提取练习 间隔重复 遗忘曲线 复习排期
很多学生以为提取练习和间隔重复是两种互相竞争的记忆方法，挑一个用就行。它们管的不是同一件事：提取练习决定坐下来学的当下做什么动作，间隔重复决定这些动作在日历上怎么分布。把它们当成对手来比较，往往是因为复习时只用了一半，要么反复重读却从不合书回想，要么排好了日程却每次都在看书。真正拉开分数的，是把两个动作接在同一张时间表上。Dunlosky 等人在 2013 年发表于《Psychological Science in the Public Interest》的综述评了十种常见学习法，只有练习测试与分散练习被评为高效用，重读与划重点属低效用。这条结论本身就是答案的一半，不是二选一，而是两个都上。
Active recall and spaced repetition are routinely framed as rivals, as if a student had to pick one. They work on different dimensions: retrieval fixes what happens inside a session, while spacing fixes where those sessions sit on the calendar. A 2013 review in Psychological Science in the Public Interest rated only practice testing and distributed practice as high utility, with rereading and highlighting at the bottom, so the honest answer is not either or, but both chained into one timetable.
提取练习和间隔重复管的不是同一件事
提取练习管的是一次学习里的动作：把书合上，用回忆把答案从脑子里调出来，而不是把答案再看一遍。间隔重复管的是这些动作在日历上的位置：同一个知识点隔多久碰一次，隔几次之后才算稳。前者改的是单次学习的强度，后者改的是同一条信息被重新激活的次数。把两者比较成谁更有效，问题本身就问偏了，因为它们作用在不同的维度上，一个回答怎么做，一个回答什么时候做。只做提取不排间隔，同一批材料会在六周里回到原点；只排间隔不做提取，日程表再满也只是一遍遍确认眼熟。两条线交叉的地方，才是复习真正发生的位置。
Retrieval governs what happens inside one session: the book closes and the answer has to come back from memory instead of being read again. Spacing governs where those sessions land on the calendar, meaning how often one item returns and after how many passes it holds. The two get compared as if they were alternatives, yet one answers how to study and the other answers when, and neither works alone for long.
合上书把答案写出来才算提取
判断有没有真正做提取，看的是眼睛有没有离开材料。常见的错误是边看边在心里点头，那种熟悉感来自识别，不是回忆，考场上没有题干提示就调不出来。Roediger 与 Karpicke 在 2006 年发表于《Psychological Science》第 17 卷的实验里，一组反复阅读，另一组读一遍后做回忆测试，五分钟后反复阅读组略好，一周后回忆组明显领先。很多学生以为多读几遍总能记住，重读带来的其实是越来越强的熟悉感。具体动作可以很土：合上书在空白纸上写下这一节的三个核心定义，写完再打开对照，用另一种颜色的笔补上漏掉的部分。写不出来比写错更值得注意，说明这条信息还没进入可提取的状态。
You know retrieval has happened when your eyes are off the page. The common mistake is nodding along while reading, because that fluency comes from recognition rather than recall, and it collapses once the prompt disappears. In a 2006 experiment in Psychological Science, Roediger and Karpicke had one group reread a passage and another recall it; the rereading group led after five minutes, while the recall group led clearly a week later. A workable routine is blunt: shut the book, write the three core definitions on blank paper, then reopen and fill the gaps in another colour.
一次复习拆成三次的落点
间隔重复要解决的是同一条信息被激活几次。Cepeda 等人在 2008 年发表于《Psychological Science》的研究给出的经验值很实用：复习间隔大约取目标记忆期限的一成到两成。目标是三个月后的考试，第一次间隔就落在十天左右，往后每次翻倍。把一次两小时的复习拆成三次四十分钟，中间隔着几天，效果通常好过连坐两小时。拆的时候按知识块拆，不按页数拆：一个知识块要能被一次完整回忆，拆到半页那种碎片反而记不住。三次里第一次在学完当天，第二次隔两三天，第三次隔一周以上，后面按考试日拉长。
Spacing answers a simpler question: how many times does the same item get reactivated. Cepeda and colleagues reported in a 2008 Psychological Science paper that a useful rule of thumb is to set the gap at roughly ten to twenty per cent of the period you need to remember the material. Three forty-minute sessions spread across days usually beat one two-hour block, and the split should follow knowledge blocks rather than page counts.
遗忘曲线不是一条人人相同的曲线
艾宾浩斯在 1885 年用无意义音节做的自我实验，画出了那条被反复引用的遗忘曲线。很多学生以为曲线上的数字可以直接套到自己的复习表上，问题是他测的是无意义音节，与你背的公式、定义、案例不是同一种材料。有意义的材料遗忘慢得多，而你已经理解的部分几乎不参与那条曲线。曲线真正有用的部分是形状：前期掉得快，后期趋平，所以第一次复习要早，后面可以拉开。按这个形状排表，比背具体百分比有用得多。
Ebbinghaus measured his own forgetting in 1885 using nonsense syllables, and that curve still gets quoted everywhere. The catch is the material: syllables are not formulas, definitions or case studies, and meaningful content decays far more slowly. What transfers is the shape, steep at first and flatter later, which is why the first review should come early and the rest can be stretched.
重读和划重点为什么感觉最有效
重读和划重点之所以感觉最有效，是因为它们每一次都让材料变得更眼熟。眼熟是一种识别层面的顺畅，读第二遍比第一遍快，划线时的手感也在告诉你正在学到东西。Karpicke 与 Blunt 在 2011 年发表于《Science》第 331 卷的对比里，用提取练习的一组在一周后的理解测试上超过了画概念图的一组，而两组对自己掌握程度的估计差不多。这个差距落在判断上，不在努力上：两边花的时间接近，只有一边在测试自己。想留住重读，位置应该放在提取之后，用来核对漏掉的部分，而不是放在提取之前当预热。
Rereading and highlighting feel effective because each pass makes the page more familiar, and familiarity is a smoothness that has nothing to do with retrieval. In a 2011 comparison in Science, Karpicke and Blunt found that students who practised retrieval outperformed those who built concept maps a week later, even though both groups rated their own understanding about the same. That gap sits in judgement rather than effort, which explains why rereading belongs after a recall attempt, as a check, not before it as a warm-up.
公式表与公式册决定了哪些内容不用背
哪些内容需要能默写，取决于考试当天发不发公式。OCR 官网在 2026 年 9 月的科目通知里写明，GCSE 数学、物理与综合科学的公式与方程表在 2025 至 2027 年照发，这个安排在现行大纲有效期内继续有效，DfE 与 Ofqual 在 2024 年确认过。Edexcel 的 A-Level 数学（9MA0）每场考试都提供 Mathematical Formulae and Statistical Tables，商法则在册，乘积法则与链式法则不在册。这两条直接决定提取练习的清单：在册的公式只需要会挑会用，不在册的必须能默写并代入。常见的错误是把整页公式表都当不用背，考场上翻不到那一条就卡住。
What you must be able to write from memory depends on what the exam hands you. OCR's subject update of September 2026 confirms that the formula and equation sheets for GCSE maths, physics and combined science continue through 2025 to 2027, a position DfE and Ofqual confirmed in 2024. Edexcel's A-level Maths 9MA0 supplies the Mathematical Formulae and Statistical Tables booklet in every paper, with the quotient rule included but the product and chain rules left out. That distinction decides which items go on a recall list and which only need to be located and applied.
从考试日往回推的间隔表怎么排
排间隔表最省事的起点是考试日，不是今天。AQA 官网公布的 2027 年夏季时间表把 GCSE 场次排在 5 月 10 日到 6 月 22 日之间，每天 9:00 与 13:30 两个时段，JCQ 官网写明 2027 年夏季 GCSE 成绩在 8 月 19 日发给考生。从最后一个场次倒推六周，把这段留给整套卷子的限时训练。再往前四周留给按知识块的提取与补漏。最早的四周只做一件事，把每个知识块做到能被完整回忆。这样倒推每个阶段都有明确产物，而不是靠感觉判断进度。很多学生以为计划要从今天往前铺，结果到考前两周才发现还有三章没碰过。
Build the timetable from the exam date backwards, not from today forwards. AQA's published 2027 summer timetable runs GCSE papers from 10 May to 22 June in 09:00 and 13:30 sessions, and JCQ lists GCSE results for summer 2027 as 19 August. Working back six weeks gives a block for full timed papers, the four weeks before that belong to recall and gap-filling by knowledge block, and the earliest month exists to make each block retrievable at all.
交错练习该放在间隔表的第几次
交错练习解决的是题型辨识，不是记忆本身。同一篇综述把交错练习评为中等效用，位置排在提取练习与分散练习之后，原因是它增加的是辨别能力，在需要判断用哪个公式的卷子上收益明显。放的位置有讲究：同一个知识块连续做两三次提取之后再交错，不要在还没记住的时候就开始混着做。混的时候按题型混，不按章节混，把需要判断用哪种方法的题放在一起才有意义。常见的错误是整本练习册从头做到尾，题号连续，脑子根本不用决定用哪条思路。
Interleaving trains discrimination rather than retention, which is why the same 2013 review placed it at moderate utility, behind practice testing and distributed practice. Put it after two or three consecutive retrievals of one block, never before the block is retrievable at all. Mix by question type rather than by chapter, because a page of similar-looking problems never forces the decision that the exam will force.
判断会了没有的三条判据
判断一个知识块是否过关，有三条很硬的判据，都不需要别人帮忙。第一条是隔三天不给任何提示，还能把定义和条件写出来。第二条是换一个问法，比如从求值换成解释为什么，仍然能答。第三条是能在限时里答完，而不是想十分钟才想起一半。三条里最容易骗过自己的是第二条，因为背下来的句子换个问法就用不上。做不到的时候不要归因于记性差，多半是提取的次数不够，或者间隔没拉开。把这三条写进复习表，每次复习结束前用一条去测，比凭感觉判断靠谱。
Three hard tests tell you whether a block is actually secure, and none of them needs a tutor. Can you reproduce the definition and its conditions after three days with no prompt; can you answer the same content under a different wording, for instance explaining why rather than calculating; can you finish inside the time limit rather than needing ten minutes to recall half of it. The second test is the one students pass least often, and failing it is usually about the number of retrievals rather than memory.
只做一半的两种走法
两种半途而废最常见。一种是只做提取不排间隔，考前三周每天做同一批题，短期手感很好，第四周开始遗忘，第六周回到原点。另一种是只排间隔不做提取，日程表上每天都有复习，翻开书看一遍就划掉，看起来完成了，实际每次都在做识别。前一种的补救是把同一批题推到不同的周，后一种的补救是把看改成合书回想，时间不用加。很多学生以为复习表排满就说明在进步，表的密度和记忆的强度是两件事。判断属于哪一种，看的是复习时书是开着的还是合着的。
Two half-measures account for most wasted revision hours. Some students retrieve intensely for three weeks, feel fluent, then lose it by week six. Others keep a full calendar but only reread each time, which is recognition rather than recall. The first group needs the same material pushed into different weeks, the second needs the book shut, and neither fix requires extra hours.
提取材料从真题和错题里来
提取的材料不用另造，考试材料本身就是最好的题库。真题按知识块切开放，一道题对应一个可以完整回忆的小块，切得太细就没有判断的空间。错题要留着原样，包括当时的错误答案和错因，重做时先看错因再看题。课本每节末尾的小标题是现成的自测清单，合上书按小标题复述一遍，说不出来的就是没读进去的部分。公式表反着用，遮住含义只看符号，把每个符号的条件与单位说出来。材料准备到这一步，提取练习就不需要额外花时间设计。
You do not need to invent recall material, since past papers and the textbook already contain it. Cut questions along knowledge blocks so each item tests something you can hold in one pass, and keep wrong answers verbatim, including the reasoning that failed. Cover the formula sheet and name each symbol with its conditions and units, a drill that exposes the gaps rereading hides.
复习表上要记录的三列数据
复习表上只记三列就够，写多了没人看。第一列记这个知识块上次复习的日期，用来算间隔有没有拉动。第二列记不给提示的回忆正确率，写对几题错几题就行。第三列记这次漏掉的具体内容，写一句话，不要只写不熟两个字。三列放在一起，间隔该拉长还是该缩短一眼能看出来。正确率连续两次在高位，间隔就翻倍；掉到一半以下，回到上一次的间隔。阅卷时能看出一个学生有没有真练过提取，答案的措辞会带上自己组织的痕迹，背下来的整段反而对不上题目要求。
Three columns are enough on a revision sheet. Log the date of the last review, the unprompted recall score, and one line naming exactly what was missed, never just a vague label. Read together, they show whether the gap should double or shrink back, and examiners can usually tell from the phrasing of an answer whether a student has genuinely retrieved the material or memorised a paragraph.
牛津高阶英汉双解词典 第10版 商务印书馆¥168📱 去京东购买京东精选
需要一对一辅导？Need One-on-One Tutoring?
A-Level / GCSE / IB / KS2-KS3 全科辅导
16621398022 同微信
Visit aleveler.com
或截图后用「扫一扫 → 相册」选择本图