IB TOK 知识主张 信念 真理 证成 反例
每一篇 TOK 展评与论文，真正的落点都是一句知识主张。主张写得太软，评语会说你没有立场；主张写得太硬，又会被追问凭什么。这篇把主张拆成可以检查的部件，从信念与真理的关系，走到证据门槛、反例与权力，最后落到展评与论文的评分口径上。
Every TOK exhibition and essay finally rests on one knowledge claim. Claim too little and the examiner sees no position; claim too much and the first question is what holds it up. This piece takes a claim apart into parts you can actually check, from the link between belief and truth through the threshold of evidence, counterexamples and power, and ends with how the exhibition and the essay are marked.
知识主张与知识问题的分工
知识主张是一句被当作真的陈述，说话的人希望别人也接受它。IBO 在 2020 年 2 月发布的《Theory of knowledge guide》把 TOK 的入口压成一个问题：我们怎么知道。这个问题要求暂时放下科目内容，去看知识怎么产生、怎么被检验、谁会否定它。知识主张是原料，知识问题是提炼，两者在展评与论文里必须配对出现。同一句话在不同场合身份不同，在物理卷上它是结论，在 TOK 里它是要被审的样本。很多学生以为把主张写长一点就更像分析，其实长度不带来任何论证。常见的错误是把课本结论直接抄成主张，却不说这个结论靠什么站住。
A knowledge claim is a statement put forward as true and offered for others to accept. The guide published by the IB in February 2020 reduces the whole course to one question, how do we know, and it expects that question to be asked about claims rather than about syllabus content; the sentence that closes a physics answer becomes the object of scrutiny in a TOK discussion. Length does not turn a claim into analysis, and copying a textbook conclusion without saying what holds it up is where most answers stall.
事实性、规范性、概念性三类主张
同样一句话，被要求拿出证据的方式并不相同。事实性主张谈世界是什么样，比如某座桥在 2007 年通车，可以拿档案与工程记录核对。规范性主张谈应不应该，比如学校该不该限制手机，靠的是价值判断与后果权衡，证据只能支持其中一部分。概念性主张追问一个词到底指什么，比如知识是否需要被证成，它靠定义与反例推进，而不是靠实验数据。这个区分在展评里很好用，评委看的就是你有没有意识到三类主张的检验方式不同。很多学生以为所有主张都能用实验验证，于是把伦理讨论写成数据罗列。常见的错误是给规范性主张堆统计，却不交代自己站在哪种价值立场上。
Not every claim is checked in the same way. Empirical claims describe how the world is and can be tested against archives, instruments or records; normative claims about what a school ought to do rest on values and consequences, where evidence carries only part of the weight; conceptual claims ask what a term means and move forward through definitions and counterexamples instead of data. Treating an ethical question as a data problem is a reliable way to lose marks.
信念、真理与证成三个条件
传统分析把知识拆成三个条件，你相信它、它是真的、你有理由相信它。1963 年 Edmund Gettier 用三页短文给出反例，说明三条同时满足仍可能不算知识。他的例子都是运气帮忙的情形，理由成立，结论也恰好为真，中间的联系却是碰巧的。TOK 不需要学生背原文，需要学生用这个反例去看自己举的例子有没有运气成分。比如你相信某位同学在图书馆，因为看见他的书包在座位上，这个理由不差，可书包也可能被别人借放。很多学生以为证成就是找到一个理由，其实还要问这个理由与结论之间是不是真的连得上。常见的错误是把相关当理由，写完就停。
Knowledge has traditionally been analysed as belief that is true and justified. Gettier showed in a three page paper in 1963 that all three conditions can hold while the belief still fails to count as knowledge, because the tie between the reason and the conclusion was luck; a schoolbag on a chair is decent evidence that its owner is nearby, yet somebody else may have put it there. What matters in TOK is whether your own example survives that kind of luck, not whether you can recite the paper.
四条常见证成路径及其边界
证据、权威、推理与直觉是学生最常调用的四条路径。证据把主张挂到可观察的东西上，推理从已知推到未知，权威借别人的专业判断，直觉是自己那种说不清来源的确信。四条路径的分量随领域变化，数学里推理最硬，历史里档案证据最硬，医学里临床试验与同行共识都算数。权威最容易被滥用，因为引用一个有名的人，不等于他的判断适用于你手上这个情况。IBO 指南把证成与证据列为贯穿课程的十二个概念之一，理由就在这里。很多学生以为引用专家就完成了论证，其实还要说明这位专家的结论凭什么落到当前案例上。常见的错误是四条路径混着用，一段话里既讲实验又讲感觉。
Students justify claims through evidence, authority, reasoning or intuition. The weight of each route shifts by area: reasoning dominates in mathematics, archival evidence dominates in history, and clinical trials with professional consensus carry medicine; authority is the route most easily abused, since a famous name does not guarantee that the judgement travels to your case. The IBO guide lists justification and evidence among the twelve concepts that thread through the course for exactly this reason.
证据门槛在五个领域各不相同
什么样的东西算证据，五个领域给出的答案差得很远。自然科学依靠可重复的实验，一次结果不算数，要被别的实验室重复出来。历史依靠留存下来的痕迹，文件、器物、口述与碳定年数据，缺了痕迹就只能说不知道。数学的证据是证明，一个反例就能推翻全称命题，不需要抽样。艺术与人文学科更多依靠文本细读与解释的融贯，同一幅画可以有两种站得住的读法。把这些门槛混在一起，文章就会在历史题里要求做实验，在数学题里要求找统计。很多学生以为证据越多越好，其实证据的类型对不对更关键。常见的错误是拿一篇新闻当全部依据，却没查它的原始出处。
What counts as evidence differs sharply across areas of knowledge. The natural sciences rely on repeatable experiment, so a single result carries little until another laboratory reproduces it; history works from surviving traces such as documents, objects, testimony and radiocarbon dates, and where the traces are missing the honest answer is that we do not know. Mathematics settles claims by proof, where one counterexample destroys a universal statement without any sampling at all.
相关、因果与观察者带来的偏差
两个变量一起变化，只说明它们同时出现，不说明谁推谁。冰淇淋销量与溺水人数在夏天一起上升，真正干活的是气温。人文科学多一层麻烦，被研究者会对你作出反应，问卷里的答案会向体面的方向偏移。这就是社会赞许偏差，问学生是否作弊的时候尤其明显。样本也决定结论能走多远，便利抽样得到的班级数据无法直接代表全国。TOK 里讨论这些，落脚点是主张的适用范围有多大。很多学生以为只要有一个数据点就能支撑普遍结论，其实那只是一个观察。常见的错误是把相关性写成因果，还顺手加上一个很大的结论。
Two variables rising together only show that they appear together. Ice cream sales and drowning figures both climb in summer, with temperature doing the work; the human sciences add a further complication because respondents react to being studied and answers drift towards the respectable option, which is why anonymity changes what a survey on cheating reports. Sample design also fixes the reach of a claim, and a convenience sample from one class cannot speak for a country.
确定性与概率的两种含义
把主张写成几成把握，往往比写成对错更诚实。医学筛查是个好例子，一种患病率很低的病，即便检测的假阳性率不高，阳性结果也可能大部分来自健康人。原因不在检测本身，而在基数太小。天气预报说降雨概率七成，也不等于七成的地方会下雨，它指的是相似天气条件下有七成出现降雨。这两类概率经常被学生混为一谈，展评里一旦写错，评委会立刻追问。IBO 指南把确定性列进十二个概念，要求你说清自己在哪种意义上确定。很多学生以为加上概率就变严谨，其实没说清概率指什么反而更含糊。常见的错误是把置信直接当成事实。
Stating how confident you are is often more honest than declaring a claim right or wrong. Screening shows the trap, since when a condition is rare most positive results can come from healthy people even with a fairly specific test; a seventy per cent chance of rain means rain on seventy per cent of days with similar conditions, not rain over seventy per cent of the area. Certainty is one of the twelve TOK concepts because the sense in which you are certain has to be spelled out.
反例对不同主张的杀伤力不同
一个反例对不同类型的主张，杀伤力差得很远。全称主张说所有天鹅都是白的，一只黑天鹅就够了。统计主张说大部分学生在夜间效率更高，一个反例只是让分布多一个点，并不推翻结论。写 TOK 时要先认清自己的主张属于哪一种，再决定反例怎么用。更强的做法是主动写出反主张，说明它在什么条件下成立。这样论证就从单向陈述变成两个方向的对话。很多学生以为反驳就是把对方说错，其实承认对方在某个范围内成立更有说服力。常见的错误是举了一个例子，却说它证明了整个规律。
A counterexample does different damage to different claims. One black swan ends the statement that all swans are white, yet a single student who works better in the morning does not overturn a statistical claim about most students; recognising which kind of claim you hold is what decides how a counterexample should be used. Writing the opposing claim and granting where it holds turns a one sided answer into an argument.
谁在说话：立场、经费与同行评审
主张的可信度与提出它的人有关，但关系不是简单的好人与坏人。经费来源会塑造研究问题，药企资助的试验更常报告有利结果，这不需要造假，选题与发表偏向就够了。同行评审能挡住明显错误，挡不住整个领域共同的假设。档案的保存也带权力色彩，被记录下来的往往是有资源的一方。IBO 指南把权力与视角放进十二个概念，提醒学生别把知识当成一份中立的清单。很多学生以为只要找到论文就万事大吉，其实还要看这篇论文在什么条件下产生。常见的错误是把机构名当成论据。
Who makes a claim shapes how far it should be trusted. Funding steers research questions, so industry sponsored trials tend to report favourable outcomes without anyone fabricating data; peer review catches clear errors but not the assumptions a whole field shares, and archives preserve the records of those with resources. Power and perspective sit among the twelve concepts for that reason, and naming an institution is not the same as giving an argument.
共享知识与个人知识互相修正
TOK 里的知识既有共享的一面，也有个人的一面。共享知识经过共同体检验，写进教科书，可以被别人复核。个人知识来自你自己的经历，比如你熟悉某座城市的公交系统，这类知识很难完整传给他人。两者会互相修正，你在实习里看到的病房流程，可能和课本写的标准流程有出入，这个出入本身就是很好的展评材料。文化也在中间起作用，同一个手势在不同地方含义不同，这让主张的语境变得重要。很多学生以为个人经验不算知识，其实它是检验共享知识的入口。常见的错误是只写自己的感受，不写出它和哪个共享主张冲突。
Knowledge is shared and personal at once. Shared knowledge has been tested by a community, written into textbooks and left open to challenge, while personal knowledge comes from your own experience of a bus network or a hospital ward and resists full transfer; where the two disagree, as when ward practice differs from the textbook flowchart, you have material worth analysing. Culture sits between them, since the same gesture carries different meanings in different places.
展评里三件实物与 950 词怎么分配
展评要求从 IBO 公布的 35 个提示词里挑一个，配三件真实物品，评论合计 950 词，占 TOK 总分三分之一，满分 10 分。评分看的是每件物品有没有具体的现实语境，以及它与提示词的关系是否被论证过。这意味着物品贵不贵不重要，重要的是你能不能说清它触发了哪一个知识主张。一个实用的选择是校内公告栏，它能同时连到证据、权威与谁有权决定信息。写作时每件物品大约 300 词，留出开头交代提示词的空间。很多学生以为物品越稀奇越好，其实评委更在意语境是否真实可查。常见的错误是把三件物品写成三段互不相干的描述。
The exhibition asks you to choose one of the thirty five prompts published by the IBO, attach three real objects and write 950 words of commentary, which is one third of the TOK grade and marked out of ten. Marks come from whether each object is grounded in a specific real context and whether its link to the prompt is argued rather than asserted, so a school noticeboard works as well as an expensive artefact when the claim it triggers is made clear. Three unconnected descriptions is the standard way this task goes wrong.
论文的 1600 词与常见扣分点
论文从每年公布的 6 个题目里选一个，1600 词，占三分之二，论证要落在两个领域上。题目本身通常是关于知识的一般问题，所以整篇不能只讲一个科目的内容。学生最常把描述当分析，写了一大段实验过程，却没有说这个过程为什么能产生可信的主张。例子太熟也是问题，爱因斯坦与牛顿年年出现，评委看不出你的判断在哪里。主张开得太大同样危险，一句话想解释所有知识，结果哪一头都落不下去。很多学生以为引用名人名言能加分，其实那只是装饰。常见的错误是结论重复开头，没有回应中间的转折。TOK 与 EE 合计最多贡献 3 分加分，成绩按 A 到 E 计。
The essay takes one of six prescribed titles released each year and runs to 1600 words, carrying two thirds of the grade with an argument that has to work across two areas of knowledge. The commonest weakness is description dressed as analysis, where a long account of an experiment never explains why that method yields trustworthy claims; familiar examples such as Newton and Einstein, and claims stretched to cover all knowledge at once, both leave the examiner with nothing to credit. A conclusion that merely repeats the introduction wastes the strongest part of the essay.
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
A-Level / GCSE / IB / KS2-KS3 全科辅导
更多咨询请联系16621398022（同微信）
Visit aleveler.com
或截图后用「扫一扫 → 相册」选择本图