Consider a question on the nerve supply of a muscle, missed by two candidates. One picked the nerve that supplies the neighbouring muscle. The other picked a nerve from an entirely different plexus. Both scored zero. They have almost nothing in common.
The first candidate knows the region and has confused two adjacent structures — half an hour with a diagram fixes it. The second does not have the region at all and needs to learn it from the beginning. A percentage score cannot tell these apart, and neither can a list of topics you got wrong.
Distractors are a diagnostic instrument
In a properly written question, each wrong option corresponds to a specific misunderstanding the examiner expects to find. That is not incidental — it is what makes the item discriminate between candidates. It also means the option you chose is a far richer signal than the fact that you were wrong.
Most candidates throw this signal away. You look at the key, register that you were wrong, feel briefly annoyed, and move on. The information about which wrong idea you hold goes unrecorded, so the same misconception produces the same error again next month.
Being wrong is data. Being wrong in a consistent direction is a diagnosis.
The patterns worth looking for
- The same confusion pair, repeatedly. Two conditions or drugs that keep swapping. Almost always fixed by an explicit contrast rather than more reading on either one.
- Stem-reading failures. You chose an option that is true in general but not for the patient described. The gap is not knowledge; it is that you decided the topic before finishing the stem.
- Rules pushed too far. A principle you know, applied where it does not hold. You need the exception, not the rule.
- Whole-region absence. Scattered options with no pattern at all, which usually means you were guessing and the topic is genuinely missing.
Doing this by hand
You can extract most of the value with a notebook and a rule: every time you miss a question, write one line naming the idea you were actually using. Not the topic — the idea. "I thought the ostium was at the floor." After a hundred questions the repeated lines are your real syllabus, and they will be far fewer and far more specific than the topic list you would have guessed at.
The reason people do not do this is that it is tedious in exactly the moment you least want more work. That is a fair reason, and it is why the job suits software.
Misconceptions are sticky in a specific way
A gap is easy to fill. A wrong belief has to be displaced, and the old version does not politely vacate — for a while both are present, and under time pressure the older, more practised one tends to win. This is why a confident error needs several encounters rather than one correction, and why reading the right answer once feels sufficient and is not.
Write the correction as a contrast
"The ostium is high on the medial wall" is a fact. "Not at the floor, as I assumed — high on the medial wall, which is why drainage is uphill" is a correction. The second names the wrong version explicitly, and naming it is what lets you notice yourself reaching for it again.
A correction that does not name the mistake it replaces leaves the mistake intact alongside it.
The pattern is usually smaller than you fear
Candidates who start recording their errors this way are generally surprised by how few distinct misconceptions account for most of their losses. Twenty or thirty recurring ideas is common, spread across what felt like hundreds of unrelated mistakes. That is encouraging: a list of thirty specific things is a tractable piece of work in a way that "get better at pathology" never is.
What Recallab records
Every attempt stores the option you chose, not just whether you were right. That is what makes Second Chance and Blind Spots possible — both live — and what the Why I got it wrong feature reads when it explains your specific error rather than the general one. That last feature is in pilot, but the underlying record has been kept from the start, so the history will be there when it opens up.