈥檚 colleagues have also found that the learning benefits are greatest when the tests are composed of short-answer rather than multiple-choice questions. That鈥檚 a point on which several of the studies converge: Learning seems to stick when students are forced to generate their own answers to questions or their own definitions of words and concepts.

When to Review?

Mr. Pashler and his UC-San Diego research team taught new material to college students, gave them review sessions at different intervals鈥攖he next day, for instance, a week later, or a month later鈥攁nd then tested them again after a year.

The students who had reviewed the material at the one-month point remembered three times more than those whose practice sessions took place a day or a week later. Those and other findings have led Mr. Pashler and his colleagues to conclude that there may be an optimum ratio for spacing out study sessions: 10 percent to 20 percent of the interval over which students are expected to recall the information.

鈥淚f you want to teach people things so they鈥檒l remember them for five years, ideally you would want them to have a one-year study interval,鈥� Mr. Pashler explained. 鈥淭hat means if you have an exam the next day and you cram the night before, that鈥檚 the least efficient thing to do in terms of what you remember five years from now.鈥�

Such findings, he added, could have implications for both instruction and textbook design. That may especially be true for learning vocabulary, concepts, and definitions. This fall, the researchers will further test their results in San Diego-area high school classes.

Feedback Proves Key

In another series of IES-financed experiments, Janet Metcalfe and her research colleagues at Columbia University in New York City took some of the teaching principles emerging from the new literature鈥攖he findings, for instance, on spaced practice and repeated testing鈥攁nd incorporated them into a computer-based tutoring program for teaching advanced vocabulary to 6th graders in the South Bronx. The students, most of whom were considered at risk for academic failure, learned four times more than peers who were given flashcards, papers, and other supplies and told to study the same material themselves.

The pattern of results was the same when the researchers tested the program on graduate students. But which cognitive-science-inspired component, the researchers wanted to know, had been most successful? They wondered, too, whether students鈥� learning suffered when they made incorrect guesses to the questions posed in the tutoring program.

In a paper due to be published in a forthcoming special issue of Psychonomic Bulletin and Review, the researchers say the answer to the second question is no. Students who guessed incorrectly as part of the tutoring program performed on par with those who got the answers right the first time. The key was to provide corrective feedback. In fact, among all the program鈥檚 experimentally driven components, feedback turned out to be most important for learning.

鈥淭o err may be human,鈥� writes Ms. Metcalfe, a professor of psychology and neurobiology and behavior, 鈥渂ut giving corrective feedback is divine.鈥�

John T. Bruer, the president of the St. Louis-based James S. McDonnell Foundation, a leading private funder of cognitive-science research, said the interest in practical uses for memory research marks a shift from the period stretching from the 1970s to the 1990s, when researchers in the field looked to artificial intelligence for learning applications.

鈥淣ow, I think there鈥檚 a perception鈥攁nd I don鈥檛 necessarily agree鈥攖hat that thread of cognitive-science research has run out,鈥� he said. 鈥淩esearch on human memory is a good place to look for new tools and ideas.鈥�