In a Yale psychology study published in 2002, 16 graduate students were handed a list of ordinary objects and asked how well they understood each one. A zipper. A flush toilet. A cylinder lock. A quartz watch. They rated themselves from 1 to 7, and they were fairly generous, because these were clever people who had zipped their coats every day of their lives. What the psychologists did next is the best place we know to start with metacognition techniques, the small habits of thinking about your thinking that adults can actually use.
Leonid Rozenblit and Frank Keil asked the students to write out, step by step, how each device worked. Then they asked for the rating again.
It fell. In the paper, the researchers note that nearly all participants showed drops in what they believed they knew, once they had tried to explain it. They gave the effect a name that has stuck: the illusion of explanatory depth. A detail at the end of the study is more unsettling than the drop itself. After reading an expert’s explanation, the students’ sense of their own understanding climbed right back up. The feeling of knowing is quick to return, and it returns whether or not anything was learned.
Why metacognition techniques matter more after school
The developmental psychologist John Flavell gave the field its name in a 1979 paper on metacognition and cognitive monitoring, built on studies of children who were asked to study a set of items until they were sure they could remember them. The younger ones often said they were ready and then could not recall them. Most of what has been written since is aimed at classrooms, and the advice tends to assume a test is coming.
Adults mostly don’t get the test. Nobody grades the explanation you gave your team about why the launch slipped, or the theory about interest rates you shared over dinner. The errors don’t come back marked in red. They just quietly fail to matter, or matter months later in a way you never connect to the original belief.
Compare that with weather forecasters. When a forecaster says there is a 30% chance of rain, it gets checked by the next evening, every day, for years. In a 1977 study of US forecasts, Allan Murphy and Robert Winkler found that forecasters could produce probability forecasts in a reliable way, meaning their stated chances lined up well with what actually happened. The rest of us make forecasts constantly. Almost none of us ever go back and score them.
So the work for an adult is mostly about building the feedback that school used to supply. The four techniques below each do that in a different way, and none of them needs more than a notebook.
Write out how it works before you trust the feeling
The simplest technique comes straight from the zipper study. Before you defend an idea, sit down and write the causal chain: this happens, which causes that, which leads to this. No bullet points that say “improves engagement.” Actual mechanism, the way you would explain it to a curious 12-year-old who keeps asking why.
Rozenblit and Keil found something useful about where the illusion lives. When they asked people about facts, like the capitals of countries, the drop in confidence was much smaller. When they asked about procedures, the steps of doing a familiar task, there was no drop at all. The overconfidence clustered around explanations, around how one thing causes another. That tells you where to point this habit: at your theories about why the economy is doing what it is doing, why a colleague behaves the way they do, how a medication works, why your sleep is bad. Those are the beliefs most likely to feel solid and turn out hollow.
The feeling of knowing is quick to return, and it returns whether or not anything was learned.
You will often get two sentences in and stall. That stall is the information. It marks the exact spot where your understanding was borrowed from a headline or a confident friend, and it is far cheaper to find it alone with a pen than in the middle of a meeting.
Judge what you learned tomorrow, not tonight
Right after you finish a book chapter or a training session, it feels like you have it. The material is still warm. That warmth is a poor guide. In a 1991 paper in Psychological Science, Thomas Nelson and John Dunlosky showed that people’s judgments of how well they had learned something were much more accurate at predicting later recall when made after a short delay instead of immediately. Researchers now call it the delayed-JOL effect, for judgment of learning.
The practical version is almost embarrassingly small. When you finish something you want to keep, don’t rate yourself then. The next morning, before you look at any notes, ask what you could explain from it. Whatever comes back is a much better estimate of what you actually kept. We wrote about why this kind of recall also strengthens memory in how to remember what you read. Recalling something from memory turns out to strengthen it, so the morning check pays twice.
Metacognitive strategies borrowed from forecasters
In the early 2010s, a US intelligence research agency ran a geopolitical forecasting tournament, and one of the competing teams, the Good Judgment Project, recruited ordinary volunteers to predict world events. Barbara Mellers and her colleagues later analyzed roughly 150,000 forecasts from 743 participants on 199 questions over two years. Among the things that predicted accuracy were open-mindedness, training in probabilistic reasoning, time spent deliberating, and how often people updated their beliefs as news came in.
You can borrow the core of this without joining a tournament. Keep a short page of predictions. Write the claim, a percentage and a date: “The new hire will still be here in a year, 80%.” “This renovation will run over budget, 70%.” Then actually go back and check. After a few months a pattern shows up. If the things you call 90% certain come true about two times in three, you have learned something about yourself that no amount of introspection would have told you.
Most people who try this find they are overconfident in exactly the areas where they feel most expert. That fits the research on self-assessment we looked at in the real story behind the Dunning-Kruger effect, and a prediction page is one of the few honest ways to see it in your own record instead of someone else’s.
Imagine it already went wrong
The psychologist Gary Klein described a method he called the premortem in Harvard Business Review in 2007. Before a project starts, the team imagines it is a year later and the project has failed. Each person writes down why. Klein drew on earlier research by Deborah Mitchell, Jay Russo and Nancy Pennington, who found that imagining an outcome had already happened increased the number of reasons people came up with for it by about 30%.
It works on single decisions too. Before you take the job or sign the lease, give yourself five minutes in the future where it went badly. Write the story of how. The point is to catch the reasons you already half knew and had been politely ignoring, which tend to show up fast once you grant yourself permission to assume failure.
How to improve metacognition without turning it into homework
None of these techniques works as a one-off. Their value comes from repetition, the way a forecaster gets calibrated by checking every day rather than once a year. The good news is that each one can be very small: two sentences of mechanism, one morning-after recall, one line on a prediction page, five minutes of imagined failure before a big decision.
It also helps to have a regular moment where the question is simply “do I actually believe this?” We argued in why your phone runs on System 1 that most of a day is spent in fast, automatic thinking, and that a slower moment has to be made on purpose. A short daily thinking practice is one way to make it. Meeting one new idea and deciding whether it holds up is a tiny, repeatable act of thinking about your own thinking.
Tomorrow morning, when you zip your coat, ask yourself how the teeth actually lock together. Then stand there a moment, coat half done, and notice how much you were sure of a second ago.