CAN TWO WRONGS MAKE A RIGHT?: Not knowing the answer can be a good thing.
Magdalena Tworkowska

For years, many educators have championed “errorless learning," advising teachers (and students) to create study conditions that do not permit errors. For example, a classroom teacher might drill students repeatedly on the same multiplication problem, with very little delay between the first and second presentations of the problem, ensuring that the student gets the answer correct each time.

The idea embedded in this approach is that if students make errors, they will learn the errors and be prevented (or slowed) in learning the correct information. But research by Nate Kornell, Matthew Hays and Robert Bjork at U.C.L.A. that recently appeared in the Journal of Experimental Psychology: Learning, Memory and Cognition reveals that this worry is misplaced. In fact, they found, learning becomes better if conditions are arranged so that students make errors.

People remember things better, longer, if they are given very challenging tests on the material, tests at which they are bound to fail. In a series of experiments, they showed that if students make an unsuccessful attempt to retrieve information before receiving an answer, they remember the information better than in a control condition in which they simply study the information. Trying and failing to retrieve the answer is actually helpful to learning. It’s an idea that has obvious applications for education, but could be useful for anyone who is trying to learn new material of any kind.

In one of their experiments, students were required to learn pairs of “weak associates,” words that are loosely related such as star-night or factory-plant. (If students are given the first word and asked to generate an associate, the probability of generating the target word is only 2 or 3 percent.) In the pretest condition, students were given the first word of the pair (star- ???) and told to try to generate the second member that they would have to later remember. They had 8 seconds to do so. Of course, almost by definition, they nearly always failed to generate the correct answer. They might generate bright in the case of star-???. At that point they were given the target pair (star–night) for 5 seconds. In the control condition, students were given the pair to study for 13 seconds, so in both conditions there were a total of 13 seconds of study time for the pair.