What Is Metacognition? Thinking About Your Own Thinking

You read a page, close the book, and feel sure you understood it. Then someone asks you to explain it back, and the sentences don’t come. The feeling of understanding and the fact of understanding turned out to be two different things.

That gap is what metacognition studies. Not what you think, but how well you track your own thinking while you’re doing it.

The term, and where it comes from

The psychologist John Flavell coined the modern sense of the word in a 1979 paper for American Psychologist. His model splits the monitoring of your own mind into a few working parts:

Metacognitive knowledge is what you’ve come to believe about how thinking works — in general, and about yourself in particular. Flavell defines it this way:

A simple example from the paper: a child learning that it’s easier to recall the gist of a story than its exact wording. That’s not knowledge about the story. It’s knowledge about how your own memory behaves — which is the whole category in miniature.

Metacognitive experiences are the feelings that ride alongside thinking: the sudden sense that you didn’t understand what someone just said, or the opposite, a feeling of confidence before you’ve actually checked your answer. Flavell’s point is that these feelings are real data, but they are not the same thing as being right.

A decade earlier, researchers studying memory had already split this same territory into two working parts: monitoring, which is reading your own mental state, and control, which is what you do in response to that reading (Nelson and Narens, 1990). Deciding you need to re-read a paragraph is control, acting on a monitoring judgement you’ve just made. The two usually work together so smoothly you don’t notice either one — until the monitoring is wrong and the control acts on bad information.

Why the feeling and the fact come apart

This is where metacognition stops being an abstract definition and starts being useful.

Confidence and accuracy are generated by different processes, and they can drift apart. A 2010 study published in Science separated people’s raw performance on a simple perceptual task from their ability to judge, afterward, how confident they should be in each answer. The two didn’t move together in lockstep: people with the same task performance varied considerably in how accurately they rated their own confidence, and this variation tracked measurable differences in anterior prefrontal cortex structure (Fleming et al., 2010). Knowing how well you know something turned out to be its own skill, with its own basis in the brain — not a simple readout of how well you actually did.

A separate line of research shows the same gap from a different angle. Most people, asked to rate their own objectivity against other people’s, rate themselves as less biased than average — even while correctly recognising that bias is common in people generally. Pronin, Lin and Ross (2002) named this the bias blind spot: people can describe the mechanics of bias accurately when talking about others, and still feel, from the inside, that their own judgement isn’t operating that way. The knowledge is there. The monitoring of your own case is where it breaks down.

The reverse failure is just as well documented, though how large it really is has become a live argument. Kruger and Dunning (1999) tested people on logic, grammar and humour, then asked each one to rate their own performance against the group. The people who scored worst overestimated themselves the most — not because they were careless, but because the same skills a task requires are often the skills needed to notice you’ve gotten it wrong. A later re-analysis pushes back on the size of that effect: Gignac and Zajenkowski (2020), testing self-assessed and measured intelligence in 929 adults with statistical methods built for exactly this question, found no significant distortion at the low end and a close-to-linear relationship overall — concluding the effect is real for some skills but likely smaller than widely reported. The direction of the finding survives the challenge. The size of it doesn’t, and a fair account says both.

Put these together and a pattern shows up. Metacognition isn’t one faculty that’s either switched on or off. You can be accurate about how thinking works in general, and still miss your own case specifically — because checking your own thinking is a separate, trainable act, not a free side effect of being smart, confident, or well-informed.

What this has to do with a daily reflection practice

None of this is a claim that reflection fixes the gap between feeling and fact. It doesn’t, by itself. What a daily practice can do is give the monitoring part of metacognition somewhere to happen on a regular schedule, instead of only in hindsight, after a decision has already been made and lived with.

Writing something down before acting on it, and again afterward, is a crude but real form of what Flavell called a metacognitive experience: a moment where you notice what you think you know, separately from whether it’s actually true. That noticing is the whole mechanism. It doesn’t require getting the answer right. It requires doing it often enough that checking becomes a habit rather than an event.

This is also why the hardest version of this gap usually shows up at night. The 2am overthinking loop feels like thinking, but it is mostly the absence of a stopping signal — monitoring without a decision attached to it. And the sense that your mind is “doing something” even at rest has a real basis: the brain’s default mode network stays active during exactly this kind of unanchored self-referential thought. A reflection written down is one way of giving that activity somewhere to land.

Frequently Asked Questions

Is metacognition the same as self-awareness?

They overlap but aren’t identical. Self-awareness is broader — it can include emotions, values, how you come across to others. Metacognition specifically concerns your thinking and knowing: how accurately you can judge what you understand, remember, or are likely to get right.

Can you improve your metacognition?

The research above doesn’t test an intervention, so no study here proves a fix. What it does establish is that confidence and accuracy are separable — which is itself the useful starting point: treating your own certainty as a feeling to check, not a fact to trust automatically.

Why does confident misjudgement feel so convincing?

Because the feeling of confidence and the accuracy of the judgement are generated by different processes (Fleming et al., 2010). A wrong answer can arrive with exactly the same felt certainty as a right one. The bias blind spot (Pronin, Lin and Ross, 2002) and the overestimation Kruger and Dunning (1999) measured both show the same thing from a different angle: the feeling of being right is not evidence of being right.

Short films from this research — follow on TikTok.