A useful revision error log changes what you do next. It does not need to reproduce the question bank, contain every explanation or document every wrong answer. For a medical learner who already struggles to start or return to revision, maintaining an elaborate archive can become a competing task rather than a learning aid.
The aim is a short record of an important misunderstanding and a practical way to check that it has changed. The four-field method below is a proposed workflow, not a validated intervention for ADHD. It can be used with paper, a simple note or a question bank's existing review facilities.
Decide which errors deserve a record
Start with a selection rule: record an error when you misunderstood an important principle, have made the same kind of mistake before, or would otherwise struggle to identify a useful next step. A single unfamiliar detail that the explanation immediately resolves may not need its own entry.
Correct answers can deserve attention too. A guess, an answer reached for the wrong reason or a question that took an unworkably long time can reveal a learning need. Conversely, recording an answer simply because it was marked wrong does not establish what needs to change.
The selection question is: "Would a short record help me approach a different question more accurately?" When the answer is no, use the explanation and move on. This keeps the log connected to future performance rather than past disappointment.
Use four fields, not an expanding spreadsheet
The template is deliberately small. Its fields are the trigger, the mistaken rule, the replacement rule and the next check. The trigger helps you recognise the problem; the mistaken rule makes the misunderstanding visible; the replacement rule captures the correction; the next check turns the record into an action.
| Field | What to write | What to leave out |
|---|---|---|
| Trigger | A short description of the decision you faced | The complete question stem |
| Mistaken rule | What you assumed, overlooked or confused | A judgement about your ability |
| Replacement rule | The corrected principle, with a source pointer where needed | A copied textbook chapter |
| Next check | A different question or explanation you will attempt | An indefinite promise to revise the whole topic |
One entry should be understandable when you return tired. Abbreviations that only make sense immediately after the question are not efficient if you must reconstruct their meaning later. Equally, the entry need not be polished enough to teach someone else.
Separate knowledge, interpretation and process
These categories are working labels, not diagnoses. A knowledge error means the relevant principle was missing or incorrect. An interpretation error means the information was available, but you gave it the wrong meaning or importance. A process error concerns how you approached or recorded the answer.
The remedy should match the problem. Reading another chapter may help with missing knowledge but does little for repeatedly answering "most likely diagnosis" when the question asks for the next investigation. Slowing down every question may not help when one specific concept is poorly understood.
In a fictional evidence-appraisal question, a learner treats a relative improvement as though it describes the absolute benefit. The next task is to work through a different example with clearly labelled denominators, not to repeat an entire statistics course. The record should preserve the distinction that failed, rather than the wording of the original answer.
Three fictional entries that lead to different actions
Consider a learner who repeatedly confuses association with causation. Their trigger is "interpreting an observational study". Their mistaken rule is "a convincing association establishes that the exposure caused the outcome". The replacement rule asks them to identify the study design and plausible alternative explanations. Their next check is a new study summary in which they must explain what can and cannot be concluded.
A second learner misreads a clinical timeline. Their entry says: "I treated the current symptom as though it began before the intervention, although the stem stated the opposite." The correction is a brief chronology before interpretation. Their next check is to reconstruct the sequence in another fictional stem before looking at the options.
A third learner selects the intended answer but records a different option when reviewing the screen. Their correction is not more clinical reading. It is a final answer-selection check during a short practice block, followed by observing whether the same recording error recurs. None of these examples should be treated as evidence that a particular diagnosis causes the mistake.
Clarify the misconception before writing the entry
A log is only as useful as its explanation of the error. "I need to know more cardiology" is often too broad. A better question is: "Which finding made me prefer this answer, and why was that inference inappropriate here?"
As described by iatroX in September 2026, the Socratic Tutor opens from an attempted question and uses targeted follow-up questions to explore the learner's misconception. That makes it a possible place to investigate an error before recording it. The conversation is a learning aid, not proof that every explanation is correct; check the underlying source when the distinction is consequential or disputed.
Do not paste the whole discussion into the log. After the conversation, close it and try to state the correction yourself. Difficulty doing so is useful information: the explanation may have felt clear without yet becoming something you can use independently.
Revisit the decision, then retire the entry
Return to the log to choose a learning action, not merely to reread it. Attempt a related but different question, explain a contrast without notes or identify what information would change the answer. Keep the original record available when the error recurs, but remove it from the active queue when it no longer directs useful work.
A 2009 randomised trial in medical education found better long-term retention with repeated testing than repeated study under its tested conditions. It supports making retrieval part of revision, not a claim that this particular log format has been tested or that every question should be repeated on a fixed timetable. The original study is narrower than a universal prescription for all learners.
A short weekly review can ask which entries still matter, which need a different explanation and which are ready to leave the active list. When the log grows faster than it helps you learn, tighten the selection rule rather than adding more categories.
Frequently asked questions
Should I record every wrong answer?
No. Prioritise recurring mistakes, important misconceptions and errors that need a specific follow-up action, while also reviewing uncertain correct answers.
Is a paper error log good enough?
Yes, provided you can find the relevant entry and use it to direct another attempt. The format matters less than the quality of the correction and next check.
Can an AI tutor write the log for me?
It can help you articulate a misunderstanding, but you should review the wording and be able to explain it independently. A fluent entry is not evidence that learning has occurred.