By Nate Kornell, Williams College
Teachers give tests to find out what their students know. But tests do a lot more than that and can have a powerful effect on what a student remembers. In a typical research study looking at the links between tests and memory, one group of participants might be told “Batman’s butler’s name is Alfred”. Another group would be asked “What is the name of Batman’s butler?” and then told it’s Alfred. If you ask again weeks later, the group that was tested remembers that the answer is Alfred better than the group that was just told the information.
But what happens if you get the answer wrong? For example, what if you say Jeffrey (a common error, in my research back when The Fresh Prince of Bel Air was on TV), and I say no, it’s Alfred. Would you have been better off not being tested?
Common sense says if you practice making errors you learn to make errors. From this has grown a long tradition in psychology aimed at promoting “errorless learning”. On the other hand, common sense also says we learn most from making mistakes. So which is it?
Does age matter?
Existing research in this area has shown that it depends on age. If you’re young and healthy, mistakes enhance learning. But people with memory impairments, including the mild impairment that comes with normal ageing, benefit most from errorless learning. In short: nimble mind, errors are fine; for granddad, errors are bad.
New research challenges all that. Psychologists Andrée-Ann Cyr and Nicole D. Anderson noticed that in prior studies with younger adults, the tests typically involved asking people about concepts, such as to name a kind of fruit. But in studies with older adults the tests typically involved filling in lexical cues, such as “st-____”.
Even though both groups are then told the answer is strawberry, the learning is different, with those younger people who made errors when asked to name a word based on a concept, remembering the answer weeks later. But it was unclear what mattered most: being young or old or if the cues in the test were conceptual or lexical.
Getting it wrong helps
In their study with groups of both old and young people, Cyr and Anderson have now found that the types of clue makes all the difference. In their conceptual test, both groups remembered more from a test that they didn’t get right – such as being asked to name a pastry, followed by feedback that “it was a tart” – than they did if they had just been told the answer straight away without being tested on it.
In their lexical test, both groups learned more from errorless learning. They remembered more words in a later test if they were simply shown “st and strawberry”, than if they’d been shown “st____” completely out of context, made a wrong guess about the word, and were then shown the answer “strawberry”. The same pattern held for both age groups.
Most of what we learn is conceptual, in the sense that it involves relating new learning to information we’ve learned before. For example, if you were to take a quiz about this article, it wouldn’t have questions on it like “er____?”, it would have questions like “when does errorless learning help and when does it hurt?”
So the practical lesson of this new study is that making errors (and then getting feedback) is a good way to learn and retain conceptual information. If you want to learn non-conceptual information, such as linking words to meaningless non-words, making errors will not help. In rhyme, it goes like this: to really understand, errors are grand; for meaningless stuff, being right is enough.
The test’s the thing
Yet these studies are, understandably, not completely realistic. The participants were basically just guessing at the answers. Also, in real-life tests it’s common to write down a wrong answer and not find out the correct answer for a while, for example when you take a test and then it is returned a week later.
Another recent study might allay these concerns. Participants were asked questions such as: “What is the world’s tallest grass” that they could really think about, instead of blindly guessing. After making errors on an initial test, they had to wait 24 hours to find out the correct answer. Even so, trying to answer the question on the first day and getting it wrong led them to remember more in a later memory test than not being tested on the information at all.
Kids need to be challenged
We all want kids to do well in school. But these results raise a difficult question: what does it mean to do well? We often assume it means doing well in the classroom, when answering questions or taking tests. But if the ultimate goal is to do well after graduation, then doing too well in school might be a problem. Perhaps we should actually be concerned when kids aren’t making errors in school because they could be learning more if they faced bigger challenges.
There is no evidence that it’s good to make errors on purpose. But teachers do need to make sure that the set of problems a child faces is challenging enough so that he or she is engaged in productive struggle. If you’re not making mistakes, you might not be learning.
Nate Kornell receives funding from the James S. McDonnell Foundation.