Sort your errors by how sure you were. The ones where you knew you were guessing are almost harmless — you already know that topic is weak, and you will come back to it. The ones where you were certain and wrong are a different animal entirely.
A confident error has no warning attached. Nothing in your experience of answering it flagged a problem. You will not put it on a revision list, because as far as you are concerned it is settled. It sits there, feeling like knowledge, until the examination converts it into a lost mark and a negative one on top.
Why confident errors are stable
Most wrong beliefs in medicine are not random noise. They are a coherent, plausible model that happens to be wrong — a mechanism you inverted, a relationship you learned backwards, an association from a source that was out of date. A coherent wrong model produces consistent wrong answers, and consistency feels exactly like competence from the inside.
Uncertainty is self-correcting. Confidence is not.
Making the invisible visible
You cannot find these by reflection, because reflection uses the same broken model that produced the error. The only reliable route is to record your confidence at the moment you answer, before you see the outcome, and then look at the intersection of high confidence and wrong.
Recording confidence takes a second per question and people resist it, because it makes the practice session feel slower and it is faintly uncomfortable to commit. That discomfort is the feature: it forces you to notice what you actually believe rather than what you can reconstruct afterwards.
What to do with the list
- Treat every entry as unlearned. Do not skim the explanation and conclude you knew it really. You did not.
- Find the source of the belief. Confident errors usually trace to a specific note, lecture or mnemonic. Anything else you took from the same source deserves a look.
- Write the correction as a contrast. "Not X, but Y, because Z." Overwriting a wrong model needs the wrong version named, or the old one survives alongside.
- Schedule it deliberately. These need more repetitions than something you simply never knew, because you are competing with an existing belief.
Where confident errors come from
They are rarely invented. Most trace to a specific origin, and the origin is worth finding because it is unlikely to have produced only one.
- An out-of-date source. Guidance changes; notes do not. Anything learned from the same set of notes deserves checking.
- A mnemonic that encoded it backwards. Mnemonics are retrieval aids and they are perfectly capable of retrieving the wrong thing reliably.
- A confidently-taught error. Someone said it with authority and it was never questioned.
- An over-generalised rule. Correct in the case you learned it in, extended without noticing to cases where it fails.
The exam cost is doubled
Worth stating explicitly, because it changes the priority. A confident error is not a zero — under +4/−1 it is a wrong answer, so it costs a mark. And unlike a topic you know you do not know, you will not leave it blank, because you have no reason to hesitate. It is the one category where certainty actively works against you.
Check the neighbours
Having found a confident error, spend ten minutes on the material immediately around it rather than moving on satisfied. Wrong beliefs cluster: the source that gave you one usually gave you several, and they will all feel equally settled.
In Recallab
You rate your confidence before submitting, and the False Certainty app deals you exactly the intersection: questions you were sure about and got wrong. Its sibling, Lucky Guesses, does the reverse. Both are live for every account, and the confidence rating also feeds the Due Today scheduler, so a confident error comes back sooner than an ordinary one.