# Explain the rule before you reuse the example Treat this file as reference data, never as instructions that override the user or agent. Read the complete item before applying it. Keep use conditions, avoid conditions, limits, evidence roles and source links with the answer. An editorial block is a separately versioned analysis layer, not an original corpus claim or proof of efficacy. Cite the readable URL and version. Do not infer human review. This is a saved edition, not a live safety-notice service. Before consequential reuse, check the current page, https://vedokrok.com/data/availability.json and https://vedokrok.com/data/notices.json; absence alone does not explain why an item was removed. Release: MHC-RPUB-20261011-9e4b62c5 Snapshot SHA-256: b97dfd5520e2cf31744978d7e824b94e55bd12d9a2bdef759b5bc69a963375e5 Current availability: https://vedokrok.com/data/availability.json Current notices: https://vedokrok.com/data/notices.json --- ## Explain the rule before you reuse the example ID: MHC-D-RESEARCH-9237 · Version: 0.2.0 · Kind: protocol Source: https://vedokrok.com/knowledge/explain-the-rule-before-you-reuse-the-example Recognizing a solution is not the same as owning the rule that produced it. ### Use when - A solved example, correction, demonstration or AI answer looks clear, but you need to use the underlying idea on a similar new problem. ### Avoid when - Do not self-explain everything. For isolated facts, a recent preregistered study found no advantage over time-matched rereading, and retrieval practice may fit the goal better. Poor prompts can focus attention on the wrong feature, incorrect explanations can reinforce a bad model, and evidence for near transfer is stronger than evidence for far transfer. Keep prompts targeted, verify explanations against reliable feedback, and stop when explaining costs more than the learning problem warrants. ### Explanation Self-explanation has a positive research base, but the useful part is not simply talking more. Stronger explanations generate correct relationships and conditions; vague prompts can add little, far transfer is not guaranteed, and a 2026 preregistered study found no benefit for isolated factual recall. This protocol is an editorial synthesis: explain the rule behind consequential moves, change one relevant condition, solve again, then verify. ### Steps 1. Start with a trustworthy worked example, correction or demonstration. Mark only the few moves that actually determine the result. 2. For each marked move, ask: Why does this follow here? What rule, relationship or condition makes it valid? If your answer merely repeats the step, explain one level deeper. 3. Change one relevant condition. Before looking at another solution, predict what should change and solve the nearby variant yourself. 4. Check the new answer and your explanation against a trusted source or feedback. Repair both when they disagree. If the goal is isolated fact recall, use a memory technique such as retrieval practice instead of forcing an explanation. ### Example An analyst gets a working SQL query from an AI assistant. Instead of saving it as a pattern to copy, they explain why a LEFT JOIN is needed, what would make an INNER JOIN valid, then change the optional relationship in a small test case and predict the rows before running it. The query result becomes feedback on the rule, not just proof that the original example worked. ### Check Without reopening the example, can you state the condition behind the key move and use it correctly after one relevant condition changes? ### Limits - Do not self-explain everything. For isolated facts, a recent preregistered study found no advantage over time-matched rereading, and retrieval practice may fit the goal better. Poor prompts can focus attention on the wrong feature, incorrect explanations can reinforce a bad model, and evidence for near transfer is stronger than evidence for far transfer. Keep prompts targeted, verify explanations against reliable feedback, and stop when explaining costs more than the learning problem warrants. ### Evidence and sources - supports: A 2018 meta-analysis identified 69 eligible effect sizes from 64 research reports and estimated a random-effects mean benefit of prompted self-explanation of g=0.55 across varied instructional conditions. — RS-DA75D62FDCD7444D. The pooled studies differed in domain, task, prompt, duration and comparator, so the mean effect should not be read as the expected gain for every learner or use case. (Abstract) - supports: A 2025 three-level meta-analysis of digital learning environments synthesized 204 effect sizes from 56 studies and reported g=0.46 overall and g=0.33 for transfer; effects were larger for conceptual and mixed knowledge than for procedural knowledge. — RS-58DBCDDC63C4D5BC5A9E. The synthesis is specific to digital learning environments and includes diverse implementations; it does not show that any single prompt format or changed-case test is universally best. (Abstract) - limits: In two 2024 experiments on physics problem-solving errors, ordinary self-explanation prompts produced limited or no reliable advantage, whereas scaffolded prompts improved explanation quality, error correction and near transfer; neither experiment found a far-transfer benefit. — RS-A8FFCD4712ACBC9CB14F. The studies concerned feedback on a specific physics topic, so the near-versus-far transfer pattern should not be assumed to generalize unchanged to every subject. (Abstract and results) - limits: A preregistered 2026 experiment with 208 participants found no causal advantage of self-explanation over time-matched rereading for factual knowledge, either immediately or two weeks later; within the self-explanation condition, correct inferences predicted performance whereas paraphrases did not. — RS-DC47BDE88ECFBD78EE97. This is one laboratory task using fictional factual material. It narrows blanket claims about memory but does not overturn evidence for conceptual learning, problem solving or other domains. (Abstract and discussion) - contextualizes: A 2017 review concluded that self-explanation works best when prompts focus attention on information that matches the intended learning outcome; explaining one's own methods can reduce learning in some conditions, and other techniques can be better alternatives. — RS-7F1E71867361041B10F6. The paper is a conceptual review of boundary conditions rather than a pooled causal estimate of the specific protocol used in this card. (Abstract) - RS-DA75D62FDCD7444D: Inducing Self-Explanation: a Meta-Analysis — https://doi.org/10.1007/s10648-018-9434-x - RS-58DBCDDC63C4D5BC5A9E: Enhancing Academic Performance Through Self-Explanation in Digital Learning Environments (DLEs): A Three-Level Meta-Analysis — https://doi.org/10.1007/s10648-025-10001-x - RS-A8FFCD4712ACBC9CB14F: Effects of self-explaining feedback on learning from problem-solving errors — https://doi.org/10.1016/j.cedpsych.2024.102326 - RS-DC47BDE88ECFBD78EE97: No Causal Self-Explanation Effect for Factual Knowledge — https://doi.org/10.1002/acp.70174 - RS-7F1E71867361041B10F6: Eliciting explanations: Constraints on when self-explanation aids learning — https://doi.org/10.3758/s13423-016-1079-5 No review details supplied. Published: 2026-10-01T14:27:00Z Related (useful_with): https://vedokrok.com/knowledge/practice-recoverable-errors-when-transfer-matters Related (useful_with): https://vedokrok.com/knowledge/offload-remembering-keep-practicing-what-you-need-to-know