A Really Hard Test Really Helps Learning

Challenging tests and falling short may be hard on the ego, but they can do more than mere studying for eventually getting it right.
Publish date:
Social count:
Challenging tests and falling short may be hard on the ego, but they can do more than mere studying for eventually getting it right.

The first time Sarah Patterson got pimped by her attending doctor, it was a distinctly unpleasant experience.

A medical student at the University of California, San Francisco, Patterson had just begun a rotation on the wards of the city's General Hospital. While doing rounds, the doctor asked her, in front of their entire medical team, to list all of the causes of atrial fibrillation — a kind of medical school pop quiz that Patterson and her fellow students refer to as "getting pimped."

"I didn't know all of them, and I fumbled and tried to string together what I knew into a coherent answer," she recalls. "It wasn't fun." But, Patterson says, as unpleasant as the incident was, it was effective. "I know a lot about atrial fibrillation now. Fear and failure are good motivators."

Patterson's experience is one with which anyone who has ever taken a pop quiz can doubtless identify. And her interpretation of it — that failure inspired her to learn everything she could about atrial fibrillation — is supported by research I recently published in the Journal of Experimental Psychology: Learning, Memory, and Cognition. Working with Robert Bjork, a professor of psychology at UCLA, and Matthew Hays, a former graduate student there, we found that two widely held assumptions about learning — one about testing, and the other about making mistakes — are based on misconceptions.

Education theory has long upheld the virtue of errorless learning. Learn something right the first time around, the idea goes, and you won't be plagued by errors in the future. This belief isn't confined to the classroom. "Don't practice mistakes," coaches, music teachers and educators across the country tell their pupils.

But our research indicates that errors are not necessarily the enemy of learning; they can, in fact, enhance it. Likewise, the widely held belief that testing serves no purpose other than assessing performance is built on a similar misconception. In reality, testing — whether self-testing or testing in the classroom — can, under the right conditions, better promote learning than can studying.

A recurring theme in the field of cognitive psychology is the human tendency toward overconfidence. Research shows that people routinely overestimate not only their ability to understand the world, but also how well they perform both simple and complex tasks, from driving to accurately recalling the details of an event that occurred just a few days prior.

A recent study by researchers at Cornell and the University of Illinois, for example, found that students in a sophomore psychology class overestimated their performance on a test even after being promised up to $100 if they could accurately assess how well they did. As the researchers pointed out, the same pattern has emerged in studies of hunters (quizzed about their knowledge of firearms), medical residents (evaluating their patient-interviewing skills) and college debaters (assessing their performance after an event), among many others.

The frequent gap between confidence and performance makes clear the value of testing as a diagnostic tool. Tests permit us to relinquish responsibility for assessing our own performance — the test does it for us, and, if it's at all well designed, does so more accurately than we would be able to manage on our own.

What is less well understood is the value of testing as learning tool, quite aside from its diagnostic worth. Not long ago I, along with Lisa Son, a professor of psychology at Barnard College, conducted a study in which we found that students studying word pairs recognized that testing was the best way to monitor their learning — to find out what they did and didn't remember about the word pairs after studying them.

What they didn't appreciate, however, was the virtue of testing as a learning strategy in its own right. Students who studied the word pairs one time through and then took a quiz on them were better able, later on, to remember the pairs than students who simply studied them twice. But the students in the first group thought they performed worse than they would have had they been able to study the words again.

This confusion is widespread, and it has implications for everyday life. Anyone who has to renew their license, or has decided to take up photography as a hobby, or has become interested in birding, or would like to be able to identify important congressional players in the fight over health care — these and any number of other disparate occasions for learning can be made more productive with the help of a test. Next time you have to commit a chunk of material to memory, consider devising one to improve the process.

But it also means something for national education policy. The merits — or lack thereof — of testing as an educational tool has, since the introduction of the No Child Left Behind Act under President George W. Bush, become a polarizing topic in American politics. NCLB instituted a system of standardized tests nationwide, which are meant to act as a kind of national barometer for educators — and for federal officials deciding where to send education funding — that is, as a tool for comparing performance across schools and school districts.

Critics of NCLB argue, among other things, that it forces educators to "teach to the test" and takes away time from studying, which, the assumption goes, is where actual learning occurs. But as our research indicates, so long as students are made to review their mistakes, testing performs a vital function as a learning tool. Educators and policymakers might consider this when they debate the merits not only of NCLB, but also when they formulate curricula and design lesson plans.

If testing is a good way to learn, then, an obvious question arises: How difficult should tests be? The answer, according to our study, is quite difficult — a fact that runs counter to at least one well-established strain of American educational theory.

Since at least the early 1960s, some education theorists have upheld the virtue of errorless learning. The idea is straightforward, and it makes intuitive sense: The best way to learn is the way that involves allowing into the process as few initial errors as possible. B.F. Skinner, a Harvard psychologist in the 1950s and '60s, was an important proponent of this theory. In 1958, Skinner published a study in which he proposed, based on his work with rats and pigeons, a set of guidelines about how people should approach complex tasks, such as education.

Skinner argued that based on his results with pigeons and rats, the best way to learn is also the most error free. As he put it, "It is a salutary thing to try to guarantee a right response at every step in the presentation of a subject matter." He also said, "There is no evidence that what is easily learned is more readily forgotten."

Skinner was wrong on both counts. Challenging materials lead to errors, but they also make learners more active participants in their learning and, as a result, create long-lasting memories (even in monkeys). Cognitive psychologists sometimes use the term "desirable difficulty" to refer to this phenomenon. The idea shouldn't be taken too far — obviously, getting drunk before a test is unlikely to improve one's prospects. But it's remarkable just how dramatically a challenge inspires focus and memory.

In our study, we asked nearly 200 UCLA undergraduates to answer questions that they could not, in fact, answer — in some cases because the questions were too difficult, and in others because the questions were, unbeknownst to the students, fictional. "What peace treaty ended the Calumet War?" we asked, and "What is the name of the sailor who took the first solo voyage around Cape Evergreen?"

In one condition, the questions and answers were shown together; in another, the questions were shown alone at first, and then the answer was shown. Students in the second condition, faced with questions they were extremely unlikely to answer correctly, nevertheless ended up performing significantly better on a later test of their knowledge than their counterparts, who had been exposed to the answers — along with the questions — at the outset of the study.

We drew an obvious conclusion from this result, but it runs counter to what many people might consider common sense. Clearly, students in the second condition learned more by trying unsuccessfully to think of the answers on their own than their counterparts, who simply studied all of the information we gave them.

In case this seems too strange to be credible, consider another study we conducted demonstrating the same phenomenon. We gave 400 University of California, Irvine undergraduates a brief text about colorblindness written by the neurologist Oliver Sacks. We found that the students learned more when they were asked questions about the text before they read it — questions they could not answer — than when they were given additional time to study Dr. Sacks work.

Both studies independently indicate a striking fact. We tend to assume that the best way to consume and remember information is through the application of rigorous, extended study. What we fail to see, however, is that the process of trying to work through a problem to which we don't know the answer focuses our attention on it in a way that simply studying it does not. The desire to get the answer right, and the frustration of failure, is partly to account.

But there's another element as well. When we struggle to learn something, and fail, the moment we finally get the answer it imprints itself more deeply on our mind than it would have had struggle and failure not preceded it.


Chick Hearn, the late broadcaster for the Los Angeles Lakers, used to preach "don't practice mistakes" — advice Sam and I considered gospel, at least on the basketball court. In retrospect, Chick was partly right: Yes, you shouldn't practice mistakes, but neither should you make practice so easy as to avoid making mistakes.

If I had to identify one overarching lesson from our study it would be this: When you make mistakes, don't just let them slip by — correct them. Create challenging learning environments, make mistakes and then learn from them.

Sign up for our free e-newsletter.

Are you on Facebook? Become our fan.

Follow us on Twitter.

Add our news to your site.