Predicting Performance Improves Learning
By Cindy Nebel
Metacognition refers to the concept of “knowing what you know.” It involves the ability to make accurate predictions about how well you will be able to then retrieve information later on. The ability to make those predictions is very important. If you are aware of the fact that you do not know much or that you are unlikely to recall it later, you will probably allocate more time to studying. Students are notoriously bad at this (1), but see this blog for advice about improving metacognition.
Even though we know that students are not likely to predict their own learning, recent research indicates that asking them to do so might actually improve retention (2). Witherby and Tauber (3) were interested in this effect and whether it worked for long(ish) time periods between studying and testing. Here’s what they did:
Participants studied unrelated word pairs (for example: CHAIR – ZEBRA). Half of the participants were asked how likely (from 0 to 100%) they were to remember the second word in each pair (ZEBRA), when given the first word (CHAIR - ?) on a later test. Participants were then tested after 3 minutes or they came back after 2 days and were tested. There were therefore four conditions in the study:
1) Make predictions – Test at 3 min
2) Make predictions – Test at 2 days
3) No predictions – Test at 3 min
4) No predictions – Test at 2 days
The researchers also looked at whether it mattered if their instructions or the predictions were out loud or on a computer.
In all of the different manipulations and at both time points, those people who predicted their learning scored about 10% better than those who didn’t. This was true at 3 minutes and 2 days (although the overall scores went down at 2 days) and it was true regardless of instructions or format.
Why do these predictions work? The Witherby and Tauber argue that the only way to make this prediction is to look at the word pair and really analyze how they go together, which makes their similarity (or not) much more clear and that information becomes part of the memory. For example, if I saw the word CHAIR, I might think “hm, I remember that the word that went with it had almost nothing to do with furniture, except maybe with what it looked like… oh yeah! ZEBRA-print!” Realizing how strongly they are associated might help with recall.
While these data are certainly compelling, there are a few limitations that need to be addressed. First, note that participants were studying unrelated words and their study was limited to 8 seconds for each word. That’s a pretty short amount of time for materials that look quite different from what we might see in the classroom. Second, the delays are rather short. While, yes, we do want students to remember things from one lesson to another (as might be relevant in the 2 day conditions), ideally they will remember material for quite a bit longer… at least until their next course and hopefully later in life!
Sometimes laboratory studies are so different from real-world situations that I am hesitant to make recommendations based on that work. However, in this case, I think there is an easy implementation that is likely to help students succeed. My recommendation is that they simply ask themselves, “How likely am I to remember this later?” while they are studying. While they don’t necessarily have to determine a percentage, the act of considering how much they know will likely change the way they are analyzing the information and provide insight into how well they understand the material.
Note that these effects are on an item by item basis. There are other predictions that can be made overall and do not have these same kinds of effects, so students should be asking how well they will remember each concept they are studying and not just how well they will do on the exam.
For more information on these topics, consider reading the following blog posts:
(1) Maki, R. H. (1998). Test predictions over text material. In D. J. Hacker, J. Dunlosky, & A. C. Graesser (Eds.), Metacognition in educational theory and practice (p. 117-144). New York: Routledge.
(2) Soderstrom, N.C., Clark, C.T., Halamish, V., Bjork, E.L. (2015). Judgments of learning as memory modifiers. Journal of Experimental Psychology: Learning, Memory, and Cognition, 41, 553-558.
(3) Witherby, A.E., & Tauber, S.K. (2017). The influence of judgements of learning on long-term learning and short-term performance. Journal of Applied Research in Memory and Cognition, 6(4), 496-503.