Why feeling confident doesn't mean you know it

Metacognition - thinking about your own thinking - includes a specific, well-studied skill called calibration: how well your confidence in an answer matches your actual accuracy. Most people assume their gut sense of "I know this" is reasonably trustworthy. The research on study behaviour says it's biased in a very consistent direction.

The evidence problem

When you judge how well you know something, you don't have direct access to the strength of the memory trace itself - nobody does. Instead, your brain uses proxies: how fluently the material comes to mind right now, how familiar it looks, how recently you saw it. All three of those are influenced heavily by how you just studied, not by how well you'll be able to retrieve it later under different conditions (like, say, an exam room a week from now).

Re-reading inflates every one of those proxies without touching actual retrievability - which is exactly why students who re-read consistently rate themselves as better prepared than students who test themselves, then perform worse. This is the same gap covered in the illusion of competence; calibration is the more general research term for the underlying skill that's failing.

Why the bias is so consistent

It's not that people are bad at judging in general - it's that the moment you're making the judgment (right after studying), fluency is artificially high for everyone, regardless of method. The bias isn't a personality trait. It's baked into the timing of when the judgment gets made.

How to get a more honest signal

Judge your knowledge after a delay, not immediately after studying. Fluency fades with time; what's left afterward is closer to the truth.
Judge it by attempting retrieval, not by rereading. A "does this look right" check inflates confidence; a "can I produce this cold" check doesn't.
Track your actual hit rate on self-tests over time rather than trusting a single in-the-moment feeling. A system that records whether you got something right or wrong - and when - is a far better source of "do I actually know this" than your gut, because it isn't subject to the same fluency bias.

This is part of why a scheduler that tracks real answer history is worth more than a study plan based on how confident you feel about a topic. The confidence is measuring the wrong thing.