On this page
You finish a chapter, close the book and think, "I've got this." Two days later the test says otherwise. That gap is not laziness. It is a known flaw in how people judge their own memory, and it can be measured and corrected. This post explains why your sense of knowing misleads you and gives you a routine to check it.
What is a judgment of learning?
A judgment of learning is your prediction, while studying, of how likely you are to remember something later. You make these judgments all the time: when you decide a page is done, when you skip a topic in revision, when you stop doing problems from a chapter. If the judgments are wrong, every decision built on them is wrong too.
A review by Bjork, Dunlosky and Kornell (2013) in the Annual Review of Psychology sums up the research: people often have a faulty mental model of how they learn, which makes them prone to misjudging and mismanaging their own learning. Two reasons come up again and again: what you can do right now, and how easy the material feels.
Why does easy-feeling material fool you?
In one study, Rhodes and Castel (2008) showed people words in large or small print. People predicted they would remember the large-print words better. They did not; font size had little to do with recall. The effect held even after people were warned that font size does not matter. When large words were made harder to read, the effect went away, which suggests that ease of reading was driving the judgment.
The same thing happens with study methods. Kornell and Bjork (2008) had people learn painting styles either one artist at a time or mixed together. Mixing produced better learning, yet participants rated the one-at-a-time method as more effective, even after their own test results showed the opposite. Smooth study feels like good study.
Why do we expect to forget less than we will?
Kornell and Bjork (2009) ran 12 experiments on what they called a stability bias. People who would study a list one to four times predicted little or no learning across those study rounds, even though their real scores rose a lot. Together with earlier work on forgetting, the authors concluded that people tend to assume their memory will stay about the same over time, neither improving with more study nor fading without it.
For a student, that bias works both ways. You underrate what another round of practice would give you, and you overrate how much of today's chapter will survive the weekend.
When are your judgments accurate?
There is good news. Nelson and Dunlosky (1991) found that judgments of learning, which are usually poor predictors of later recall, became extremely accurate when people made them a little while after studying, not immediately. The likely reason is that a delayed judgment forces you to try retrieving the item. If it comes back after a gap, it probably will on the test too.
That matches what students actually do. In a survey of 177 college students, Karpicke, Butler and Roediger (2009) found that most repeatedly reread their notes or textbook, and relatively few tested themselves. Rereading keeps the material in front of you, so it never gives you the delayed, honest check. Our post on why rereading feels like learning goes further into that trap.
How to check your own learning: a calibration routine
Calibration means comparing what you predicted with what you scored. Do this for one chapter a week.
Wait before you judge. After studying a section, move on to something else for at least 20 minutes. Do not rate yourself while the page is still in your head.
Predict, then test. Write ten short questions on the section. Before answering, write how many you expect to get right.
Answer without the book, then mark your answers.
Log the gap. If you predicted 8 out of 10 and scored 5, your overconfidence for that chapter is 3. Keep a running list by chapter.
Act on the gap. Chapters with the biggest gaps get the next revision slot, ahead of chapters that only feel weak.
After three or four weeks you will know which subjects you tend to overrate. The aim is for your predictions and your scores to move closer together.
What to do when the check shows a gap
A wrong answer tells you where the gap is, not what it is. Sometimes it is a forgotten fact; often it is a step you never really understood. Fermi's AI tutor checks understanding as it goes and works out what you already know, so a gap shows up during the session instead of on the test. Pair it with writing your own questions and you have two honest checks instead of one.
Tonight, pick one section you studied today, wait an hour, then predict and test yourself on ten questions.
