# 24-hour cooling-off before sentence passes > Depth-first rabbithole dossier for 24-hour cooling-off before sentence passes; source-anchored research pack. 6 facets · 44 facts · page: https://llms-explorer.com/tree/24-hour-cooling-off-before-sentence-passes/ ## Structure and components - - **C21.** Tom Stafford (Sheffield) says authors read what they meant to write, so "hte" gets read as "the". — https://kirkville.com/whats-up-with-that-why-its-so-hard-to-catch-your-own-typos-wired/ (mirror of http://www.wired.com/2014/08/wuwt-typos/) [M,P] - **C22.** Stafford's explanation: writing is a high-level task, so the brain generalizes the low-level parts (letters, words) to focus on meaning. — https://www.mentalfloss.com/article/633063/reason-its-so-hard-spot-your-own-typos [M,P] - **C23.** UNC: when material is too familiar, "you tend to skip over a lot of errors." — UNC URL [E] - — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/rabbithole-synthesis.md#c-proposed-mechanism-1-overfamiliarity-causes-top-down-reading` - 4. The core claim is that the author knows what the text is *meant* to say, and that knowledge fills gaps while they read. Stafford (University of Sheffield), quoted in WIRED 2014: when rereading your own work you already know what you wanted to say, so you miss errors such as "hte" because your brain supplies "the." https://kirkville.com/whats-up-with-that-why-its-so-hard-to-catch-your-own-typos-wired/ (republishes http://www.wired.com/2014/08/wuwt-typos/) 5. Stafford explains why this happens: writing is a high-level, meaning-focused task, so the brain generalizes low-level parts (letters in — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/mechanism.md#proposed-mechanism-overfamiliarity-drives-top-down-reading` ## How it works - In scope: the practice of letting one's own draft rest for about 24 hours before doing sentence-level editing or proofreading passes. The report covers the mechanism the practice assumes, the conditions under which it works or fails, the evidence against it, and the disagreements about it. Out of scope: other revision techniques (reading aloud, reading backwards, changing the font), structural or global revision, peer review, and the parent domain "Editing and Revision". The report mentions these only where a cited source uses them to explain the delay. — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/edge-cases.md#scope` - 1. Writing-center guidance states the rule as "Ideally, give yourself 24 hours before editing. At the very least, give yourself an hour break before proofreading." — https://bowvalleycollege.libguides.com/writingsupport/proofreading 2. Other writing-center guidance names no fixed interval. It says "Put the paper aside for a few hours, days, or weeks." — https://writingcenter.unc.edu/tips-and-tools/editing-and-proofreading/ 3. The same UNC page gives the mechanism the practice relies on: when material is too familiar, "you tend to skip over a lot of errors." — https://writingcenter.unc.edu/tips — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/edge-cases.md#what-the-practice-asserts-and-where-the-number-comes-from` - In scope: the practice of setting a finished draft aside for about a day before the sentence-level pass (line edit or proofread). That covers how people use it, the claimed mechanism, the evidence for it, how it has been tested, and what it means in practice. Out of scope: other de-familiarization tricks as practices in their own right, peer review, the wider revision process, and incubation in creative problem solving. Those appear only where they explain or challenge the cooling-off claim. — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/practice.md#scope` - 1. University writing guides state the rule directly: "leave your final draft for at least 24 hours before proofreading." https://www.wgtn.ac.nz/student-learning/studyhub/academic-writing/essay-writing-process/step-5-revising-drafts,-editing-and-proofreading 2. The same guide gives the reason: "When you are too familiar with your writing, you might not see the mistakes." https://www.wgtn.ac.nz/student-learning/studyhub/academic-writing/essay-writing-process/step-5-revising-drafts,-editing-and-proofreading 3. Writing-center guidance treats 24 hours as the ideal and an hour as the minimum fallba — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/practice.md#operational-use` - **Inputs:** four independent reports from ~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/: - `mechanism.md` [M] - `history.md` [H] - `edge-cases.md` [E] - `practice.md` [P] — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/rabbithole-synthesis.md` - 1. Should I save this dossier as a file in ~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/? I couldn't list that folder because shell access was denied, so I didn't know the pipeline's naming convention. For now it exists only in this response. — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/rabbithole-synthesis.md#needs-input` - 29. In a study of 60 children in grades 3–5, postponing revision until after writing led to more revisions than revising during writing. — https://pubmed.ncbi.nlm.nih.gov/11307706/ 30. Surface revisions outnumbered meaning revisions, but the difference was significant only for younger children or when revision happened during writing. — https://pubmed.ncbi.nlm.nih.gov/11307706/ 31. The abstract does not say how long the delay was. The study therefore separates revising during writing from revising after writing, not a 24-hour delay from a shorter one. — https://pubmed.ncbi.nlm.nih.gov/11307706 — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/edge-cases.md#delayed-revision-in-children-supporting-with-limits` ## Measurements and reference values - 5. Daneman & Stainton (1993) found that students who proofread their own essay right after writing it detected about 20% fewer errors than students who proofread a familiar essay written by someone else. (Secondary report of the 1993 paper.) — https://englelab.gatech.edu/articles/2022/Burgoyne%20et%20al.%20(2022)%20Revisiting%20the%20self-generation%20effect%20in%20proofreading.pdf 6. Daneman & Stainton attributed the deficit to "overfamiliarity", which pushes the reader into top-down, "expectancy-driven" processing. — same Burgoyne et al. (2022) PDF; original: https://link.springer.com/articl — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/edge-cases.md#the-key-study-behind-it-and-its-limits` - 16. Daneman & Stainton (1993) found that people detected fewer errors in essays they had written 20 minutes earlier than in unfamiliar essays by others. They also detected more errors in familiar essays by others than in unfamiliar ones. *Reading and Writing* 5(3):297–313. https://eric.ed.gov/?id=EJ498842 17. Daneman & Stainton blamed the self-generation deficit on over-familiarity, which pushes the reader into top-down, "expectancy-driven" reading. https://eric.ed.gov/?id=EJ498842 18. In Daneman & Stainton's second experiment, writers proofread their own essay two weeks later. At that point t — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/history.md#empirical-evidence-supporting` - | Pass | Focus | New claims | Total | New-info rate | |---|---|---|---|---| | 0 | Broad draft: Horace, King, writing-center rule, Daneman & Stainton | 12 | 12 | 100% | | 1 | Mechanism and disconfirmation: Burgoyne 2022, Pilotti 2006/2009, sample sizes | 9 | 21 | 43% | | 2 | Primary-text check: Horace Latin, Quintilian's limits, Chanquoy abstract | 5 | 26 | 19% | | 3 | Origin of the "24 hours" figure | 1 | 27 | 4% | — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/history.md#pass-log` ## Problems, failure modes and limitations - - The 24-hour figure is a practitioner heuristic. No located adult study tests 24 hours specifically. The tested delays are the next day (children, positive), one week (adults, null) and two weeks (adults, positive, small n). Sources: claims 10, 15, 17. - The strongest evidence for a delay concerns revision depth (Chanquoy, claims 13–14). The evidence for catching typos is weaker (claims 16–18). A cooling-off period is therefore better justified before a sentence-level *rewriting* pass than before a mechanical proofread. - If the writer cannot wait, the claimed mechanism (overfamiliarity) is a — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/practice.md#concrete-implications` - - **In scope:** setting your own finished draft aside, conventionally for about 24 hours, before sentence-level editing or proofreading. That covers its parts, the proposed mechanisms, its history, the tests of it, edge cases, failure modes and disagreements. - **Out of scope:** other revision techniques (reading aloud, reverse reading, font changes as practices in their own right), structural revision, peer review, and the parent topic "Editing and Revision". These come up only where a source uses them to explain the delay. - **Boundary breach:** none. No pass drifted into sibling concepts. — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/rabbithole-synthesis.md#scope` - - **C1.** The practice has three steps: finish the draft, leave it alone, then come back as a reader. UNC: "Put the paper aside for a few hours, days, or weeks … Clear your head of what you've written so you can take a fresh look at the paper." — https://writingcenter.unc.edu/tips-and-tools/editing-and-proofreading/ [M,H,E] - **C2.** The delay comes before the surface pass. UNC treats editing and proofreading as separate stages, puts proofreading last, and advises checking for "only one kind of error at a time." — same UNC URL [M,H] - **C3.** UNC names no fixed interval ("a few hours, days, or — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/rabbithole-synthesis.md#a-definition-and-parts` - - **C39.** Sio & Ormerod (2009) meta-analysed 117 studies (3,606 participants). Setting a problem aside had a positive effect: mean d ≈ 0.29, 95% CI 0.21–0.39. — https://eric.ed.gov/?id=EJ827110 · https://gwern.net/doc/psychology/writing/2009-sio.pdf · https://psycnet.apa.org/fulltext/2008-18777-009.html [M,E,P] - **C40.** The effect was smaller when the break was filled with demanding tasks. — gwern PDF [M] - **C41.** For linguistic problems, a break with a low-demand task beat pure rest. — gwern PDF [E] - **C42.** The length of the break did not predict effect size (coefficient < .001). — gw — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/rabbithole-synthesis.md#e-indirect-evidence-from-incubation-research-problem-solving-not-text` - - **Does familiarity with your own text hurt or help error detection?** Daneman & Stainton (1993) say it hurts, and that the deficit fades after two weeks. Pilotti & Chodorow (2009) say familiarity helps detection of errors that others introduced. Burgoyne et al. (2022) found no significant effect. These sources are not reconciled here. https://eric.ed.gov/?id=EJ498842 · https://eric.ed.gov/?id=EJ826333 · https://link.springer.com/article/10.1007/s00426-022-01699-3 - **Is 24 hours the right interval?** Guidance ranges from one hour (Bow Valley, minimum) to 24 hours, six weeks (King), and nine — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/history.md#unresolved-disagreements` - 23. Sio & Ormerod's (2009) meta-analysis found significant incubation effects: a break improved later problem solving across creative and insight problem types. https://psycnet.apa.org/fulltext/2008-18777-009.html 24. For visual insight problems, a longer preparation phase before the break produced a larger incubation effect. A break helps more after real effort than before it. https://pmc.ncbi.nlm.nih.gov/articles/PMC6994426/ 25. Incubation is about restructuring a problem. It supports the idea that a delay helps with meaning-level choices, such as how to recast a sentence. It gives no eviden — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/practice.md#mechanism-candidate-from-incubation-research-boundary-evidence-only` - - **C12.** Horace, *Ars Poetica* 386–390 (c. 19 BC), tells poets to hold a work back until the ninth year ("nonumque prematur in annum"). — https://en.wikisource.org/wiki/Page:Hoyt's_New_Cyclopedia_Of_Practical_Quotations_(1922).djvu/641 [H] - **C13.** Horace's reason: you can delete what you have not published, but "nescit vox missa reverti" (a word once sent out cannot come back). — [H] cites a latin.it URL that contains a literal "…" and looks truncated, so it is **not used as an anchor**. The line is at the Perseus *Ars Poetica* record in C15. - **C14.** Scholars disagree on what the "nine — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/rabbithole-synthesis.md#b-history` - - **X1. Where "the following day" comes from (Chanquoy):** - [M] claim 10 quotes "the following day" and cites the Europe PMC abstract. - [H] claim 21 says the abstract does not give the interval; the phrase comes from a ResearchGate secondary summary. - [E] claim 31 says the abstract does not give the delay length. - [P] claim 15 attributes it to an academia.edu snippet. - **Status:** unresolved. [M]'s attribution to the abstract conflicts with two reports that read the same abstract. - **X2. What Daneman & Stainton compared against:** - The ERIC abstract [M,H,P] says own essays scored worse — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/rabbithole-synthesis.md#contradictions-between-the-reports-kept-side-by-side` - 1. **Invented precision:** "24 hours" has never been tested (C10, C67). 2. **Wrong premise:** the self-generation deficit may not exist (C50–C53, C60–C61). 3. **Meaning errors survive the break:** memory of what you meant outlasts memory of what you wrote (C31–C32, an inference). 4. **A single-sitting draft gains little:** the harmful kind of familiarity may take days of rework to build (C64, C43). 5. **Time without effort:** a rushed pass after 24 hours still misses errors, because effort predicts detection (C58). 6. **Blind spots the break can't fix:** wrong-word and function-word errors (C6 — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/rabbithole-synthesis.md#failure-modes` - Caveats: - Claim 15 (the next-day delay and remaining-error result) rests on a secondary snippet, because the full text was blocked. - Claim 20 is the Pilotti & Chodorow result as reported by Burgoyne et al. The primary record was located but not read in full. - No primary study tests an exactly 24-hour delay in adults. — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/practice.md#quality-gate` ## Comparisons and alternatives - 16. **Failed replication.** Burgoyne, Saba-Sadiya, Harris, Becker, Brascamp & Hambrick (2022, *Psychological Research*) ran a near-replication of Daneman & Stainton with eye tracking (n = 63 in Session 1). It "failed to reveal a self-generation effect." https://doi.org/10.1007/s00426-022-01699-3 17. In that replication, people who proofread their own just-written essay detected *more* errors than people proofreading someone else's (88.6% vs 83.3%, d = 0.49, p = .059). For word errors (wrong real words) the self-generated group did significantly better (89.4% vs 80.0%, d = 0.61, p = .018). This — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/mechanism.md#limits-and-disconfirming-evidence` - 23. A meta-analysis of 117 incubation studies (3,606 participants) found a small-to-medium positive effect of setting a problem aside (weighted mean d = 0.29, 95% CI 0.21–0.39). — https://gwern.net/doc/psychology/writing/2009-sio.pdf ; https://psycnet.apa.org/fulltext/2008-18777-009.html 24. In that meta-analysis, the length of the incubation period was not a significant predictor of effect size (coefficient < .001). This contradicts the idea that a longer rest (24 hours rather than minutes) produces more benefit. — https://gwern.net/doc/psychology/writing/2009-sio.pdf 25. A longer preparation — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/edge-cases.md#evidence-from-incubation-research-indirect-a-different-task` - 1. The practice has three parts: (a) finish the draft; (b) leave it untouched for a delay; (c) come back and edit as a reader rather than as the author. UNC's Writing Center puts it as "Put the paper aside for a few hours, days, or weeks … Clear your head of what you've written so you can take a fresh look at the paper." https://writingcenter.unc.edu/tips-and-tools/editing-and-proofreading/ 2. The delay goes before the surface passes. UNC treats editing (content, structure, clarity, style) and proofreading (surface errors) as separate stages, with proofreading last, and advises checking for "o — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/mechanism.md#definition-and-parts` - **Verdict: BUDGET_EXHAUSTED (soft stop), not SATURATED-DEPTH.** The rate is still falling, but no two passes in a row came in under 5%. At least one or two more passes look likely to pay off: - Full text of Chanquoy (2001): the actual interval and the remaining-error result (settles X1 and C38). - The original Daneman & Stainton paper (settles X2 and the 20-minute detail). - Full text of Pilotti & Chodorow (2009) (settles X3). - Pilotti (2012), "Does familiarity with text breed complacency or vigilance?" - Begg & Wickelgren (1974): how fast memory for wording fades compared with memory for mea — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/rabbithole-synthesis.md#saturation` - 16. Burgoyne et al. propose that only extreme familiarity may impair proofreading, "of the kind that is developed by working on a piece extensively over the course of days, weeks, or months." If this is right, a cooling-off period matters most for long, heavily reworked documents. It matters least for a draft written in one sitting. — same PDF 17. Across both experiments, wrong real-word errors were detected less often than misspellings (for example, 81.25% vs 90.83% in Session 2). Function-word errors were detected less often than content-word errors. A rest period does not change which error — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/edge-cases.md#boundary-conditions` - Caveats: - The Daneman & Stainton (1993) figures come from Burgoyne et al.'s summary. I did not read the original PDF. - The Chanquoy (2001) delay length is unknown because the full text was blocked (HTTP 403). - No source directly tests a 24-hour interval against shorter or longer intervals for proofreading one's own writing. That is the main evidence gap. - Saturation verdict: BUDGET_EXHAUSTED after pass 0 and one deepening pass. The new-information rate had not been measured to below 5%. Candidates for further passes: the Daneman & Stainton original, Pilotti 2012 ("Does familiarity with tex — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/edge-cases.md#quality-gate` - - **c. 19 BC:** Horace says to hold the work back for nine years and to show it to judges. The goal is to avoid publishing irreversible mistakes. - **c. AD 95:** Quintilian says to hold the work back "a certain time" so that it reads as another's work. This is the first statement of the defamiliarization mechanism. He caps the interval for deadline work and warns against over-correction. - **2000:** King gives a minimum of six weeks for book drafts, with the same "someone else's work" rationale. - **2000s–present:** Writing centers and editing guides shrink the interval to "a few hours, days, — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/history.md#how-the-practice-evolved-summary-of-claims-1-15` - 5. Psychologist Tom Stafford (University of Sheffield) calls writing a high-level, meaning-focused task, so writers read what they meant rather than what they typed. https://kirkville.com/whats-up-with-that-why-its-so-hard-to-catch-your-own-typos-wired/ (mirror of http://www.wired.com/2014/08/wuwt-typos/) 6. Stafford's remedy is to make the text look unfamiliar: change the font or background, or print it. Time is one way to do that, not the only way. https://kirkville.com/whats-up-with-that-why-its-so-hard-to-catch-your-own-typos-wired/ 7. Daneman & Stainton (1993) found that people detected f — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/practice.md#mechanism-as-claimed` - 12. Chanquoy (2001) tested French 3rd–5th graders (N=60). In some conditions revision was postponed rather than done during writing. https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=Chanquoy%20revise%20their%20writing%203rd%20to%205th&resultType=core&format=json 13. Chanquoy's abstract reports: "The two postponed revising conditions led to more revisions than the revision occurring during writing." https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=Chanquoy%20revise%20their%20writing%203rd%20to%205th&resultType=core&format=json 14. Under immediate revision, surface corr — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/practice.md#evaluation-direct-tests-of-a-delay` - 1. **Does self-authorship hurt proofreading at all?** Daneman & Stainton (1993) say yes (claims 7–9). Burgoyne et al. (2022) found a non-significant reversal (claim 16). Pilotti & Chodorow (2009) found facilitation (claim 20). If there is no self-generation deficit, the cooling-off rationale for proofreading loses its premise. 2. **Does a delay help?** A two-week delay removed the deficit in Daneman & Stainton (claim 10). A one-week delay gave no benefit in Burgoyne et al. (claims 17–18). A next-day delay helped children's revision in Chanquoy (claims 13–15). The populations, tasks (revision v — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/practice.md#unresolved-disagreements` - - **C64.** Burgoyne et al. suggest only extreme familiarity may hurt, "of the kind that is developed by working on a piece extensively over the course of days, weeks, or months." — Burgoyne PDF [E,P] - **C65.** Wrong-word errors were caught less often than misspellings (e.g. 81.25% vs 90.83% in Session 2). Function-word errors were caught less often than content-word errors. A rest period does not change these error classes, and spell check misses both. — Burgoyne PDF [M,E,P] - **C66.** In every study, experimenters inserted the errors (e.g. 20 per essay), so writers' natural errors were never — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/rabbithole-synthesis.md#h-edge-cases-and-limits` - - The self-generation effect in proofreading - Reading aloud and reverse-order proofreading - Changing the format to make text unfamiliar (font or printout) - The incubation effect (Wallas) - Peer editing vs self-editing - Satisfaction-of-search in proofreading - Memory for a sentence's form vs its meaning (Sachs) - The generation effect — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/rabbithole-synthesis.md#handoff-to-concept-family-explorer-not-pursued-here` ## Facts and statements - 11. The UNC Writing Center says to "put the paper aside for a few hours, days, or weeks" before editing. It gives no fixed 24-hour figure. https://writingcenter.unc.edu/tips-and-tools/editing-and-proofreading/ 12. UNC treats proofreading as a separate, final stage for "surface errors such as misspellings and mistakes in grammar and punctuation", done after content revision. The "sentence pass" belongs to this stage. https://writingcenter.unc.edu/tips-and-tools/editing-and-proofreading/ 13. Bow Valley College's writing guide states the 24-hour rule explicitly: "Ideally, give yourself 24 hours b — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/history.md#institutional-guidance-where-the-24-hours-figure-appears` - 21. Sachs (1967) found that recognition memory for the form of a sentence declines much more rapidly than recognition memory for its meaning. — https://www.semanticscholar.org/paper/Recognition-memory-for-syntactic-and-semantic-of-Sachs/d6190336549fff6d971aecc373c195464788a52a ; https://psycnet.apa.org/record/1968-00308-001 22. Inference, not directly tested: if wording fades quickly and meaning persists, a 24-hour gap may reduce the writer's memory of exact wording (helping surface proofreading) while leaving the writer's intended meaning intact. Errors where the page says something other tha — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/edge-cases.md#mechanism-check-what-a-24-hour-gap-does-and-does-not-erase` - **In scope:** the practice of leaving a finished draft alone for a set time before sentence-level editing or proofreading. That covers where the idea came from, how the waiting time changed from "years" to "24 hours", why it is said to work, and the evidence for and against it. — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/history.md#scope` - 22. Daneman & Stainton's samples were very small (for example, n = 10 per condition in Experiment 1). https://link.springer.com/article/10.1007/s00426-022-01699-3 23. Burgoyne et al. (2022, *Psychological Research*) ran two eye-tracking experiments. Experiment 1 found no self-generation effect. Experiment 2 induced over-familiarity and found only a weak, non-significant effect. https://link.springer.com/article/10.1007/s00426-022-01699-3 24. In Burgoyne et al., time spent proofreading and number of text fixations predicted errors found. Fluid intelligence also predicted performance. This point — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/history.md#empirical-evidence-disconfirming-or-complicating` - **In scope:** the practice of setting a finished draft aside for a delay (conventionally about a day) before doing sentence-level editing and proofreading passes. The report covers why the delay is supposed to work, what it is made of, what stays true across its variants, and where the evidence stops. — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/mechanism.md#scope` - - **D1. Does proofreading your own text right away hurt?** - Yes: Daneman & Stainton 1993 (small n, overfamiliarity). - No: Burgoyne 2022 Exp. 1, a null result trending toward an author advantage, with higher power. - It helps: Pilotti & Chodorow 2009. - Weak, non-significant harm: Burgoyne Exp. 2 (p = .051). - **D2. Does a delay restore performance?** - Two weeks: yes (Daneman & Stainton). - One week: no (Burgoyne). - Next day: more and deeper revisions in children (Chanquoy). That was a different task and population. - **D3. What does the delay act on?** Each candidate predicts a different i — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/rabbithole-synthesis.md#disagreements-in-the-research-itself-not-averaged` - - **No 24-hour dose–response.** No source tests 24 hours specifically. Tested delays are 20 minutes, one week, and two weeks, and the incubation literature finds no effect of period length (claims 7, 11, 24). - **Familiarity can help.** For wrong-word and function-word errors, self-authorship improved detection in two larger studies (claims 10, 13). - **Meaning errors survive the rest.** A writer's intended meaning outlasts their memory of exact wording (claims 21–22). This is an inference, not a tested result. - **Short, single-sitting drafts.** Overfamiliarity may need days or weeks of rewor — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/edge-cases.md#failure-modes-summary-of-the-claims-above` - - **Does familiarity hurt or help own-text proofreading?** Daneman & Stainton (1993, small n) say it hurts (overfamiliarity). Pilotti & Chodorow (2009) and Burgoyne et al. (2022, Exp. 1) found it helps or does nothing. Burgoyne Exp. 2 found a weak, non-significant harm. — https://englelab.gatech.edu/articles/2022/Burgoyne%20et%20al.%20(2022)%20Revisiting%20the%20self-generation%20effect%20in%20proofreading.pdf ; https://eric.ed.gov/?id=EJ826333 - **Does a delay restore performance?** Daneman & Stainton report a recovery after two weeks. Burgoyne et al. found no gain after one week. — same Burg — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/edge-cases.md#unresolved-disagreements` - **Gaps:** - The Chanquoy interval (claim 21) and the King quotes (claims 8–9) rest on secondary sources. - The Chanquoy, Pilotti 2006, and Daneman & Stainton full texts were not accessible (403 or abstract only). - No original source was found for the specific "24 hours" figure. — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/history.md#quality-gate` - Date: 2026-09-24 · Parent context: Editing and Revision · Method: /rabbithole (3 passes, scoped brief) — source: `~/.global-ai-hub/research-runs/frontier-2026-09-24/24-hour-cooling-off-before-sentence-passes/reports/practice.md` ## Related concepts - 24-hour — is a part of 24-hour cooling-off before sentence passes - before — is a part of 24-hour cooling-off before sentence passes - cooling-off — is a part of 24-hour cooling-off before sentence passes - sentence — is a part of 24-hour cooling-off before sentence passes - passes — is a part of 24-hour cooling-off before sentence passes