Numbers, formal logic, reading, and everyday judgement — ten items each. Timed in the background. At the end you see where your thinking slips, not how smart you are.
Take the check →About 15 minutes. No account. Nothing unlocked at the end.
| Part | What it looks at | Example of the trap |
|---|---|---|
| Numbers & probability | Percentages, independence, base rates | A price rises 20% then falls 20% — it does not return to where it started |
| Formal logic | Contrapositive, necessary vs sufficient, sets | "Only after 10 uses" is not the same as "after 10 uses" |
| Reading | Main claim, what weakens it, what the author would refuse | A correlation is reported; the confounder is left unmentioned |
| Everyday judgement | Sunk cost, framing, survivorship, regression | A bad month improves after a talking-to — mostly regression to the mean |
None of these items requires specialist knowledge. They test moves of thought: checking whether the inverse is true, asking what else changed at the same time, noticing that the sample only includes survivors. Those moves transfer to reading, to work, and to arguments you have with yourself.
That is also why the result is worth taking twice. These are habits, and habits move.
The whole value of this check is that you answer without looking things up. If the timer were visible, it would create pressure and invite clock-watching; if there were no timer at all, a 90-minute attempt would be indistinguishable from a 15-minute one. So it runs silently in the background and is reported at the end.
The time itself is also information. Getting the same score in 22 minutes that took you 38 last month is progress, and it is the kind of progress that is otherwise invisible.
Look at the two weakest domains first, then at the cluster list — it names the specific trap you fell into ("base rate", "necessary vs sufficient", "sunken cost"). That name is the useful part: once you can name the trap, you start seeing it outside the test.
Ignore the total. 28 out of 40 tells you almost nothing on its own; "I keep missing the confounder" tells you what to practise.
Is it free?
Yes — no account, no ads, no paywall. Scoring happens in your browser.
How long does it take?
About 15 minutes at a normal pace. There is no visible clock.
What if I get interrupted?
Leave the page and the attempt is void — this check is not saved as a draft. Come back when you have fifteen uninterrupted minutes.
Can I retake it?
As often as you like. The items rotate between attempts so you are not memorising answers.
数字、形式逻辑、阅读、日常判断，各 10 题。后台静默计时。做完你看到的是"思路在哪一步滑掉了"，不是"你有多聪明"。
开始 · 40 题 →约 15 分钟。不用注册。做完不会有任何东西被锁住。
| 部分 | 看什么 | 典型陷阱 |
|---|---|---|
| 数字与概率 | 百分比、独立性、基率 | 先涨 20% 再降 20%，回不到原价 |
| 形式逻辑 | 逆否、必要与充分、集合 | 「只有完成 10 次」不等于「完成 10 次就」 |
| 阅读理解 | 主旨、什么能削弱它、作者不会同意什么 | 只报了相关，没提同时发生的另一个原因 |
| 日常判断 | 沉没成本、框架、幸存者、回归 | 最差的一个月之后回升，多半只是回归 |
这 40 题不需要任何专业知识。它考的是思维动作：逆命题成不成立、同一时间还变了什么、样本里是不是只剩活下来的人。这些动作能迁移到读书、工作和自己跟自己的争论里。
也因为它们是习惯，所以值得做第二次——习惯会动。
这套题的全部价值在于：你是在没查资料的情况下答的。秒表显示出来会制造压力、诱人一直看时间；完全不计时，又分不清 15 分钟和 90 分钟。所以它在后台静默走，只在结束时报给你。
用时本身也是信息。同样的分数，上个月 38 分钟、这次 22 分钟，那就是进步，而且是平时看不见的那种进步。
先看最弱的两个方向，再看"错的地方集中在"那一行——它会点出你具体掉进哪个坑（"基率""必要 vs 充分""沉没成本"）。那个名字才是有用的部分：一旦能叫出陷阱的名字，你在题外也会开始看见它。
总分可以忽略。28 / 40 本身说明不了什么；"我老是漏掉那个共同原因"才告诉你该练什么。
免费吗？
免费。不注册、没有广告、没有付费墙，算分在你浏览器里完成。
要做多久？
正常节奏约 15 分钟。没有可见的时钟。
中途被打断了怎么办？
离开页面本次就作废——这套题不存草稿。等你有十五分钟不被打扰的时候再来。
可以重复做吗？
随时。每次的题目会轮换，不会让你背答案。
全部测试 · 大五人格 120 题 · 番茄钟