Metacognition: how to know whether you really understand

Metacognition is the ability to monitor and steer your own learning: knowing what you have mastered, what you have not, and what to do about it. In practice, it means not trusting the feeling that you already know something and checking it instead: explain without notes, apply the idea to a new case, or predict your score before you grade yourself.

The feeling of mastery usually comes from familiarity. After reading a chapter twice, the sentences look familiar, and that feels a lot like knowing them. The problem shows up when the exam asks you to produce the answer rather than recognize it among options or in a text in front of you.

Bjork, Dunlosky and Kornell (2013) reviewed research on how students manage their own learning. They describe how many people confuse in-the-moment fluency, the material feeling easy to read, with lasting learning. As a result, students favor strategies that feel productive but pay off less over time, such as rereading or cramming a whole topic into one session.

Kornell and Bjork (2008) showed this with an experiment on painters' styles. Participants got better at recognizing new paintings when examples from different artists were interleaved, yet most of them believed they had learned as much or more when each painter's works were shown together. Their impression and their results pointed in opposite directions.

Each one takes a few minutes and gives you a concrete result you can compare with the source. Pick the one that best fits the kind of exam you will take.

A statistics student writes down three questions a day, his confidence from 1 to 5, and whether he got each one right. Monday: arithmetic mean, confidence 5, correct. Median of an even number of values, confidence 4, wrong. Standard deviation, confidence 2, partly correct. A pattern appears quickly: his confidence is high for procedures he has seen many times, even when he does not always carry them out correctly.

For the median mistake he reviews the procedure: with the values 3, 5, 8 and 10, the median is the mean of the two middle values, (5 + 8) / 2 = 6.5. On Thursday he tries again with different numbers and gets it right. The log takes him no more than five minutes a day and shows where his intuition needs a check. The example is our own.

Metacognition only helps if it changes your next session. When a check shows you knew less than you thought, reading the correct answer is not enough: try again later with a different question. If the mistake keeps coming back, switch activities, for example from flashcards to full problems or to explaining the concept in writing.

It also works the other way. If you have answered correctly for several days and your confidence matches your results, you can space that topic further apart and spend the time elsewhere. That adjustment avoids two common traps: endlessly reviewing what you already know and never touching what only felt familiar.

Go through these signs at the end of each week. If two or more sound familiar, add closed-book checks before you count the topic as done.

In Lumia AI, after answering each flashcard you rate it Again, Hard, Good or Easy, and the app calculates the next review from that rating. That is a metacognitive exercise in itself: if you mark Easy an answer you only recognized after flipping the card, the card will be spaced further out than you can handle.

Custom quizzes let you choose the number of questions, the difficulty, and question types such as Multiple Choice or Fill in the Blank. Before you start, write down how many you expect to get right and compare it with your score. Lumia does not grade your open answers or measure your understanding: the useful information comes from comparing your prediction with what actually happened.