Desirable Difficulties to Create Learning
Imagine that you are a student and you have an upcoming exam to prepare for. What do you do?
I have now asked this question of hundreds of students and teachers. Invariably, one of the first responses that pops up is some variant of: “Reread my textbook or notes!” Following on fairly quickly, someone will say, “Do practice problems!” or “Take practice tests!” This isn’t just my own anecdotal experience: In surveying students’ study behaviors, Karpicke, Butler & Roediger (2009), found that repeated rereading was by far the most frequently reported strategy (84% of students, with 55% listing it as their number one strategy used when studying). In distant second was doing practice problems, with 43% of students reporting it, and 12% listing it as their #1, go-to strategy.
If I ask students when do they take these practice tests, the answer is almost always, “At the end of studying” or “Closer to the exam date.” That is, people study to learn, and the test themselves to check what they have learned. Indeed, this is what Kornell & Bjork (2007) found when they surveyed 472 introductory psychology students at the University of California, Los Angeles: 68% indicated that they quizzed themselves to figure out how well they have learned the information that they are studying. Only 18% indicated that they quiz themselves because they learn more that way than from reading, and a worrying 9% reported that they never quiz themselves. These behaviors are representative of the broader US population, and not just limited to a group of psychology students: I found a very similar distribution of responses from a survey of 450 people recruited from Amazon Mechanical Turk (age range 18-74).
We know that learners of all ages like rereading, but how effective is rereading, compared to self-testing?
Roediger & Karpicke (2006) gave participants a text passage (about the Sun or about sea otters) to study.
Half of the participants studied this passage for a total of 20 minutes (we will label this group, “SSSS”, which refers to four 5-minute sets of “studying”).
The other half studied the passage for only 5 minutes before it was taken away. They were then given a blank piece of paper and told, “write down everything you can remember”. They did this for five minutes before that piece of paper was taken away and given a new blank piece of paper to repeat this test again. In total, after the initial 5-minutes of studying, they took three tests (we will label this group, “STTT”).
Five minutes after study (left side of the graph), those who studied more, recalled more.
However, one week later (right side of the graph), there was a large reversal of the pattern of results: Those who studied more had forgotten most of what they had “learned”. On the other hand, those who tested themselves repeatedly, held on to the information that they had learned. In other words, the very act of testing oneself (i.e., retrieving information from memory) works to strengthen that learning, and it does so much more effectively than repeated rereading*.
The Need to Introduce Desirable Difficulties to Create Learning
The testing effect is just one example of a desirable difficulty. It is “difficult” because it forces the learner to be more active and expend more effort to retrieve information from memory (as opposed to passively rereading a chapter). It is “desirable”, however, because engaging in that active retrieval in fact strengthens learning and leads to better long-term retention of information.
“Desirable difficulty” is a term coined by Professor Robert Bjork of UCLA, and refers to a set of study strategies that create conditions of learning that introduce difficulties, engage learners actively, and may appear to impede performance but which, in reality, create more durable learning than the conditions that, superficially, make learning feel easy.
|Desirable difficulties include:
For example, when learning new information, instead of immediately starting by reading the chapter or listening to a lecture on a particular topic, a learner might ask himself/herself what they already know about that topic and trying to answer a few questions about they key terms or concepts from that chapter. Even if they do not know the answers, just activating relevant concepts surrounding that topic will mentally prepare you to absorb new incoming information when you do finally read the chapter. (Pretesting effect)
Or, let’s say you want to reread that chapter after you’ve read it once. One might be tempted to read it again immediately because you feel like you’ve “almost” got it. Instead, it is better to take a break and turn your mind to something else before coming back to that chapter again. This latter option of spacing out your studying might feel more difficult (e.g., you realize that you have forgotten information when you come back to it that second time, and feel as though you are having to re-tread over old ground), but this difficult in fact belies stronger, more “sticky” learning. (Spacing effect).
These are just a few examples of what desirable difficulties are and how one might incorporate them in study.
An important thing to note is that these desirable difficulties do not take more time. In fact, just about every experiment is very careful to ensure that they are comparing conditions in which people spend the same total amount of time studying, manipulating only what they do in that study time or how they spread out that study time.
The Critical Distinction between Learning and Performance
Learning = flexible and durable learning; learning that lasts
Performance = how well a learner is doing currently
Performance is what we often experience: How well do we feel like we understand something that a teacher is saying or that we are reading? Am I getting these practice questions correct? And based on the answers to these questions, we make inferences about how much we are learning. However, as the examples above show (as well as many, many research experiments), good performance does not always equate to good learning.
Reading and rereading, for example, might lead to feelings of familiarity and fluency with the to-be-learned material. Indeed, testing oneself immediately after learning, we might think we know a lot (for example, the people in the SSSS condition performed better than the people in the STTT condition when tested after five minutes). However, these feelings of fluency are misleading. We only feel like we have learned that information because it is fresh in our minds and we’ve just seen it several times in quick succession; the information, however, is not deeply anchored and will be quickly forgotten. On the other hand, effortfully retrieving information might feel very difficult and yet it is this effort that end up securing that information into our minds much more effectively for the long-term.
TL;DR:1. Introducing “desirable difficulties” leads to better long-term learning than the strategies that make learning feel easy.
2. Performance is not the same as learning, and rapid improvements in performance (e.g., feelings of fluency while studying) can often misled people into thinking that they have learned.