sciencegeekgirl » We learn by taking tests (even when we get the wrong answer)

: photo by Patrick Hannigan (click for Wikimedia link)

We think of taking tests as something to assess whether we learned something, but there is a fascinating set of literature that shows that it does more than that. Tests can be learning events in their own right. It makes sense when you think about it. How is it that we learn things? By making neuronal connections, or strenthening neuronal connections, in the brain. Each time we take a test and are asked to recall information, that neuronal path gets strengthened. That’s why flash cards are useful. One of the seminal papers on this topic is The critical importance of retrieval for learning by Karpicke and Roediger.

When someone recalls something from memory, they’re more likely to be able to recall it again later — much more so than when that information is just presented to the person.
People forget information more slowly when tested on it
Asking people questions whose answers involve numbers increase people’s retention of numbers presented in text (by directing their attention to the type of information important to learn for the test)
Questions asked before a task can activate prior knowledge and focus students on the relevant material

Robert Bjork of UCLA, who studies learning and forgetting, has written extensively on this topic, especially given that students don’t really know how it is that they learn, and their study habits don’t make the most effective use of what we know about our cognitive function. His “how to succeed in college” paper is a nice summary of this research.

But, you might ask, what if you take a test and get the wrong answer? Doesn’t that then cement the wrong answer in your brain? So isn’t there a danger to testing ourselves when we might get the wrong answer stuck in our head? Some research suggests that testing could distort knowledge in this way: When you get the wrong answer on a multiple choice test, you’re more likely to make that same mistake on a later test.

Giving someone feedback on their performance on a test can reduce memory distortions, but this is sometimes not feasible (especially in today’s climate of standardized testing). Luckily, new research shows that, no, making mistakes still helps us learn. A set of two articles authored and co-authored by Nate Kornell sheds some light on these questions:

When they tested students before they studied a text, students did better on a test after having studied the text, even though they got most of the questions wrong. This appeared to be due to the testing itself, rather than focusing students’ attention on the important aspects of the text, because students learned better when tested than when key information in the text was bolded. (Note the implications for students’ exuberant highlighting of texts!)
This kind of testing also reduced forgetting after a one week delay.
This kind of testing was also more effective than reading the same question and trying to memorize the words of the question itself (without trying to retrieve the answer).
In another study, where students were doomed to fail (being tested on word association pairs that most people get wrong), they found the following: Trying (and failing) to answer a question, and then studying it, produces better learning than studying it (for a longer time) without first trying to answer it.