Xiaoyang Yu

本文只处理一句常见的辩护。当有人指出”当初要是没那样”这类设想并没有发生过，通常会得到这样的回应：它虽然没发生，但它在物理上是可能的，并没有违反物理定律。 本文的判断是：这句话是一个需要计算才能支撑的断言，而这个计算从来没有人做过。 先说这句辩护为什么一向不必付账。一件实际发生过的事，在物理上一定是可能的——发生了，所以可能，这条不需要任何证明工作。可能性一直是搭着”实际发生”的便车来的。 而反事实场景里没有实际发生的东西，便车没了，可习惯还在。此处还有两条要点明：“没有禁令”不等于”有许可”——“不违反已知定律”是一句关于我们那套写法的话，不是关于世界的话；以及**“我一下子没看出哪里违反”不等于实际上没违反**——没看出来是一句关于看的人的话。相似同样不传递可能性：实际发生的那件事只给它自己发通行证，不给旁边站着的那些发。 那么要证的是什么？两件：该场景内部未违反物理定律，以及该场景与之前的实际情况的接缝处未违反物理定律。而程序是确定的：先在最细的颗粒度上把该场景定义出来，再从接缝处做拉普拉斯妖式的计算，最后看算出的接缝后场景与你所定义的是否完全吻合。吻合了，才说明没违反。 本文并逐一处理三样常被搬出来当许可证的东西，指出它们全都是任务书。其一，若在这件事所在的尺度上随机性不显现（日常反事实正是如此，微观涨落在这个尺度上被平均掉了），接缝处只有两条路：接缝那一刻的状态与实际的一模一样，那么算出来的就是实际历史，与你的场景对不上；或者不一样，可这一动不由前一刻算出，而不由前一刻算出的改动正就是”违反”的意思。其二，量子随机性并不构成豁免，因为随机性仍然是可以计算的——它给出的不是”什么都行”，是一族有确定分布的后继；程序一步没少，只是第三步从”看那一个吻不吻合”变成”看这一族里有没有一个吻合”。门槛降到这里，也仍然只剩一句：你算了吗？没算你凭什么说没违反？ 其三，混沌放大同样不帮忙：混沌给出的是发散，而发散不指向任何特定终点；发散越厉害，落到”我没搬家”这个指定结局上的活儿越重，不是越轻。何况主张”由某个微小涨落放大而成”的人，还欠着两笔：是哪一个涨落，沿哪条路径。 最后须把主张说准：本文不主张那个计算做不到——理论上输入取得到，思想实验也行。本文主张的是：没人做过。 这一条不依赖任何一套写法，决定论也好、量子随机也好、混沌也好，都得算，而三边都没算。所以那句辩护不是被驳倒了，是还没有被提出来：在计算拿出来之前，那一栏是空的，而我们一向把它当成已经填好的。 Abstract This paper addresses one common defence. When it is pointed out that a scenario such as “if I hadn’t done that” never happened, the reply is usually that although it did not happen, it is physically possible and violates no law of physics. The paper’s claim is that this is an assertion requiring a calculation to support it, and that calculation has never been performed by anyone. First, why the defence has never had to pay. Whatever actually happened is thereby physically possible—it happened, so it was possible, and no proof is called for. Possibility has always ridden along with actuality. A counterfactual scenario contains no actuality; the ride is over, but the habit remains. Two further points: the absence of a prohibition is not the presence of a permission—“violates no known law” is a remark about our notation, not about the world; and “I do not offhand see a violation” is not the absence of one—not seeing is a remark about the observer. Resemblance likewise transmits no possibility: what actually happened issues a pass to itself alone, not to whatever stands nearby. What, then, must be shown? Two things: that nothing within the scenario violates physical law, and that nothing violates it at the seam where the scenario joins the preceding actual course of events. The procedure is definite: define the scenario at the finest granularity; compute forward from the seam in the manner of Laplace’s demon; then check whether the computed post-seam scenario matches exactly the scenario as defined. Only then is no violation shown. The paper takes up three things regularly produced as licences and shows each to be a work order instead. First, where randomness does not manifest at the scale in question—as with everyday counterfactuals, microscopic fluctuation being averaged out at that scale—the seam admits only two routes: the state at the seam is identical to the actual one, in which case the computation yields actual history and fails to match your scenario; or it is not, in which case that alteration was not computed from the preceding instant, and an alteration not so computed is precisely what “violation” means. Second, quantum randomness grants no exemption, because randomness is still calculable—it yields not “anything goes” but a family of successors with a definite distribution. The procedure loses not a single step; only the third changes, from “does that one match?” to “does any member of the family match?” With the bar lowered this far, one question remains: have you calculated? If not, on what basis do you say no law was violated? Third, chaotic amplification does not help either: chaos yields divergence, and divergence points at no particular endpoint. The stronger the divergence, the heavier the work of landing on the one specified outcome, not the lighter. Moreover, whoever claims the outcome was amplified from some minute fluctuation owes two further items: which fluctuation, and along which path. Finally the claim must be stated exactly. The paper does not hold the calculation to be unperformable—inputs are in principle available, and a thought experiment will serve. The paper holds that no one has performed it. This does not depend on any particular notation: deterministic, quantum-stochastic, chaotic—all require the calculation, and on all three sides none was done. The defence has therefore not been refuted; it has not yet been made. Until the calculation is produced that column is blank, and we have always treated it as already filled in.

本文讲我们对待心里那些画面的真实方式。通常的说法是：我们要么信它，要么不信它。本文认为不是这样。 **默认不是"信"，是按真实入账，再减去我们不得不承认的那部分。** 一段"当初要是没那样"的画面出现时，如果没人质疑，它就整个挂在"真实"名下。有人指出"可是实际上你并没有那样啊"，我们才承认一处不同；再追问，再承认一处。承认了的那几处，是折扣；没被追问出来的部分，仍然挂在"真实"名下。 这里有三件事要说清楚。 **第一，折扣的大小，不由这段画面假的程度决定，由我们被逼的程度决定。** 折扣不是我们发现的，是被人抠出来的。没人抠，就一分不减。 **第二，开票的和抵扣的是同一个人。** 要算"这一段跟实际发生的差在哪里"，得拿两样东西来比；可比的两边都是心里的画面。这笔账没有外人来审。 **第三，也是最要紧的一条：虚构不是打了折的真实。** 折扣这条轴上量的是"差多少"，零点是完全属实，往下是有出入、大部分不实、几乎全错。可"几乎全错"仍然是**对一件发生过的事的错误描述**。而"当初我没那样"这段画面，不是对任何一件发生过的事的描述——它讲的那件事根本没有原件。所以它不在这条轴上。**在这条轴上一直减，得到的是"一件真事被说得很离谱"，永远得不到"没有这件事"。** 单位不对。 **第四，也是全篇的落点：不是"剔除掉不真实的部分之后，剩下的是真的"。** 那段画面里，那个城市确实存在，我确实有一副身体——可这些零件的真，是它们各自另有出处，是被借来当布景的；借来的真不算这张单子的真。一张假钞上的纤维是真纤维，油墨是真油墨，可没有"这张钞票有九成购买力"这回事，它的购买力是零。真实性不是画面里各个零件的属性，是整段画面所讲的那件事是不是那样；而它所讲的那件事从来没有过。所以**它的真实性不是"很低"，是零**。前面那些被抠出来的"不同之处"，因此也不是折扣——不是从一个正的真值里减掉的份额，而是布景被一件一件认出来的过程。**零不是减出来的，一开始就是零。** **第五，这不等于说这些画面没有意义或功能。** 它们能让人后悔得睡不着，能让人下决心，能让人第二天改主意——这些作用一分不假，确实发生了。但那是**这段活动**的属性，不是**它所讲的那件事**的属性。混淆出在同一个"真"字上："这话说得真"有时指贴切、打动人，有时指所述属实；前者说作用，后者说真实性。**这两样是两个互不相干的领域。** 一段画面不需要任何真实性就能起全部作用——后悔靠的不是它为真，是它被按真实入账了；反过来，一段完全属实却没人看的记录，作用是零。所以本文的结论要分两半说，免得被听成一半：**那些画面的真实性是零；那些画面的作用是实实在在的。** 这两句不冲突，因为它们说的不是同一件事。 **第六，"不真"也不等于"可能"。** 到这里通常有最后一次退守："就算它不真，它至少是可能的吧？"这一次要拆的不是"可能"这个词，是它**从哪里领的**。你**可以**去证明这些画面在物理上可能；但**在证明完成之前，你不应当认为它们在物理上可能**。而这份举证责任之所以一向没人负，是因为可能性一直是白拿的：**一件实际发生过的事，在物理上一定是可能的**——发生了，所以可能，这条不需要任何证明。**可能性一直搭着"实际发生"的便车。** 这段画面里没有实际发生的东西，便车没了，可习惯还在。此处还有两处要点明：**"没有禁令"不等于"有许可"**——"不违反已知定律"是一句关于我们那套写法的话，不是关于世界的话，我们手上没有禁令，跟世界发过许可证，是两件事；以及**相似不传递可能性**——实际发生的那件事只给它自己发通行证，不给旁边站着的那些发。最后须说清这一条主张什么：不授予"可能"，不等于断定"不可能"；本文只说**在证明拿出来之前，那一栏是空的**，而我们一向把它当成已经填好的。 本文并指出一类特别的画面：关于未来的走向。这一类连折扣都提不出来，因为没有东西可以拿来比——不是暂时缺资料，是明天还没发生。所以它们**按全额挂账，一分不减**。我们对未来那种"这是能实现的"的感觉之所以那么硬，不是因为我们对未来特别有信心，而是因为那些画面从来没有被扣过一分钱。反过来，"当初如果"那一类因为有一件真事可以拿来比，反而能被扣掉几项——而正因为它被打过折，它看起来更像一本真账。**能打折的，比不能打折的更骗人。** 这里要补一处难题。有人会说：明天我确实走了左边那条路，那昨天心里那段画面，难道不算真的吗？ 不算。请把两件事分开：一件是昨天那段画面，另一件是明天实际发生的事。两者相似，可"相似"是我们事后在心里比出来的——比的两边，一边是对那段画面的回忆，一边是对明天的回忆。**应验给的是事后的一致，不是当时的真实性。** 那段画面被放出来的时候，没有任何东西使它为真；它后来进入了一套合得上的关系，而这套关系是别处造出来的，不会回填给它。 于是这本账永远结不平：它里面没有一栏叫"作废"。看破一段画面是编的，不是把折扣一路加到十成，而是把这张单子从"真实"这本账上整个拿掉。**不是多减了一笔，是少了一栏。** ## Abstract This paper concerns how we actually treat the images in our minds. The usual account is that we either believe them or we do not. The paper argues otherwise. **The default is not belief but booking at face value, minus whatever we are forced to concede.** When an image of "if I hadn't done that" arises, and no one challenges it, the whole of it is entered under "real." Only when someone objects, "but in fact you did," do we concede one difference; pressed again, we concede another. What we concede is the discount; what was never pressed out of us remains entered under "real." Three things follow. **First, the size of the discount is set not by how false the image is but by how hard we are pressed.** The discount is not discovered; it is extracted. Unpressed, nothing is deducted. **Second, the one who issues the invoice and the one who allows the deduction are the same person.** To compute how this image differs from what actually happened, two things must be compared—and both sides of the comparison are images in the mind. No outside party audits this ledger. **Third, and most importantly: fiction is not discounted truth.** The scale of discounting measures how much something differs; its zero is complete accuracy, and below it lie minor inaccuracy, largely inaccurate, almost wholly wrong. But "almost wholly wrong" is still **a wrong description of something that happened**. The image "if I hadn't done that" is not a description of anything that happened—the event it speaks of has no original at all. It is not on the scale. **Subtract along this scale forever and you arrive at "a real event grossly misdescribed," never at "there is no such event."** The units do not match. **Fourth, and the paper's main point: it is not that what remains after the false parts are struck out is true.** The image contains a city that really exists, a body I really have—but the truth of these components is their own, borrowed from elsewhere and used as scenery; borrowed truth does not accrue to this invoice. The fibres in a counterfeit note are real fibres and the ink is real ink, yet there is no such thing as the note having ninety percent of its purchasing power: its purchasing power is zero. Truth is not a property of the components but of whether the thing the whole image speaks of is so—and that thing never was. Its truth is therefore **not low but zero**. The "differences" extracted earlier are accordingly not discounts subtracted from some positive truth-value, but the scenery being identified piece by piece. **Zero was not arrived at by subtraction; it was zero from the start.** **Fifth, none of this says the images lack meaning or function.** They can keep a person awake with regret, settle a resolution, change a mind by morning—and these effects are entirely real; they really occur. But they are properties of **this activity**, not of **the thing the image speaks of**. The confusion lies in one word: "true" sometimes means apt, moving, on the mark, and sometimes means accurate to what is. The first speaks of effect, the second of truth. **These are two unrelated domains.** An image needs no truth whatever to produce its full effect—regret works not because the image is true but because it was booked as real; conversely, a wholly accurate record no one reads has zero effect. The conclusion must therefore be stated in two halves: **the truth of these images is zero; the effects of these images are entirely real.** The two do not conflict, because they are not about the same thing. **Sixth, "not true" does not amount to "possible" either.** A last retreat usually follows: "granted it is not true, but surely it is at least possible?" What must be examined is not the word "possible" but **where the entitlement was drawn from**. One **may** try to prove these images physically possible; but **until that proof is complete, one should not regard them as physically possible**. The reason no one has ever shouldered this burden is that possibility has always come free: **whatever actually happened is thereby physically possible**—it happened, so it was possible, and no proof is required. **Possibility has always ridden along with actuality.** In these images there is no actuality; the ride is over, but the habit remains. Two further points: **the absence of a prohibition is not the presence of a permission**—"violates no known law" is a remark about our notation, not about the world, and having no prohibition in hand is not the world having issued a licence; and **resemblance does not transmit possibility**—what actually happened issues a pass to itself alone, not to whatever stands nearby. Finally, what this claim asserts must be stated exactly: withholding "possible" is not asserting "impossible." The paper says only that **until the proof is produced, that column is blank**—whereas we have always treated it as already filled in. The paper further isolates a special class: images of how the future may go. For these no discount can even be proposed, since there is nothing to compare them against—not a temporary lack of records, but that tomorrow has not happened. They are therefore **booked in full, with nothing deducted**. Our sense that a future course is realizable is so firm not because we have special confidence in the future but because those images have never had a penny deducted. Conversely, "if I hadn't" images can have items deducted, since a real event stands beside them for comparison—and precisely because they have been discounted, they look more like a genuine ledger. **What can be discounted deceives more than what cannot.** The ledger therefore never balances: it has no column for "void." To see that an image is invented is not to raise the discount to a hundred percent but to remove the entry from the ledger of the real altogether. **Not one more deduction, but one fewer column.**

本文讲我们对待心里那些画面的真实方式。通常的说法是：我们要么信它，要么不信它。本文认为不是这样。 默认不是”信”，是按真实入账，再减去我们不得不承认的那部分。 一段”当初要是没那样”的画面出现时，如果没人质疑，它就整个挂在”真实”名下。有人指出”可是实际上你并没有那样啊”，我们才承认一处不同；再追问，再承认一处。承认了的那几处，是折扣；没被追问出来的部分，仍然挂在”真实”名下。 这里有三件事要说清楚。 第一，折扣的大小，不由这段画面假的程度决定，由我们被逼的程度决定。 折扣不是我们发现的，是被人抠出来的。没人抠，就一分不减。 第二，开票的和抵扣的是同一个人。 要算”这一段跟实际发生的差在哪里”，得拿两样东西来比；可比的两边都是心里的画面。这笔账没有外人来审。 第三，也是最要紧的一条：虚构不是打了折的真实。 折扣这条轴上量的是”差多少”，零点是完全属实，往下是有出入、大部分不实、几乎全错。可”几乎全错”仍然是对一件发生过的事的错误描述。而”当初我没那样”这段画面，不是对任何一件发生过的事的描述——它讲的那件事根本没有原件。所以它不在这条轴上。在这条轴上一直减，得到的是”一件真事被说得很离谱”，永远得不到”没有这件事”。 单位不对。 第四，也是全篇的落点：不是”剔除掉不真实的部分之后，剩下的是真的”。 那段画面里，那个城市确实存在，我确实有一副身体——可这些零件的真，是它们各自另有出处，是被借来当布景的；借来的真不算这张单子的真。一张假钞上的纤维是真纤维，油墨是真油墨，可没有”这张钞票有九成购买力”这回事，它的购买力是零。真实性不是画面里各个零件的属性，是整段画面所讲的那件事是不是那样；而它所讲的那件事从来没有过。所以它的真实性不是”很低”，是零。前面那些被抠出来的”不同之处”，因此也不是折扣——不是从一个正的真值里减掉的份额，而是布景被一件一件认出来的过程。零不是减出来的，一开始就是零。 第五，这不等于说这些画面没有意义或功能。 它们能让人后悔得睡不着，能让人下决心，能让人第二天改主意——这些作用一分不假，确实发生了。但那是这段活动的属性，不是它所讲的那件事的属性。混淆出在同一个”真”字上：“这话说得真”有时指贴切、打动人，有时指所述属实；前者说作用，后者说真实性。这两样是两个互不相干的领域。 一段画面不需要任何真实性就能起全部作用——后悔靠的不是它为真，是它被按真实入账了；反过来，一段完全属实却没人看的记录，作用是零。所以本文的结论要分两半说，免得被听成一半：那些画面的真实性是零；那些画面的作用是实实在在的。 这两句不冲突，因为它们说的不是同一件事。 本文并指出一类特别的画面：关于未来的走向。这一类连折扣都提不出来，因为没有东西可以拿来比——不是暂时缺资料，是明天还没发生。所以它们按全额挂账，一分不减。我们对未来那种”这是能实现的”的感觉之所以那么硬，不是因为我们对未来特别有信心，而是因为那些画面从来没有被扣过一分钱。反过来，“当初如果”那一类因为有一件真事可以拿来比，反而能被扣掉几项——而正因为它被打过折，它看起来更像一本真账。能打折的，比不能打折的更骗人。 这里要补一处难题。有人会说：明天我确实走了左边那条路，那昨天心里那段画面，难道不算真的吗？ 不算。请把两件事分开：一件是昨天那段画面，另一件是明天实际发生的事。两者相似，可”相似”是我们事后在心里比出来的——比的两边，一边是对那段画面的回忆，一边是对明天的回忆。应验给的是事后的一致，不是当时的真实性。 那段画面被放出来的时候，没有任何东西使它为真；它后来进入了一套合得上的关系，而这套关系是别处造出来的，不会回填给它。 于是这本账永远结不平：它里面没有一栏叫”作废”。看破一段画面是编的，不是把折扣一路加到十成，而是把这张单子从”真实”这本账上整个拿掉。不是多减了一笔，是少了一栏。 Abstract This paper concerns how we actually treat the images in our minds. The usual account is that we either believe them or we do not. The paper argues otherwise. The default is not belief but booking at face value, minus whatever we are forced to concede. When an image of “if I hadn’t done that” arises, and no one challenges it, the whole of it is entered under “real.” Only when someone objects, “but in fact you did,” do we concede one difference; pressed again, we concede another. What we concede is the discount; what was never pressed out of us remains entered under “real.” Three things follow. First, the size of the discount is set not by how false the image is but by how hard we are pressed. The discount is not discovered; it is extracted. Unpressed, nothing is deducted. Second, the one who issues the invoice and the one who allows the deduction are the same person. To compute how this image differs from what actually happened, two things must be compared—and both sides of the comparison are images in the mind. No outside party audits this ledger. Third, and most importantly: fiction is not discounted truth. The scale of discounting measures how much something differs; its zero is complete accuracy, and below it lie minor inaccuracy, largely inaccurate, almost wholly wrong. But “almost wholly wrong” is still a wrong description of something that happened. The image “if I hadn’t done that” is not a description of anything that happened—the event it speaks of has no original at all. It is not on the scale. Subtract along this scale forever and you arrive at “a real event grossly misdescribed,” never at “there is no such event.” The units do not match. Fourth, and the paper’s main point: it is not that what remains after the false parts are struck out is true. The image contains a city that really exists, a body I really have—but the truth of these components is their own, borrowed from elsewhere and used as scenery; borrowed truth does not accrue to this invoice. The fibres in a counterfeit note are real fibres and the ink is real ink, yet there is no such thing as the note having ninety percent of its purchasing power: its purchasing power is zero. Truth is not a property of the components but of whether the thing the whole image speaks of is so—and that thing never was. Its truth is therefore not low but zero. The “differences” extracted earlier are accordingly not discounts subtracted from some positive truth-value, but the scenery being identified piece by piece. Zero was not arrived at by subtraction; it was zero from the start. Fifth, none of this says the images lack meaning or function. They can keep a person awake with regret, settle a resolution, change a mind by morning—and these effects are entirely real; they really occur. But they are properties of this activity, not of the thing the image speaks of. The confusion lies in one word: “true” sometimes means apt, moving, on the mark, and sometimes means accurate to what is. The first speaks of effect, the second of truth. These are two unrelated domains. An image needs no truth whatever to produce its full effect—regret works not because the image is true but because it was booked as real; conversely, a wholly accurate record no one reads has zero effect. The conclusion must therefore be stated in two halves: the truth of these images is zero; the effects of these images are entirely real. The two do not conflict, because they are not about the same thing. The paper further isolates a special class: images of how the future may go. For these no discount can even be proposed, since there is nothing to compare them against—not a temporary lack of records, but that tomorrow has not happened. They are therefore booked in full, with nothing deducted. Our sense that a future course is realizable is so firm not because we have special confidence in the future but because those images have never had a penny deducted. Conversely, “if I hadn’t” images can have items deducted, since a real event stands beside them for comparison—and precisely because they have been discounted, they look more like a genuine ledger. What can be discounted deceives more than what cannot. The ledger therefore never balances: it has no column for “void.” To see that an image is invented is not to raise the discount to a hundred percent but to remove the entry from the ledger of the real altogether. Not one more deduction, but one fewer column.

本文讲我们对待心里那些画面的真实方式。通常的说法是：我们要么信它，要么不信它。本文认为不是这样。 默认不是”信”，是按真实入账，再减去我们不得不承认的那部分。 一段”当初要是没那样”的画面出现时，如果没人质疑，它就整个挂在”真实”名下。有人指出”可是实际上你并没有那样啊”，我们才承认一处不同；再追问，再承认一处。承认了的那几处，是折扣；没被追问出来的部分，仍然挂在”真实”名下。 这里有三件事要说清楚。 第一，折扣的大小，不由这段画面假的程度决定，由我们被逼的程度决定。 折扣不是我们发现的，是被人抠出来的。没人抠，就一分不减。 第二，开票的和抵扣的是同一个人。 要算”这一段跟实际发生的差在哪里”，得拿两样东西来比；可比的两边都是心里的画面。这笔账没有外人来审。 第三，也是最要紧的一条：虚构不是打了折的真实。 折扣这条轴上量的是”差多少”，零点是完全属实，往下是有出入、大部分不实、几乎全错。可”几乎全错”仍然是对一件发生过的事的错误描述。而”当初我没那样”这段画面，不是对任何一件发生过的事的描述——它讲的那件事根本没有原件。所以它不在这条轴上。在这条轴上一直减，得到的是”一件真事被说得很离谱”，永远得不到”没有这件事”。 单位不对。 本文并指出一类特别的画面：关于未来的走向。这一类连折扣都提不出来，因为没有东西可以拿来比——不是暂时缺资料，是明天还没发生。所以它们按全额挂账，一分不减。我们对未来那种”这是能实现的”的感觉之所以那么硬，不是因为我们对未来特别有信心，而是因为那些画面从来没有被扣过一分钱。反过来，“当初如果”那一类因为有一件真事可以拿来比，反而能被扣掉几项——而正因为它被打过折，它看起来更像一本真账。能打折的，比不能打折的更骗人。 于是这本账永远结不平：它里面没有一栏叫”作废”。看破一段画面是编的，不是把折扣一路加到十成，而是把这张单子从”真实”这本账上整个拿掉。不是多减了一笔，是少了一栏。 Abstract This paper concerns how we actually treat the images in our minds. The usual account is that we either believe them or we do not. The paper argues otherwise. The default is not belief but booking at face value, minus whatever we are forced to concede. When an image of “if I hadn’t done that” arises, and no one challenges it, the whole of it is entered under “real.” Only when someone objects, “but in fact you did,” do we concede one difference; pressed again, we concede another. What we concede is the discount; what was never pressed out of us remains entered under “real.” Three things follow. First, the size of the discount is set not by how false the image is but by how hard we are pressed. The discount is not discovered; it is extracted. Unpressed, nothing is deducted. Second, the one who issues the invoice and the one who allows the deduction are the same person. To compute how this image differs from what actually happened, two things must be compared—and both sides of the comparison are images in the mind. No outside party audits this ledger. Third, and most importantly: fiction is not discounted truth. The scale of discounting measures how much something differs; its zero is complete accuracy, and below it lie minor inaccuracy, largely inaccurate, almost wholly wrong. But “almost wholly wrong” is still a wrong description of something that happened. The image “if I hadn’t done that” is not a description of anything that happened—the event it speaks of has no original at all. It is not on the scale. Subtract along this scale forever and you arrive at “a real event grossly misdescribed,” never at “there is no such event.” The units do not match. The paper further isolates a special class: images of how the future may go. For these no discount can even be proposed, since there is nothing to compare them against—not a temporary lack of records, but that tomorrow has not happened. They are therefore booked in full, with nothing deducted. Our sense that a future course is realizable is so firm not because we have special confidence in the future but because those images have never had a penny deducted. Conversely, “if I hadn’t” images can have items deducted, since a real event stands beside them for comparison—and precisely because they have been discounted, they look more like a genuine ledger. What can be discounted deceives more than what cannot. The ledger therefore never balances: it has no column for “void.” To see that an image is invented is not to raise the discount to a hundred percent but to remove the entry from the ledger of the real altogether. Not one more deduction, but one fewer column.

本文承接《故事B》与《三维的眼》，追到一个此前两文都停在其外的环节。《故事B》指出：要理解”甲因祸得福”，须自费建造一个永不可查证的世界；《三维的眼》指出：岔路的Y形是空间中的形状，选项一个也没有。两文都保留了一侧未动——“事实”那一侧，仿佛它是我们脚下的实地。本文要说：那一侧也是视频。 论证如下。关于过去的反事实，是我们能用心灵之眼看到的，就像看视频一样；关于未来的走向，同样是我们能用心灵之眼看到的，不论我们看到了几种走向、几个视频。而关于过去的事实本身，其实也是我们能用心灵之眼看到的，就像看视频一样。三者由同一台放映机放出，同样清晰，同样连贯，同样带着”这是真的”那种笃定；而我们把这些视频信以为真。 本文指出”记忆是取回过去”这一说法不能成立：若被取回的真是过去，那它就得是1947年那一段时空本身，而1947年那一段时空并没有被搬进任何人的头里。头里发生的，是此刻的一段活动，其形式与放映一段虚构无异。过去不在视频里，正如反事实不在视频里。 本文并指出妄想这一例的位置。若有妄想的话，视频内容就包含了妄想；而视频本身不会因此显得不同——它照样清晰、照样连贯、照样带着”这是发生过的”那种笃定。这一点通常被当作病理的特例，本文要说的相反：它是一枚显影剂，显出的是正常情况下的那份笃定，同样不是从视频内部挣来的。一段真的记忆之为真，不因为它里面有什么；它里面并没有一枚”真”的印章。 于是我们从形式上难以识别关于过去的事实与关于过去的反事实的不同，因两者都是视频而已。本文把”难以”推到”不可能”：区分所需的那样东西根本不在视频里。真与假不是视频的性质，是视频与视频之外某物的关系；而那个关系，从放映厅里永远看不到。 本文据此对前两文各作一处必要的修正。对《故事B》：A可查证、B永不可查证，这份不对称仍然成立，但须挪位——它在世界那一侧（1947年那笔钱留下了账目、人证、文件；乙还钱那件事没留下任何东西），不在看的那一侧（两段视频一样清晰、一样笃定）；而日常里我们几乎从不走出放映厅去核对，说”1947年甲把钱交给了乙”时，做的是放一段视频，不是查一本账。地位不对称，消费方式相同。 对《三维的眼》：该文第七节说”往过去的方向看，只看到一条已经发生的唯一的四维轨迹”，这里的”看到”须收回——我们往过去看，看到的也是一段视频，不是那条线；那条线唯一，这一点不来自看，也不需要看来支持。此修正不削弱该文，反而让它更稳：该文说定与开的不对称不在线上而在看上，现在可再进一程——在看这一侧，连不对称也没有：往过去是视频，往未来也是视频；差别只在于关于过去的那些视频，原则上还有个外部的东西可以去碰一碰，而关于未来的与关于反事实的，连碰的对象都没有。 本文守数条须格外郑重划定的界线：这不是”记忆都不可信”的怀疑论，本文不主张我们关于过去的说法多半是错的，只主张真假不由视频内部裁定，查证仍然有效——只是查证发生在放映厅之外；这更不主张妄想与事实等价，恰恰相反，正因为真假是视频之外的关系，视频之内的笃定就不足以作准，而这一条对妄想成立，对我们自己那些从未核对过的确信同样成立；至于记忆、证词、历史该如何在此之后被重新安放，本文不下判决，明确悬置为尚待研究。本文这一篇，自己也是一段视频；它没有让谁走出放映厅，它只是在放映厅的墙上，写了”放映厅”三个字。指可弃，月不可弃。 Abstract This paper continues Story B and The Three-Dimensional Eye, reaching a step at which both earlier papers stopped short. Story B argued that to understand “a blessing in disguise” one must build, at one’s own expense, a permanently unverifiable world; The Three-Dimensional Eye argued that the Y-shape of a fork is a shape in space and that there are no options at all. Both left one side untouched—the side of “fact,” as though it were solid ground beneath us. This paper argues that that side is a video too. The argument runs as follows. Counterfactuals about the past are things we can see with the mind’s eye, just as we watch a video; the courses the future might take are likewise things we can see with the mind’s eye, however many courses, however many videos. And the facts of the past themselves are also things we can see with the mind’s eye, just as we watch a video. All three are played by the same projector, equally vivid, equally coherent, equally attended by the conviction that this is real—and we take these videos to be real. The paper argues that “memory retrieves the past” cannot stand: if what were retrieved really were the past, it would have to be that stretch of spacetime in 1947 itself, and that stretch was never carried into anyone’s head. What occurs in the head is an episode occurring now, formally indistinguishable from the screening of a fiction. The past is not in the video, any more than the counterfactual is. The paper then places the case of delusion. Where delusion is present, the content of the video contains the delusion; yet the video does not thereby look any different—it remains as vivid, as coherent, as attended by the conviction that this happened. This is usually treated as a pathological special case; the paper argues the reverse: it is a developing agent, and what it develops is that in the ordinary case the conviction is not earned from inside the video either. A true memory is not true because of anything in it; it carries no stamp reading “true.” Hence we find it formally difficult to tell facts about the past from counterfactuals about the past, since both are merely videos. The paper presses “difficult” to “impossible”: what the distinction requires is simply not in the video. True and false are not properties of videos but relations between a video and something outside it—and that relation is never visible from inside the screening room. On this basis the paper offers a necessary correction to each earlier paper. To Story B: the asymmetry whereby A can be verified and B never can still holds, but must be relocated—it lies on the side of the world (the money of 1947 left ledgers, witnesses, documents; B’s returning the money left nothing), not on the side of the looking (the two videos are equally vivid, equally confident); and in ordinary life we almost never leave the screening room to check, so that in saying “in 1947 A gave the money to B” we are playing a video, not consulting a ledger. The standing is asymmetrical; the consumption is identical. To The Three-Dimensional Eye: its seventh section says that looking toward the past one sees a single unique four-dimensional trajectory already occurred—here “sees” must be withdrawn. Looking toward the past we too see a video, not the line; the line’s uniqueness does not come from the looking and needs no looking to support it. This correction does not weaken that paper but steadies it: it held that the asymmetry between fixed and open lies not in the line but in the looking; now one may go further—on the side of the looking there is no asymmetry either. Toward the past, a video; toward the future, a video. The only difference is that for videos about the past there is in principle something external one might go and touch, whereas for videos of futures and of counterfactuals there is nothing to touch at all. The paper guards several lines with particular gravity: this is not skepticism about memory—it does not claim that our accounts of the past are mostly wrong, only that truth is not adjudicated from inside the video, and verification remains effective, but takes place outside the screening room; still less does it hold delusion and fact to be equivalent—on the contrary, precisely because truth is a relation outside the video, conviction inside the video is insufficient warrant, and this holds for delusion and equally for those convictions of our own we have never checked; and how memory, testimony, and history ought to be re-situated thereafter, the paper renders no verdict, leaving it expressly still under study. This very paper is itself a video; it has let no one out of the screening room—it has only written the words “screening room” on the wall. The finger may be discarded, the moon may not.

本文承接《故事B》，追问一个更前置的环节：当我们比较故事A与故事B，并发表评论”乙可以把钱还给甲”时，这个”可以”由什么支撑。可供选择的只有两个，而两个都退票：其一，故事B已经真实发生过——那样它就不是”可以”而是”已经”，且《故事B》已表明B原则上永不可查证；其二，世界的物理结构容许B出现——可从来没有人在形式上成功地证明过这一点。我们所拥有的只是一组含自由参数的方程，而多解性住在我们的写法里，不住在世界里：世界只走了一个解，它不曾走过一族解，也不曾在一族解中挑过。方程的一般性本是我们不知道是哪一组值的记号，我们却把它读回世界，当成世界自己的复数。这是一张以形式证明为担保而从未兑付的欠条。 我们之所以从不索要这张欠条，是因为我们不是推出了它，是看见了它。我们感觉看到了，就是一个option：我们看到脚下的道路在前方分成两个岔，看到这两个岔，就感觉这两个岔似乎是两个option，而option给我们一种可实现感。本文指出这里有一次范畴错置：岔路的Y形是一件此刻摆在空间中的东西，不是一份未来的清单；在四维中，属于我的那个Y根本不存在，我只是一条不分岔的曲线，分岔的是路而不是我，而”路分岔了”是一件关于沥青的事实。 本文并对这一洞见作一处必要的推进：不只是另一个option是假的，而是剩下的那一个也不是option。说我的轨迹所经过的那条岔道”真正可实现”，只是重复说它实现了；而”选项”是一个要求复数的词——一份只有一项的清单不是清单，一个没有替代者的选项不是选项。于是两边都塌：假的那个是空间形状的误读，真的那个是同语反复。选项一个也没有。 本文另指出：未走的那一支，在四维中不过是我的轨迹从未经过的一片区域，而这样的区域数量无穷——仙女座星系是一片，三世纪的洛阳是一片。它凭什么被叫做option而仙女座不是？只凭空间上离我近、时间上排我前面。邻近不是可能，近不是能；所谓选项，不是世界中的一类对象，是一种距离关系被读成了模态地位。本文并指出唯一性不因看的方向而改变：往过去看是一条唯一的四维轨迹，往未来看不论看到几个option，都并不代表真实的未来轨迹是这几个；并不是只往过去看才唯一，而往未来看不唯一——往未来看不唯一的，是option而已。定与开的不对称不在线上，在看上；我们把”不知道是哪一条”换算成了”有好几条”，而这一步换算从未获得授权。 本文并指出”看到”是借来的动词，而借来处并没有可看的东西：你现在已经在头脑中看到了前方的道路分岔，虽然你的眼前真正看到的只是屏幕上的这篇论文——那里没有沥青，没有光子，没有对象。这不是一次对象缺席的观看，是根本没有观看；而”看到了什么”这个句式一用上，那个”什么”就凭空取得了地位。甲在把钱交出去之前”看到”乙还钱，1950年甲”看到”自己当上盐场老板——这两幅都没有出现在四维时空中。我们却给这种看图改了名字，叫它分析、推理、考虑各种可能，一改名它就取得了论证的身份。其实我们只是在看，只是在虚构地看，只是在看虚构。 本文最后指出：选项被端上来，是因为主角不能出场——我们无法预测自己的坐标在四维时空中的未来轨迹，因而无法谈论它，只好转而谈论眼前在空间中所看到的，似乎那就是未来；替身上台太久，我们忘了主角从未出场。同一层遮蔽也向后蔓延：我们看到四维中真实的过去轨迹时，也没意识到其唯一真实性，而是立刻读成”诸多可能中实现了的那一个”——这个说法已预设了未实现的其余，而其余从未被出示过。未来被预先复数化，过去被追认复数化，唯一的那条线从未被当成唯一。我们缺少四维的眼；但补上四维的眼也不会看到选项，它只会看到那条线，而”选项”一词到那时将无所指——这才是”看错”的分量：错的不是看漏了几个选项，是压根没有选项可漏。 本文守数条须格外郑重划定的界线：这一诊断不否认选项语言在决策、计划、制度中的实际功用；它不是”未来已被写定因而可以被知道”的预测神话——不可知与不唯一是两回事，本文所主张的正是不可知而唯一；至于审议、责任、后悔该如何在此之后被重新安放，本文不下判决，明确悬置为尚待研究。本文这一篇，自己也是一幅被心眼看见的图；它是指月之指；指可弃，月不可弃。 Abstract This paper takes up a question prior to the one in Story B: when we compare story A with story B and say “B could have returned the money to A,” what supports that “could”? Only two supports are available, and both fail. First, that story B actually occurred—but then it would not be “could” but “did,” and Story B has shown B to be in principle permanently unverifiable. Second, that the physical structure of the world permits B—yet no one has ever formally demonstrated this. What we possess is a set of equations with free parameters, and the plurality of solutions lives in our notation, not in the world: the world took one solution; it never traversed a family of solutions, and never picked among them. The generality of the formalism is a token of our not knowing which values obtain, and we read that generality back into the world as the world’s own plurality. This is an IOU underwritten by a formal proof that was never delivered. We never call in the IOU because we did not infer the possibility—we saw it. What we feel we see, we take to be an option: we see the road ahead divide in two, and seeing the two branches we feel that the two branches are two options; options give us a feel of realizability. Here lies a category mistake: the Y-shape of the fork is a thing standing now in space, not a menu of futures. In four dimensions there is no Y belonging to me; I am a single curve that does not fork. The road forks, I do not—and that the road forks is a fact about asphalt. The paper presses the insight one step further: it is not only that the other option is counterfeit, but that the remaining one is not an option either. To say that the branch my trajectory passes through is “genuinely realizable” is only to repeat that it was realized; and “option” is a word requiring plurality—a list with one item is not a list, and an option with no alternative is not an option. Both sides thus collapse: the false one is a misreading of a spatial shape, the true one a tautology. There are no options at all. The paper further notes that the branch not taken is, in four dimensions, merely a region my trajectory never enters—and such regions are infinite in number: the Andromeda galaxy is one, third-century Luoyang is another. What entitles one of them to be called an option while Andromeda is not? Only spatial nearness and temporal priority. Nearness is not possibility; being near is not being able. An option is not a kind of object in the world but a distance relation misread as a modal status. Uniqueness, moreover, does not change with the direction of looking: toward the past one sees a single unique four-dimensional trajectory; toward the future, however many options one sees, they do not constitute the real future trajectory. It is not that looking backward yields uniqueness while looking forward does not—what is non-unique when we look forward is only the options. The asymmetry between fixed and open lies not in the line but in the looking; we have converted “not knowing which one” into “there being several,” and this conversion was never licensed. The paper also argues that “to see” is a borrowed verb, and that where it is borrowed to, there is nothing to see. You have already seen, in your mind, the road ahead dividing, though what your eyes actually see is this paper on a screen—there is no asphalt there, no photons, no object. This is not a seeing whose object is absent; it is no seeing at all, and once the construction “what did you see?” is applied, the “what” acquires standing out of nothing. Before handing over the money, A “saw” B returning it; on learning of the 1950 policy, A “saw” himself as a salt-field owner. Neither ever appeared in four-dimensional spacetime. Yet we rename this looking-at-pictures analysis, inference, weighing the possibilities—and with the renaming it acquires the standing of an argument. In truth we are merely looking; looking fictively; looking at fiction. Finally: options are brought on stage because the protagonist cannot appear. We cannot predict the future trajectory of our own coordinate in four-dimensional spacetime, so we cannot speak of it, and speak instead of what the eye sees in space, as if that were the future; the stand-in has held the stage so long that we forget the protagonist never came on. The same occlusion spreads backward: seeing the real past trajectory in four dimensions, we likewise fail to register its unique reality, reading it at once as “the one among many possibilities that was realized”—a phrase presupposing an unrealized remainder never exhibited. The future is pluralized in advance, the past retroactively, and the one line is never taken as one. We lack the four-dimensional eye; but the four-dimensional eye would not see options either—it would see the line, and by then the word “option” would have nothing to refer to. That is the weight of the error: not that we overlooked some options, but that there were none to overlook. The paper guards several lines with particular gravity: this diagnosis denies the language of options none of its practical utility in decision, planning, or institutions; it is not the predictive myth that the future, being fixed, is knowable—unknowability and non-uniqueness are two different things, and the position here is precisely that the line is unknowable yet unique; and how deliberation, responsibility, and regret ought to be re-situated thereafter, the paper renders no verdict, leaving it expressly still under study. This very paper is itself a picture seen by the mind’s eye; it is a finger pointing at the moon; the finger may be discarded, the moon may not.

本文追一个从未被正面处理过的环节：**"理解"一句评论，究竟需要我们做什么。** 论证从一个具体的历史情境出发：**1947年，甲把所有积蓄交给朋友乙保管；之后甲准备买下一个盐场，但乙不承认甲的钱在他处保管，吞没了甲的钱，甲没开成盐场。但1950年的政策规定盐场老板等类人群都要枪毙，甲因祸得福。** 本文指出：**当我们思考这个故事（A）时，如果我们自称理解了这个故事，则我们必须想象出一个当了盐场老板而被枪毙的甲，存在于一个与1950年的中国几乎一模一样的世界中。** 这个被想象出来的世界，就是**故事B**——而它的虚构性，对我们几乎是隐形的。**在未被质疑的情况下，我们会感觉这个被枪毙的甲所在的世界就是1950年的真实世界**；若被质疑"真实世界里甲未被枪毙啊"，我们会退一步说，这个真实世界里有一处不同，那就是甲由未被枪毙变成了被枪毙。**可再仔细想想，不同之处还应有一些，比如乙把钱还给了甲（而不是没还）。** 而这条线一旦拉开，就停不下来：乙若还了钱，乙便不是那个会吞没朋友积蓄的乙，甲当初的托付、二人的关系，全都跟着变了；甲若从1947到1950一直是盐场老板，这三年的社会位置、来往的人、家人的处境，同样全部不同——**故事B从来不是"1950年的中国加一处修改"，而是一个从1947年就开始分岔、且在除"甲被枪毙"这一点之外的所有地方完全未被定义的世界。** 本文进而指出这份虚构性何以隐形：**故事B之所以难以被识破，恰恰因为它被刻意造得尽量贴近真实世界**——我们造B时只改动一个点，其余全部沿用A的世界，而正是这份"最大程度的沿用"，让B在感觉上几乎就是那个真实世界；**相似性，本是为了让B可理解、可对照，却同时成了它虚构性的最佳掩护：越像真的，越不像是被造出来的。** 本文另指出A与B在证据地位上的绝对不对称：**故事A的事实部分可被查证**（有账目、人证、政策文件），**而故事B永远无法被查证**——不是因为暂时缺资料，是因为没发生过的事不会留下任何记录，原则上不存在任何证据能支持或反驳它；可我们说出B时的笃定语气，与陈述A时几乎无异，这份**语气上的平权**，是虚构性隐形的另一半原因。本文并对A作一处必要的修正：严格讲，A本身也已是一次筛选（从1947至1950那片无限复杂的实际发生里，只挑出了钱、盐场、政策几条线），因此准确的对照不是"真实世界对虚构故事"，而是**"一个真的故事对一个假的故事"**——A的真，是"所述属实"，不是"未经中介地直通实在"。本文最后点明代价：**"因祸得福"是一次价值涂色，而要给一段真实历史涂上这层色，我们必须先自费建造一座完整的、永不可查证的世界，作为涂色的参照系；而这座建筑物，因为造得太像真的，我们从未把它当成建筑物看待过。** 本文守数条须格外郑重划定的界线：这一诊断不否认评论性语句在生活中的实际功用；不构成"故一切评价都不成立"的裁决；至于因果、责任、评价该如何在此之后被重新安放，本文不下判决，明确悬置为**尚待研究**。本文这一篇，自己也是一个为了理解而被建造出来的故事；它是指月之指；指可弃，月不可弃。 ## Abstract This paper examines a step never directly addressed: **what "understanding" an evaluative statement actually requires of us.** The argument proceeds from a concrete historical situation: **in 1947, A entrusts his entire savings to his friend B for safekeeping; A then prepares to buy a salt field, but B denies holding A's money and embezzles it, so A never opens the salt field. Yet in 1950, policy dictates that salt-field owners and similar classes be executed—A's misfortune proves a blessing in disguise.** The paper argues: **when we consider this story (A), if we claim to have understood it, we must imagine an A who became a salt-field owner and was executed, existing in a world almost exactly like China in 1950.** This imagined world is **story B**—and its fictionality is nearly invisible to us. **Unchallenged, we feel that the world containing this executed A simply is the real world of 1950**; challenged with "but in the real world A was not executed," we retreat one step and say this real world differs in one respect: A goes from not-executed to executed. **Yet on further reflection there must be other differences too—for instance, B returned the money rather than keeping it.** And once this thread is pulled, it cannot be stopped: if B returned the money, B is not the man who would embezzle a friend's savings, so A's original entrustment and the pair's relationship both change; if A were a salt-field owner from 1947 to 1950, his social position, his associates, his family's circumstances across those three years all differ—**story B was never "China in 1950 plus one modification," but a world diverging from 1947 onward and left wholly undefined everywhere except at the single point of A's execution.** The paper then explains why this fictionality stays hidden: **story B evades detection precisely because it is deliberately built to resemble the real world as closely as possible**—we alter one point and inherit everything else from A's world, and this maximal inheritance makes B feel like the real world itself; **the resemblance, intended to make B intelligible and comparable, becomes the best cover for its fictionality: the more real it looks, the less it looks constructed.** The paper further notes an absolute asymmetry in evidential status: **story A's factual portion can be verified** (ledgers, witnesses, policy documents), **whereas story B can never be verified**—not from a temporary lack of records, but because what never happened leaves no records, so that in principle no evidence could support or refute it; yet we assert B with a confidence scarcely distinguishable from our assertion of A, and this **parity of tone** is the other half of why the fiction stays hidden. The paper also offers a necessary correction regarding A: strictly speaking, A is itself already a selection (from the infinitely complex actuality of 1947–1950, only the money, the salt field, and the policy were picked out), so the accurate contrast is not "real world versus fictional story" but **"a true story versus a false one"**—A's truth consists in accuracy of what it states, not in unmediated access to reality. The paper closes on the cost: **"a blessing in disguise" is an act of evaluative coloring, and to apply this color to a stretch of real history we must first build, at our own expense, an entire, permanently unverifiable world to serve as the reference frame for that coloring—and this edifice, built too well to look built, we have never regarded as an edifice at all.** The paper guards several lines with particular gravity: this diagnosis denies evaluative statements none of their practical utility; it constitutes no verdict that all evaluation is void; and how causation, responsibility, and evaluation ought to be re-situated thereafter, the paper renders no verdict, leaving it expressly **still under study**. This very paper is itself a story built in order to be understood; it is a finger pointing at the moon; the finger may be discarded, the moon may not.

本文立一个极简、却足以撼动一整类思考习惯的判断，并把它贯彻到底：**我们对事件动力的分析，是一种解释，解释的对象是不内带任何动力的发生本身；因此不要把对事件动力的分析当真。** 这把刀必须被彻底贯彻——**发生里，甚至没有"相互的作用"；看上去似乎存在的相互作用，是我们感觉、解释出来的。** 我们习惯说"空气被台球推开"，可**空气和台球同时改变了位置，并不等于台球推开了空气：前者是发生，后者是解释。** 同样，**这一判断并非在说"重力不存在"——重力恰恰是我们捏造出来的解释；把重力用时空弯曲替代，同样是解释：我们既看不到重力，也看不到时空弯曲。** 刀还须落到时间上：**时间的存在，是对变化的某种解释——我们其实看不到时间的存在；我们先把时间解释、想象成类似空间维度的东西，再用这种东西来替代时间。** 本文进而把这一诊断，扩展到承载一切的那个坐标框架本身：**我们看到各种物体，然后想象它们位于一个由三条相互正交的线所度量的空间／容器／盒子中，给它们这个盒子里的坐标，然后就神奇地感觉这个盒子似乎真的存在了——这个盒子并不真的存在，而只是一种解释／想象。我们再添加一条正交的线来代表"时间"，然后感觉时间就似乎真的存在了——这个四维的新盒子也并不真的存在。** 关键在于，是**先有盒子才给得出坐标，不是坐标证明了盒子**。而本文由此正面回应一个最强的反驳——"可物理公式管用啊"：**即使套用物理公式时，我们想象这个四维盒子的存在，然后物体在这四维上的坐标恰好符合我们所套用的物理公式，这时我们感觉物理公式和四维盒子同时被证明存在了。并没有！只是我们假设它俩存在后，我们的后续推理、预测，仍能自圆其说。但自圆其说的东西／系统，并不证明其中所有要素的存在性。** 剥净之后，"发生本身"只剩并置的变化，没有作用、没有力、没有作为容器的时空。本文进而处理一个实际需求：若非要谈动力，该安放在哪里？**做事的是 condition，而不是人这样的 agent——不需要看到 agent，只需要看到 condition，穿过 agent 的边界。** 这是改进而非又一次虚构，因为 agent 要求一道封闭的边界（而这边界是我们划出的），condition 则不要求，它可以穿过皮肤与系统壁，少虚构一层。本文守数条须格外郑重划定的界线：这不否认物理学的实用效力与预测能力，所拆的仅是把"规律性"读作"内在动力"、把"自洽好用"读作"要素实存"这两步；**"不要当真"不等于"不要再做"——方法照用，地位下调**；这也不构成"故一切解释都无价值"的虚无主义。至于责任、归因这类实践该如何被重新安放，本文不下判决，明确悬置为**尚待研究**。本文这一篇，自己也是一层叠加在发生之上的解释；它是指月之指；指可弃，月不可弃。 ## Abstract This paper advances a claim of extreme simplicity and carries it through without exception: **our analysis of an event's dynamics is an explanation, and what it explains is occurrence itself, which carries no intrinsic impetus; therefore, do not take analyses of event-dynamics as real.** The blade must be applied thoroughly—**within occurrence there is not even "interaction"; what appears to be interaction is something we feel and explain into being.** We habitually say "the air is pushed aside by the billiard ball," yet **the air and the ball changed position simultaneously; this does not amount to the ball pushing the air: the former is occurrence, the latter explanation.** Likewise, **this is not saying "gravity does not exist"—gravity is precisely an explanation we fabricated; replacing it with spacetime curvature is equally an explanation: we see neither gravity nor spacetime curvature.** The blade must fall on time as well: **the existence of time is a certain explanation of change—we do not actually see time existing; we first explain and imagine time as something akin to a spatial dimension, then substitute that for time.** The paper then extends this diagnosis to the very coordinate framework that carries everything: **we see various objects, then imagine them located within a space/container/box measured by three mutually orthogonal lines, assign them coordinates within this box, and then marvelously feel that this box seems really to exist—yet the box does not really exist; it is merely an explanation, an imagining. We then add a fourth orthogonal line to represent "time," and feel that time too seems really to exist—yet this new four-dimensional box does not really exist either.** The crux is that **the box comes first and makes coordinates possible; coordinates do not prove the box.** The paper thereby confronts the strongest rebuttal—"but the physics works": **even when applying physical formulae, we imagine this four-dimensional box's existence, and when objects' coordinates along these four dimensions happen to satisfy the formulae we apply, we feel that both the formulae and the box have been proven to exist. They have not! It only shows that, having assumed both exist, our subsequent reasoning and predictions remain self-consistent. But a self-consistent thing or system does not prove the existence of all its constituent elements.** Once stripped, "occurrence itself" retains only juxtaposed change—no action, no force, no spacetime as container. The paper then addresses a practical need: if one must speak of dynamics, where to locate it? **What does things is condition, not an agent such as a person—one need not see the agent, only the condition, crossing the agent's boundary.** This is an improvement rather than one more fabrication, because agent demands a closed boundary (drawn by us) while condition demands none, crossing skin and system walls alike, fabricating one layer less. The paper guards several lines with particular gravity: this denies neither physics' practical efficacy nor its predictive power—what it dismantles is solely reading "regularity" as "intrinsic impetus," and reading "self-consistent and useful" as "its elements exist"; **"do not take it as real" does not mean "do not do it"—retain the method, demote its status**; nor does this constitute nihilism about explanation. As to how practices of responsibility and attribution ought to be re-situated, the paper renders no verdict, leaving it expressly **still under study**. This very paper is itself a layer of explanation superimposed upon occurrence; it is a finger pointing at the moon; the finger may be discarded, the moon may not.

本文提出对话的一项隐秘风险：当听到的话让人觉得有道理时，这句话便在那一刻获得与自己的内心独白同等的可靠感，被当作自己的思想去调用与承重；然而它是相对陌生的思想，未经自己的测试史检验，驾驭不了它的人会摔跤——如在不熟悉的地形上行走，无人行骗，误导照旧发生。自己的思想是熟的，别人的思想是生的，但我们不感觉它像生的：认知系统登记的是整合的顺滑度，而非来源与测试史。与人工智能的对话把这一风险推至极端：人工智能没有骗人，它的话语甚至比你自己的话语更像你的——它是减去了摩擦的你的声音——归化手续在抵达的瞬间已经办完；其风险的正确模型不是毒物而是换水：整缸混合水可以让鱼生病，纵然新水无毒，只因换水速率超过了代谢速率。本文进而给出时代诊断：思想的输入速率与消化速率之比已改变数量级，而消化端由生理定速——这是一个前所未有的、思想没有时间熟化的时代；我们由此重返婴儿期，必须像婴儿学习使用四肢那样，靠误差信号去适应大脑里新增的内容。文末给出思想检疫的操作方法，并指明熟化的本体论地位：消化不是自我在吸收思想，消化过程本身在铸造下一个自我；困水期不是等待成熟，困水期就是成熟。 Abstract This paper identifies a hidden risk of dialogue: the moment an utterance feels plausible, it acquires the same felt reliability as one’s own inner monologue and is put to use as one’s own thought. Yet it is a relatively foreign thought, unverified by one’s own history of trials; one may prove unable to handle it and may stumble—as one stumbles on unfamiliar terrain—misdirection without any deceiver. One’s own thoughts are cooked; others’ thoughts are raw; but the raw does not feel raw, because cognition registers smoothness of integration rather than provenance or testing history. Dialogue with artificial intelligence pushes this risk to its extreme: the AI does not deceive; its discourse is even more like you than your own—your voice minus friction—so that naturalization is complete upon arrival. The correct model of the risk is not poison but water change: mixed water can sicken the fish even when the new water is clean, because the rate of change exceeds the rate of metabolism. The paper then diagnoses the era: the ratio of input rate to digestion rate has shifted by orders of magnitude, while digestion is physiologically rate-fixed—an unprecedented era in which thought has no time to cure. We have thereby returned to infancy: we must adapt to the new contents of our brains by error signals, as infants learn the use of their limbs. The paper closes with operational methods of quarantine and with the ontological status of curing: digestion is not a self absorbing thoughts; the process of digestion is what forges the next self. The quarantine period is not a wait for maturity; the quarantine period is maturation itself.

本文是《许可的妄想》的姊妹篇，为其补足前提。前篇论证系统化哲学与临床妄想同构、仅隔一纸文化许可；本文追问哲学是从什么之中提纯而来，以及提纯意味着放弃了什么。本文提出：常人的思想不是一个系统，而是一堆互不咬合、彼此牵制的妄想的杂多；它之所以不崩坏，恰因其不成系统——内部矛盾之多，使任何单一读数都无法统摄全局，诸味相互抵消，故无一味走向极端。系统化哲学则是从这团杂多之中，抓住一条读数、剔除其上全部矛盾配重、令其贯彻到底的产物。此即提纯：原则愈纯，而那些使杂多得以安稳的相互牵制愈被弃尽。由此，纯度成为贯穿三者的唯一变量：临床妄想、系统化哲学、常人思想，依”单一原则统摄全局的纯度”由高至低排列。纯度既是”深刻”的度量，亦是”逼近病理”的度量——二者本是同一根标尺。常人之”清醒”不来自看清，而来自不纯；哲学家为深刻所付之代价，是自愿将自己推向谱系中纯净而危险的一端。 Abstract This paper is the sister to “Licensed Delusion,” supplying its missing premise. The former argues that systematic philosophy and clinical delusion are isomorphic, divided only by a sheet of cultural license; this paper asks out of what philosophy is refined, and what refinement entails relinquishing. It proposes that ordinary thought is not a system but a manifold of mutually incompatible, mutually restraining delusions; it does not collapse precisely because it is not systematic — the abundance of internal contradictions prevents any single reading from subsuming the whole, so the many flavors cancel one another and none proceeds to an extreme. Systematic philosophy is the product of seizing one reading out of this manifold, stripping from it every contradictory counterweight, and carrying it through to the end. This is refinement: the purer the principle, the more thoroughly discarded are the mutual restraints that had kept the manifold stable. Purity thus becomes the single variable running through all three: clinical delusion, systematic philosophy, and ordinary thought arrange from high to low along “the purity with which a single principle subsumes the whole.” Purity is at once the measure of “profundity” and the measure of “proximity to pathology” — the two are one and the same rod. The “lucidity” of ordinary people comes not from seeing clearly but from impurity; the price the philosopher pays for depth is to push himself, willingly, toward the pure and dangerous end of the spectrum.

本文提出：系统化哲学与临床妄想是同一种活动。二者同以单一原则统摄万象，同对反例施行辩护性吸收，其持有者同样终生无法自其体系中抽身。区分二者的，既非结构，亦非真假，而仅是一纸文化许可：哲学落在现代文化的协调均衡之内，遂领得"思想"之名；另一种落在均衡之外，遂领得"病"之名。本文据此论证，"病非病"一语的真义不是为病脱罪，而是揭示——被尊为哲学者与被诊断为妄想者，乃同一种心灵活动，只隔着一层文化的命名。文章依次剖析妄想的三项结构特征如何在哲学中逐一重现，说明"可共享性"何以是唯一的分界线而非健康的标志，并指出既有的两道辩护（体系可被主动修改、哲学家生活未受损）如何本身即许可机制的运作。 Abstract This paper proposes that systematic philosophy and clinical delusion are one and the same activity. Both subsume all phenomena under a single principle, both practice the defensive absorption of counter-evidence, and the holders of both are alike unable, throughout their lives, to step outside their system. What separates the two is neither structure nor truth but a mere cultural license: philosophy falls within the coordination equilibrium of modern culture and thereby earns the name "thought," while the other falls outside that equilibrium and earns the name "illness." On this basis the paper argues that the true import of the phrase "sickness is not sickness" is not to acquit illness but to reveal that what we revere as philosophy and what we diagnose as delusion are one and the same activity of mind, divided only by a layer of cultural naming. The paper shows in turn how the three structural features of delusion each recur within philosophy, why "shareability" is the sole line of division rather than a mark of health, and how the two standard defenses (that a system can be actively revised; that the philosopher's life is unimpaired) are themselves the operation of the licensing mechanism.

本文合并七记洞见，把"自然延续"这一概念，从一个看似客观独立的标准，逐层还原成它真正的身份——**我们对自己预测的一次事后追认**，并进一步揭示，这份预测所依赖的方法，正是我们做一切计划、谋划、乃至后悔与犹豫时，共享的同一套通用机制：**功能卡推演**。论证从一个精确的物理例子起步：台球沿斜坡滚下，是最接近普适物理定律的可靠知识，可我们不会说小球本该在斜坡上静止不动——与之相对，"某人的一句话对另一人的功能"这类描述，远不如物理因果可靠，却被感知成同等可靠，且唯有后者会滑向"本该不同"这一步判语。本文追问这份差异的来源：**斜坡台球之所以不产生自然延续本应不同的感觉，是因为物理状态设定简单，我们难以作出错误的预测**；一旦情形复杂——乌云翻滚——我们便可能预测暴雨，若不下，便觉得现实"偏离"了自然延续；**可其实不下暴雨，只是偏离了错误的预测。所谓"自然延续"，只是我们做出的预测，无论对错，都被感知为自然延续本身**——这一诊断，收在一个归谬上：若真准确预测到自然延续，现实又怎能偏离它？本文进而追问，为什么预测出错时，我们责怪的是一个"外来干扰"，而非自己：**狂风没有给我们线索，我们便把"没有线索因此没有预测到"，错读成了"位于待预测的系统之外"——可没有线索，只说明认知失败，不构成关于世界结构的证据。** 本文由此揭示我们实际预测的机制：**这种所谓的预测，其实只是我们心中根据模块的"功能"而做的推演，比如"翻滚的乌云有产生暴雨的功能"；狂风这个模块，虽具有"吹散乌云"的功能，但此时不在我们的视野之内，因此没有进入我们的推演，于是被感知为似乎位于系统之外，实际只是位于推演之外。恰如在我们的推演中，关公可以战秦琼，因为两者都被想象为具有某些武力功能的模块，因此可以推演其相互作用——其实就是想象这些功能是如何被使用的。** 本文进而把"系统边界"这个概念，逼到彻底的悖论——若系统是整个宇宙，没有狂风的系统在物理上是个悖论；若系统缩小到一个小范围，小系统的自然延续又如何定义？不考虑边界外情况的所谓"预测"，与预测关公战秦琼的胜负，又有何不同？关公战不了秦琼，小系统也并非独立。本文由此指出，这套功能卡推演，**也被我们用来在格斗中思考自己的招式，在与他人的非暴力对抗中思考自己的对策，在旅游时计划明天的活动安排，在团队中规划任务的分配——这是我们做任何计划、谋划的基础，因此它显得如此自然，似乎在直接处理现实本身，而非只是处理自己头脑之中的模块。可是，只有在思考台球滚下斜坡这样的简单配置时，头脑之中的模块才能与现实的物理演变严丝合缝；这也解释了为什么我们对复杂配置的推演总不能与现实吻合——这并不说明现实脱离了自然延续，而是说明我们的推演脱离、误判了自然延续，且在复杂配置下，无论采用多么精妙的思维模块组合，也不可能吻合现实的物理演变，因为我们没有采用拉普拉斯妖式的计算方式。** 本文最后把这套机制，用在两种最贴身的心理体验上：**当我们为过去感到后悔时，我们是在用功能卡构造出了另一种（更优的）可能性；当我们思考未来的走向时，用功能卡构造出一种可能性后，又构造出另一种，这时未来就似乎有两个分岔了——分别指这两种可能性。** 两者共享同一个错误：把推演的次数，错当成了世界本身的分支数。本文守数条须格外郑重划定的界线：这一诊断不否认物理因果的真实可靠性，也不否认功能卡推演在日常计划与沟通中的实践价值；不构成"故一切归因与规划都不成立"的裁决；至于"本该如此"这类判语在责任实践中该如何被重新安放，本文不下判决，明确悬置为**尚待研究**。本文这一篇，自己也是一次此刻用功能卡拼出的推演，一场关公与秦琼的对局；它是指月之指；指可弃，月不可弃。 ## Abstract This paper merges seven insights, progressively reducing "natural continuation" to its true identity—**a retroactive medal we award our own prediction**—while revealing that the method underlying such prediction is exactly the shared, general-purpose mechanism we use in all planning, scheming, regret, and hesitation: **function-card reasoning.** The argument begins from a precise physical example: a billiard ball rolling down a slope is the most reliable knowledge, closest to universal physical law, yet we never say the ball should have remained at rest—by contrast, descriptions of "one utterance's function toward another person" fall far short of this reliability, yet are perceived as equally reliable, and only the latter slides into the verdict "should have been otherwise." The paper asks where this difference originates: **the billiard-ball case produces no such feeling because the physical setup is simple, making an erroneous prediction hard to form**; once matters grow complex—churning clouds—we may predict rain, and if none comes, feel reality has "deviated"; **but the absence of rain only deviates from a mistaken prediction. The so-called "natural continuation" is nothing but the prediction we made, perceived as such whether right or wrong**—landing on a reductio: had we truly predicted accurately, how could reality deviate? The paper then asks why, when prediction fails, we blame an "external disruption" rather than ourselves: **the gale gave us no clue, so we misread "unpredicted for lack of a clue" as "outside the system"—yet lacking a clue only reveals cognitive failure, no evidence about the world's structure.** The paper thereby exposes the actual mechanism of prediction: **this so-called prediction is merely a deliberation we run in our minds based on each module's "function"—"churning clouds have the function of producing rain"; the gale, though possessing the function of "dispersing clouds," was not within our field of view and so never entered our deliberation, hence felt to lie outside the system when it in fact lay only outside the deliberation. This is exactly like our deliberation imagining Guan Yu fighting Qin Qiong—both imagined as modules bearing certain martial functions, their interaction deliberated—which is simply imagining how these functions would be deployed.** The paper pushes "system boundary" to a full paradox—if the system is the universe, a windless system is a physical paradox; if shrunk to a small region, how is its natural continuation even defined? How does a "prediction" ignoring its boundary differ from predicting a duel between Guan Yu and Qin Qiong, who could never have fought? The paper then shows this function-card reasoning **also underlies how we think through our moves in combat, our tactics in non-violent contests, our itinerary while traveling, our allocation of tasks within a team—the very basis of all planning, and precisely because of this ubiquity it feels as though it directly handles reality itself, rather than merely the modules in our own heads. Yet only in simple configurations like the billiard ball does what is in our head align exactly with physical evolution; this explains why our deliberation over complex configurations never quite matches reality—not because reality deviated from its natural continuation, but because our deliberation deviated from, misjudged, that continuation; and in complex configurations, no matter how sophisticated the combination of thinking-modules we employ, it can never match physical evolution, because we have not adopted a Laplace's-demon style of computation.** The paper finally applies this mechanism to two of our most intimate psychological experiences: **when we feel regret over the past, we are using function-cards to construct another (superior) possibility; when we contemplate the future's direction, having constructed one possibility with function-cards, we construct another, and the future then seems to have forked into two—corresponding to these two possibilities.** Both share the same error: mistaking the number of times we have deliberated for the number of branches reality itself possesses. The paper guards several lines with particular gravity: this diagnosis denies neither the genuine reliability of physical causation nor the practical value of function-card reasoning in daily planning and communication; it does not verdict that all attribution and planning are void; and how judgments of "should have been otherwise" ought to be re-situated within practices of responsibility, the paper renders no verdict on, leaving it expressly **still under study**. This very paper is itself a deliberation just now assembled from function-cards, one more match between Guan Yu and Qin Qiong; it is a finger pointing at the moon; the finger may be discarded, the moon may not.

本文追一族彼此嵌套的判断，最终收束到一个统一的诊断：**我们所有讲述宇宙中模块间关系的语言与认知，都是在普适性物理定律发明之前，用来讲述宇宙规律的一种原始方式；而故事与反事实情境的设想，正是这种原始讲述方式结构上必然要求的一个（错误的）前提。** 论证从故事的内部结构起步：**故事自带寓意或中心思想，隐性地指明了相关反事实设想的内容**——奋斗的故事，自然带着一个颓废堕落的反事实作对照；可现实本身，并不内在地对应某一个确定的故事，而是同时可以被叙述成无穷多不同的故事，各自配套着不同的反事实。本文进而指出，当我们想象一个人若遭遇不同、此刻性格便会不同时，我们实际上把此刻的他，感受成了一个和过去不连续的陌生人——**我们没有感到他的恒常不变性，相当于认为过去那个他已经消失了**，同时又感觉"似乎有一条他还是他的路径"；这暴露了一个默认设定：我们认为人保持不变才是"自然""正常"的，而这份设定，使我们无法整合与该人互动的体验，等于把该人认知成了两个人。本文由此追到反事实的根：**设想某反事实情景本可发生，相当于认为实际发生的情景并非过去状态的自然延续**——反事实，表明我们认为世界被偷走了，我们被抛入了一个拙劣的替代品中；而"自然延续"，只有在不被认知为自然延续时，才第一次有了被谈论的必要，这时才有了反事实充当自然延续的候补。可**整个宇宙，本来就是其自身的自然延续**——没有切分之前，"这本该是别的样子"这句话无处安放；反事实只在我们把宇宙切成模块之后，才第一次有了栖身之地，而这一步想象，整个颠倒了自然延续与非自然延续的概念，却又能自圆其说。本文进而给出一个形式化的表述：设两模块在两个时刻的状态为A1、B1及A2、B2，A2+B2就是A1+B1的自然演变；可若只看A1到A2这一段，就总能说B破坏了A的自然演变——**这句话的言外之意，是"宇宙本应是A1+B1'而不是A1+B1"，可宇宙本来就是A1+B1而不是A1+B1'**。这类基于自然语言语义的粗略预测，本质上是对模块B之功能的描述，而功能描述天生带着**颗粒度**：狂风与暴雨这类词，从无精密定义，可我们照样能想象"昨天若是暴雨而非狂风，庄稼会怎样"——这想象不会因模糊而瘫痪，只是**我们容忍了它与真实结果之间一道从未被核算过的鸿沟**，而这份轻松的可想象性，恰恰制造了"实际发生的狂风并非必然"这一错觉。本文最后指出一处被系统性忽略的不一致：谈论B对A的影响时，我们往往不再追溯C对B的影响是如何穿行传递到A的，仿佛B屏蔽了C——**这唯一的原因，是思考这种传递会带来的复杂度；我们的文化刻意忽略了影响的传递，只因思考它太复杂。** 本文由此得出统一诊断：以上种种，都不是零散的认知怪癖，是"用模块间关系讲述宇宙"这一种前科学认知方式的结构性产物——在能够用普适物理定律一次性描述整个宇宙演变之前，把宇宙切成模块、谈论模块间关系，就是我们所有知识的本质；而故事与反事实，正是这种原始讲述方式必然要求、却从未被兑现的前提：一个本不存在的"模块独立自然演变"。本文守数条须格外郑重划定的界线：这一诊断不否认任何一次具体归因的实践价值；不构成"故一切归咎与解释都不成立"的裁决；至于责任究竟该如何在模块化的语言之外被重新安放，本文不下判决，明确悬置为**尚待研究**。本文这一篇，自己也是这场思考此刻切出的一个模块、讲出的一个故事；它是指月之指；指可弃，月不可弃。 ## Abstract This paper traces a nested family of judgments to one unifying diagnosis: **all our language and cognition for narrating relations among modules of the universe are a primitive mode of describing cosmic regularity, predating universal physical law; and stories, together with counterfactual imaginings, are a (false) premise this primitive mode structurally requires.** The argument begins within the story itself: **a story's moral or central theme implicitly specifies its matched counterfactual**—the story of striving naturally carries a contrasting counterfactual of decline; yet reality itself corresponds to no single determinate story, but can be narrated into countless different stories, each carrying its own matched counterfactual. The paper then shows that imagining a person's character would differ had they met different circumstances treats the present self as a stranger discontinuous with the past—**we fail to feel his constancy, tantamount to believing the past self has vanished**, while simultaneously feeling "as if there were a path where he remained himself"; this exposes a default assumption that remaining unchanged is what counts as "natural," and this assumption renders us unable to integrate our experience of interacting with that person, effectively cognizing him as two people. The paper traces this to the root of counterfactuals: **to imagine that some counterfactual scenario could have occurred is to hold that what actually happened was not the natural continuation of the prior state**—the counterfactual expresses a feeling that the world has been stolen, that we have been thrown into a shoddy substitute; and "natural continuation" only needs to be spoken of once it is no longer recognized as such, at which point the counterfactual steps in as its candidate replacement. Yet **the universe as a whole is already its own natural continuation**—prior to any slicing, "it should have been otherwise" has nowhere to attach; the counterfactual finds its foothold only once we have cut the universe into modules, an imagining that inverts the very concepts of natural and unnatural continuation while remaining perfectly self-consistent. The paper offers a formal statement: letting two modules' states at two moments be A1, B1 and A2, B2, A2+B2 just is the natural evolution of A1+B1; yet looking only at A1 to A2, one can always say B disrupted A's natural evolution—**the implication being "the universe should have been A1+B1' rather than A1+B1," when the universe simply was A1+B1, not A1+B1'.** Such rough, natural-language predictions amount to a description of module B's function in relation to A, and functional description inherently carries **granularity**: words like gale and rainstorm have no precise definitions, yet we can still imagine "had yesterday been a rainstorm rather than a gale, how would the crops have fared"—an imagining that does not collapse under vagueness, only that **we tolerate a never-measured gap between the imagined and the real outcome**, and it is precisely this easy imaginability that manufactures the illusion that the actual gale was not necessary. The paper finally identifies a systematically overlooked inconsistency: when discussing B's influence on A, we typically decline to trace how C's influence on B might have propagated through to A, as though B shielded C's influence—**the sole reason being the complexity such tracing would demand; our culture deliberately ignores the transmission of influence simply because thinking it through is too complicated.** The paper concludes that none of the above are scattered cognitive quirks but structural products of "narrating the universe through module relations"—prior to describing the universe's evolution once and for all with universal physical law, cutting the universe into modules and discussing relations among them just is the substance of all our knowledge; and stories together with counterfactuals are exactly the premise this primitive mode necessarily requires, yet never actually delivers: an independent "natural evolution of a module" that was never coherently defined. The paper guards several lines with particular gravity: this diagnosis denies no specific attribution its practical value; it does not verdict that all blame and explanation are void; and how responsibility ought ultimately be re-situated beyond modular language, the paper renders no verdict on, leaving it expressly **still under study**. This very paper is itself a module this thinking has just now cut out, a story just now told; it is a finger pointing at the moon; the finger may be discarded, the moon may not.

本文追一条从”故事自带反事实”出发、最终抵达”宇宙本是其自身唯一自然延续”这一判断的论证。起点是：故事自带寓意或中心思想，隐性地指明了相关反事实设想的内容——奋斗的故事，自然自带一个与之对比的颓废、放弃的反事实情境。可一段现实本身，并不内在地对应某一个确定的故事，而是同时可以被叙述成无穷多个不同的故事，各自配套着不同的反事实。本文进而展示这一机制在人格判断上的具体后果：当我们想象一个人若遭遇不同，此刻的性格便会不同时，我们实际上把此刻的他，看作了一个不同于过去的他的、陌生的人——我们没有感到他的恒常不变性，相当于认为过去那个他已经消失了；同时，我们又感觉似乎有一条”他还是他”的路径存在，仿佛这份改变并非必然。这表明我们认为，人保持不变（或只朝我们欢迎的方向改变）才是”自然”“正常”的；我们因此无法整合自己与该人互动的体验，只能分两段处理——等于把该人认知成了两个人。本文由此追问一个更根本的问题：“自然延续”这个概念，究竟是从哪里获得它的立足点的？ 答案是：反事实设想，表明我们认为世界消失了、被偷走了，我们被抛入了一个拙劣的替代品中——而”哪里不是自然延续”，答案是哪里都可以不是；描述”自然延续是什么”，本身就已经是在讲一个故事了。 本文据此立出全篇最重的判断：一切——整个宇宙——本来就是其自身的自然延续。反事实，只在我们把宇宙切分为模块之后，才第一次有了立足之地——设想模块A的实际发生并非自然延续、而应是另一版本A’，这一想象整个颠倒了自然与非自然的概念，却能自圆其说；而这种颠倒，恰恰是我们理解模块间关系的唯一办法，尽管两模块本来并无关系。把宇宙划分成模块、谈论模块间关系，这就是我们全部知识的本质——在我们能用普适物理定律一次性描述整个宇宙演变之前。本文将这一诊断形式化：设两模块在两个时刻的状态为A1、B1与A2、B2，若二者合起来即是整个宇宙，则A2+B2无可争议地是A1+B1的自然演变；可若只看A1到A2这一段，我们便总能说”B影响或破坏了A的自然演变”——而这句话，言外之意是”宇宙本应是A1+B1’而不是A1+B1”，可宇宙本来就是A1+B1，不是A1+B1’；“作为宇宙局部的A的自然演变”这一提法，从未被定义过，因为A从来不是独立于B而存在的整体，只是A+B这个自然延续的一部分。 本文进而指出，这套”B对A的功能”式的粗略预测，天生带着粗颗粒度：狂风与暴雨之间，没有精密的定义边界；可即便如此，我们仍能想象”昨天若是暴雨而非狂风，庄稼会怎样”，这份想象不会因颗粒度粗、定义模糊而瘫痪——只是这份想象，同样可能与真实的暴雨结果相差甚远，我们只是容忍了这份差异，而从未核算它到底有多宽；正是这份毫不费力的可想象性，制造了”实际发生的狂风并非必然”这一错觉。本文最后指出一处影响追溯的不一致性：设宇宙分三个模块A、B、C，我们评判B对A的影响时，往往不再追溯C对B的影响如何穿行至A，仿佛B屏蔽了这份传递——可这一停止追溯，唯一的原因，是思考这种传递会带来的复杂度；我们的文化刻意忽略了影响的传递，只因继续追溯会使思维变得极度复杂，于是简单地停止了这种思考，而非因为这里真有一个天然的、有原则的断点。 本文守数条须格外郑重划定的界线：这一诊断不否认任何具体故事、具体判断在其适用条件内的真实与有用；不构成停止归责、停止理解他人性格变化的理由；至于反事实与责任究竟该如何最终安放，本文不下判决，明确悬置为尚待研究。本文这一篇，自己也是把宇宙这唯一的自然延续，切成模块、讲成故事的又一次尝试；它是指月之指；指可弃，月不可弃。 Abstract This paper traces an argument from “stories carry built-in counterfactuals” to the culminating judgment that the universe is its own sole natural continuation. It begins: a story’s moral or central theme implicitly specifies its matching counterfactual—a story of striving naturally carries, as its foil, a counterfactual of decline or surrender. Yet a stretch of reality does not intrinsically correspond to one determinate story; it can equally be narrated as countless different stories, each carrying its own matched counterfactual. The paper shows this mechanism’s concrete consequence for judging character: when we imagine that had a person met different circumstances, his present character would differ, we in fact regard the present him as a different, unfamiliar person, not a continuation of his past self—we fail to feel his constancy, tantamount to holding that the past him has vanished; yet we simultaneously feel there existed a path on which he “remained himself,” as though this change were not necessary. This reveals that we hold constancy (or change only in directions we welcome) to be “natural,” “normal”; we thereby fail to integrate our experience of interacting with this person, splitting it into two segments—effectively cognizing him as two people. The paper then presses a deeper question: where does “natural continuation” get its footing at all? The answer: counterfactual imagining expresses a feeling that the world has vanished, been stolen, and we have been thrown into an inferior substitute—and as to where is “not” a natural continuation, the answer is anywhere can be; to describe what “natural continuation” consists in is already to be telling a story. The paper thereby states its heaviest judgment: everything—the whole universe—already is its own natural continuation. The counterfactual only gains a foothold once we carve the universe into modules—positing that module A’s actual occurrence is not its natural continuation, but should instead be some A’, entirely inverts the concepts of natural and unnatural, yet remains self-consistent; and this inversion is precisely the only way we have of understanding the relation between two modules, even though the two modules originally bore no relation at all. Carving the universe into modules and discussing the relations among them is the very essence of all our knowledge—prior to being able to describe the whole universe’s evolution at once via universal physical law. The paper formalizes this: let the two modules’ states at two moments be A1, B1 and A2, B2; if together they constitute the whole universe, then A2+B2 is, beyond dispute, the natural evolution of A1+B1; but looking only at the segment A1 to A2, we can always say “B influenced or disrupted A’s natural evolution”—yet this statement implicitly asserts “the universe should have been A1+B1’ rather than A1+B1,” when the universe simply is A1+B1, not A1+B1’; “A’s natural evolution as a piece of the universe” was never defined, since A was never a whole independent of B, only a part of the natural continuation that is A+B. The paper further notes that this “B’s function relative to A” style of rough prediction is inherently coarse-grained: there is no precise boundary between a gale and a rainstorm; yet even so, we can still imagine “had yesterday been a rainstorm rather than a gale, how would the crops have fared”—an imagining not paralyzed by coarseness or vagueness—only this imagining may likewise diverge greatly from the true outcome of an actual rainstorm, a gap we merely tolerate without ever reckoning its width; it is precisely this effortless imaginability that manufactures the illusion that the actual gale was not necessary. The paper finally points to an inconsistency in how far we trace influence: dividing the universe into modules A, B, C, when we judge B’s influence on A, we typically do not further trace C’s influence on B propagating on to A, as though B shielded that transmission—yet the sole reason for stopping the trace is the complexity that continuing it would bring; our culture deliberately neglects the transmission of influence simply because tracing it further would render thought unmanageably complex, and so thinking is simply halted, not because there is any principled, natural stopping point. The paper guards several lines with particular gravity: this diagnosis denies no specific story or judgment its truth or usefulness within its conditions of applicability; it is no reason to stop attributing responsibility or understanding a person’s change of character; and how the counterfactual and responsibility ought ultimately to be settled, the paper renders no verdict on, leaving it expressly still under study. This very paper is itself one more attempt to carve the universe’s sole natural continuation into modules, into a story; it is a finger pointing at the moon; the finger may be discarded, the moon may not.

本文追一个从未被正面问过的问题：同一段真实的物理过程，比如一只猫捉住一只老鼠，可以被压缩成无穷多个互不相容、却都真实的故事——"猫捉老鼠"，也可以是"老鼠甲的惊险的一天"、"感冒病毒对老鼠甲行为的影响"、"血液中的某白细胞捕食了细菌"、"某水分子中发生的量子力学现象"、"适者生存、优胜劣汰"、"猫的捕食活动在代际是如何传递的"。**到底哪个故事，才是这里真正发生的唯一故事？本文的判断是：并没有所谓唯一的故事，所有这些故事都在真实地发生着，即使它们在中心思想或寓意上彼此矛盾。** 可人类个体，对于自己所体验到的人生所涉及的物理过程，往往只会以一到两个故事来记忆一段物理过程，而不是同时用无穷多个故事来记忆同一段过程。本文称这一现象为**故事化坍缩**：无穷多个原则上同时成立的版本，在被记忆的那一刻，被压缩成了一到两个，其余的，即使同样真实、甚至足以推翻我们最终采纳的判语，也被永久错过。本文进而指出，这份错过的根源，往往不是选择性偏见，而是**信息在当时的条件下根本不可得**：设想两个旁观者，同时看着这场猫捉老鼠——我捕捉到的细节，编成了"黑猫捉到了大老鼠"；在场的另一人，捕捉到了一个我也看见、却因情绪基调不同而未曾编入我的故事的细节（一个老鼠本可逃入的洞），编成了"老鼠再跑快一点就能进洞了，真是遗憾"；而"老鼠因为感冒而跑慢了"这个同样真实的故事，则因为信息完全无法获得，被我们两人共同忽略——**正因为这一版本从未进入候选清单，"老鼠跑得不够努力，因此被捉了"这个中心思想，才得以立得住；唯有全知的上帝，才知道老鼠不是不努力，是感冒了。** 本文由此揭示一种寄生在这份永久信息缺失上的辩护策略：**自由意志论者可以声称，老鼠即使感冒了，凭着自由意志，还是能跑得比它实际跑的快一点——因为没有人能够证明这个选项不存在。** 这套辩护的力量，不在于它论证了什么，而在于它展示了一种永远不会输的策略的形状：它不需要证明老鼠"确实能"跑得更快，只需要指出对手"无法证明它不能"——这是一次彻底的举证责任倒置。而这套策略可以被无限地向前递归：自由意志论者同样可以声称，老鼠昨日本可自由地选择留在洞里避雨、从而不会感冒，因为对手同样无法证明老鼠昨天"必须"出去觅食。**"你无法证明我必须如此"与"我确实能够不如此"，中间隔着举证责任转移这一整层，从未被兑现过**——这是一套永不认输、却也永不断言任何具体内容的辩护。本文最后给出这一切的落点：**我不应把记忆中的故事认作现实的本质，而应将其视作一种标识符、一块牌子，这个牌子后面才是现实，现实与故事无关，故事只是一种可被替换、更改的表象。** 本文守数条须格外郑重划定的界线：这一诊断不否认任何一个具体故事本身的真实性，只否认其唯一性与穷尽性；不构成"故一切归责都不成立"的裁决；至于自由意志与责任究竟该如何最终安放，本文不下判决，明确悬置为**尚待研究**。本文这一篇，自己也是这场思考此刻坍缩出的一个故事、一块可被替换的标识符；它是指月之指；指可弃，月不可弃。 ## Abstract This paper pursues a question rarely asked directly: a single real physical process—a cat catching a mouse—can be compressed into countless mutually incompatible, yet equally true, stories: "the cat caught the mouse," or equally "mouse A's harrowing day," "the cold virus's effect on mouse A's behavior," "a white blood cell preying on bacteria," "a quantum event within some water molecule," "survival of the fittest," "how a cat's predatory behavior is passed down across generations." **Which story is the one true story actually occurring here? The paper's judgment: there is no single true story; all of these are truly occurring, even where they conflict in their central moral.** Yet an individual, remembering an episode of lived experience, collapses it into only one or two stories, rather than holding countless versions simultaneously. The paper names this **story-collapse**: countless versions, in principle equally valid, compressed at the moment of memory into just one or two, the rest—however true, however capable of overturning the judgment eventually adopted—forever missed. The paper argues this loss is typically not selective bias but that **the relevant information was simply unavailable at the time**: imagine two observers watching the same chase—the details I captured become "the black cat caught the big mouse"; the other observer, who caught a detail I too saw but, given a different emotional register, never wove into my story (a hole the mouse might have reached), produces instead "if only the mouse had run a bit faster, it would have made the hole—what a pity"; while "the mouse ran slow because it had a cold" is a story neither of us could tell, since that information was simply never available to either—**and it is precisely because this version never entered the candidate pool that the standing moral, "the mouse wasn't trying hard enough, hence was caught," gets to stand; only an omniscient god would know the mouse was not lazy but ill.** The paper thereby exposes a rebuttal strategy that parasitizes this permanent unavailability: **a free-will theorist may claim that even with a cold, the mouse, by virtue of free will, could still have run a bit faster than it actually did—since no one can prove this option did not exist.** The force of this rebuttal lies not in what it proves but in the shape of an unfalsifiable strategy: it need not prove the mouse "could" run faster, only point out the opponent "cannot prove it could not"—a thorough inversion of the burden of proof. And this strategy can be extended recursively backward: the free-will theorist may equally claim the mouse could have freely chosen, the day before, to stay in its burrow out of the rain and thus never catch cold, since the opponent likewise cannot prove the mouse "had to" forage that day. **Between "you cannot prove I had to" and "I really could have done otherwise" lies the entire burden of a proof-shift that is never actually discharged**—an unfalsifiable defense that never loses, and never asserts anything specific either. The paper concludes: **I should not take the story in my memory to be the essence of reality, but should regard it as a label, a tag—behind this tag lies reality, and reality has nothing to do with the story; the story is merely a replaceable, alterable appearance.** The paper guards several lines with particular gravity: this diagnosis denies no specific story its truth, only its uniqueness and exhaustiveness; it does not verdict that "so no attribution of responsibility holds at all"; and how free will and responsibility ought ultimately to be settled, the paper renders no verdict on, leaving it expressly **still under study**. This very paper is itself a story this thinking has just now collapsed into, a replaceable tag; it is a finger pointing at the moon; the finger may be discarded, the moon may not.

本文合并四条彼此独立发展、却共享同一个根源的论证，构成一部关于**我们对因果与责任的判断，如何被语法的恒常性预设、信息的不可得性、以及记忆的坍缩，系统性地扭曲**的整体论述。第一条线索是**语义创世**：一句原本只用来区分两具躯体的话——"他的想法"——在区分对象退场之后，被想象填满了它留下的语法空位，凭空造出一个躯体之外的主体；这一机制依次推广到动词与代词（"想"与"我"，及"我思故我在"之荒谬化）、空间（色盲房间、反事实、柏拉图洞穴）、内部房间的具体挂牌流程（僵尸悖论的自我拆解）、虚拟房间与建模的实存感、时间中的恒常感（塑料袋方案、看病诊断），以及自由意志辩护的收缩策略——辩护者把自由压缩进越来越窄的选择瞬间，这一收缩动作本身，正是他们承认瞬间之外不自由的行为证据。第二条线索是**认知边界如何被错当本体边界**：我们事后能给自己的行为找出各种原因，事前却不将外在的未知因素纳入计划或担忧，因为未知因素没有具体内容可填入图景，而这份"未填"被悄悄读成了"不存在"；这一空缺，在面对未来时比面对过去时更彻底，因为过去有具体结果倒逼我们承认"总有别的原因"，未来却没有这份倒逼——"唯一主宰"之感由此而生。第三条线索是**影响链的无限延伸与可描述性的门槛**：若认了"甲影响了乙"，这条线便会无限延伸，以致任何人此刻的行为都是所有前人的总和，这正面反证了"源于自己"这一感觉的不可靠；而我们判定一份影响是否存在，唯一的操作性标准，是它能否被讲通——这道门槛远比通常以为的容易跨过，只需一个通用的、无法被证伪没有影响的中间变量，便能弥散地覆盖后续一切；而这一门槛，本身又受着道德兴趣的预先筛选：涂不上道德色彩的因果关系，会被搁置，直到我们发明出涂色的办法。第四条线索，本文在此首次系统展开：**故事化坍缩**。一段真实的物理过程，可以被压缩为无数个互不相容、却都真实的故事版本——"猫捉老鼠"，也可以是"老鼠甲的惊险的一天""感冒病毒对老鼠甲行为的影响""某白细胞捕食了细菌""适者生存"；这些故事，没有一个是唯一正确的那个，它们都在真实地发生着，即使彼此矛盾。可个体记忆一段经历时，只会坍缩出一到两个故事，其余版本，即使同样真实、甚至足以推翻我们采纳的判语，也会永远错过——这份错过，往往不是因为选择性采纳，是因为相关信息在当时的条件下根本不可得。而**自由意志论**恰恰寄生在这份永久的信息不可得上：它可以无限地宣称"本可以更努力""本可以昨天就避免"，因为对手永远无法证明"必须如此"——这是一套永不认输、却也永不断言任何具体内容的辩护策略。本文最后给出统一诊断：我们记忆中的故事，从来不是现实本身，只是挂在现实门口、可被替换的一块标识符；现实，不因换了哪块牌子而有分毫改变。贯穿这四条线索的，是同一种对"话语具有脱离情境之恒常性"的误信，与同一种对"信息不可得即等于不存在"的错读。本文守数条须格外郑重划定的界线：这一切不触及第一人称此刻真实的确定性；不构成停止与他人交流、信任、协作的理由；不构成忽视他人痛苦感受的许可证；至于自由意志与责任究竟该如何最终安放，本文不下判决，明确悬置为**尚待研究**。本文这一篇，自己也是这场思考此刻坍缩出的一个故事、一块可被替换的标识符；它是指月之指；指可弃，月不可弃。 ## Abstract This paper merges four independently developed lines of argument, sharing one root, into a unified account of **how our judgments of causation and responsibility are systematically distorted by grammar's presumption of constancy, by the unavailability of information, and by the collapse of memory.** The first thread is **semantic genesis**: a phrase originally meant only to distinguish two bodies—"his thought"—once the disambiguated party recedes, has its grammatical subject-slot filled by imagination, conjuring a subject beyond the body; this mechanism recurs across verbs and pronouns (Cartesian absurdity), space (the color-blind room, counterfactuals, Plato's cave), the concrete labeling of an "inner room" (the zombie paradox self-undoing), the virtual room and its sense of reality, constancy in time (the plastic-bag solution, the diagnosed illness), and the shrinking defense of free will—defenders compress freedom into an ever-narrower instant, the compression itself being evidence they concede everything outside it is unfree. The second thread is **the cognitive boundary mistaken for the ontological boundary**: we readily find causes in hindsight, yet fail to factor unknown future factors into our plans, because the unknown has no specific content to fill a slot, and this blank is quietly read as nonexistence—more thoroughly so for the future than the past, since the past has an actual outcome forcing some acknowledgment while the future has none, whence the sense of sole mastery. The third thread is **the infinite extension of influence and the threshold of describability**: once one grants that A influenced B, the chain extends without limit, refuting the reliability of "sourced from oneself"; and the sole operative criterion for judging whether an influence exists is whether it can be told coherently—a threshold far easier to clear than supposed, requiring only a generic, unfalsifiable intermediate variable; this threshold is itself pre-filtered by moral interest, uncolorable causation being set aside until a means of coloring is invented. The fourth thread, developed here for the first time systematically, is **story-collapse**. A real physical process can be compressed into countless mutually incompatible yet equally true stories—"the cat caught the mouse," or equally "mouse A's harrowing day," "the cold virus's effect on mouse A's behavior," "a white blood cell preying on bacteria," "survival of the fittest"; none is the one correct story, all are truly occurring even where they conflict. Yet an individual, remembering an episode, collapses it into only one or two stories; the rest, however true, however capable of overturning the judgment we adopted, are forever missed—not usually from selective bias but because the relevant information was simply unavailable at the time. **Free-will apologetics** parasitizes exactly this permanent unavailability: it can endlessly claim "could have tried harder," "could have avoided it yesterday," since the opponent can never prove "had to be so"—an unfalsifiable strategy that never loses, and never asserts anything specific either. The paper concludes with a unifying diagnosis: the story in our memory is never reality itself, only a replaceable tag hung at reality's door; reality is not altered in the least by which tag we happen to hang. Running through all four threads is the same false belief in semantic constancy independent of context, and the same misreading of informational unavailability as nonexistence. The paper guards several lines with particular gravity: none of this touches first-person present certainty; none of it is reason to cease communicating, trusting, or cooperating with others; none of it licenses disregarding another's suffering; and how free will and responsibility ought ultimately to be settled, the paper renders no verdict on, leaving it expressly **still under study**. This very paper is itself a story this thinking has just now collapsed into, a replaceable tag; it is a finger pointing at the moon; the finger may be discarded, the moon may not.

本文追一个统一的机制——**语义创世**——它在代词、空间、时间与自由意志辩护这四个看似无关的领域里，反复地凭空造出一个"躯体之外的主体"。机制的原型极朴素：**"他的想法"这句话，本只是一个区分算子。** 两具躯体同时在场时，这句话只做一件事——把话题从"关于乙的情况"切成"关于甲的情况"；此刻，"想法需不需要有个主语"这个问题从未被激活。可一旦乙的躯体退场，只剩甲一具躯体，我们再度说"他的想法"时，语法留下的主语空位，被**想象**填了上去——这个填充没有任何新证据支撑，只因语法结构"读起来该完整"，而这份闭环带来的顺畅感，被错当成了对应真实的正确感。我们遗忘了这句话原是区分算子、其历史任务已经完成，却将错就错，用新的方式解读旧语义，合理化了新的想象而不自知——这背后，是对"话语具有脱离情境之恒常性"的误信。本文进而展示这同一机制的四次推广。其一，**推广到动词与代词**："我想吃饭"与"他想吃饭"共用一个"想"字，可后者对我而言纯属外在、物理性的报告，前者却是我正经历的体验；这份共用制造了对称的假象，而这假象仅在第三方视角下才真的成立。"我"字亦然：只有在第三方发起区分性提问（"谁想去吃饭？""我！"）时，"我"字才在语义上真正对称；离开这一情境（如笛卡尔式独白），"我"字的所指便发生了隐秘的畸变。本文据此论证：**"我思故我在"这句话，以其原初语义（区分算子）理解，是荒谬的**——躯体的存在肯定先于其思考，第三方从不会有论证笛卡尔是否存在的动力；这句话之所以显得深刻，全赖一次隐秘的置换——把"我"从指着躯体，偷换成指着一个被想象出来的灵魂。其二，**推广到空间**：我们想象色盲者"看到"的是一个内部的、同构的灰度房间，这表明我们把日常所见，也想象成了内部的、认知中的东西；可这想象是次生的，原初语义下，所见之物是外部的、独立于认知的。反事实情境与柏拉图的洞穴隐喻，进行的是同一场语义次生。其三，**推广到"挂牌"的具体流程**：我们想象他人"内部房间"的方式，其实是复制一个和外部房间一模一样的房间，只在门口挂一块"内部"的牌子；再把这人的外部形象克隆一份塞进去，挂上"灵魂"的牌子；再挂上"盲人"的牌子，就得到了哲学僵尸——**若这具僵尸的内部房间从不可被访问，这房间存在的唯一理由也随之消失，这正是僵尸概念自身的悖论。** 而这同一套虚拟房间装置，也被我们用来理解自己的思维（如提前关门的假想情景）：虚拟房间与虚拟房客彼此支撑，显得"活了"，其实两者都不存在；某种恒常感更让我们误以为这场景"一分钟前就已构想出"，可它此刻才第一次被造出。这隐喻的实质，只是大脑里的某种物理情况，隐喻的空间形式，却带来了物理情况本不具备的恒常性。其四，**推广到时间与选择**：一个此刻才被制造或确证出的方案／答案，会给人一种"它一直存在、我只是没瞥一眼"的错觉；"当初有几个选项"这类叙事，是选项被事后安放到那个时间点上的明证——实际发生的，只是一段连续、不可切分的物理过程，被硬生生切成几步，再在每一步里倒推出选项。同一机制，也解释了为什么自由意志的辩护者，总是把自由压缩进一个越来越窄的选择瞬间：这套压缩没有天然的止点，可以反复地对任何"自由区间"施加"内部再分出计算与拍板"的操作；而当"大脑的自由选择不可想象"这一物理事实无法回避时，辩护者便把选择的主体，偷偷换成大脑之外的意识／灵魂——这正是语义创世在自由意志辩护里的重演。**辩护者把防线一路收缩到一个瞬间，这个收缩动作本身，恰恰是他们承认瞬间之外不自由的行为证据。** 本文最后指出，贯穿这四次推广的共同根源，是同一种**恒常感**：它把此刻才被制造、填补、确证的东西，错觉成了早就存在、独立于此刻的东西——这份错觉不可见，因为它就发生在语言与想象最基本的语法零件里。本文守数条须格外郑重划定的界线：这一诊断不触及第一人称此刻真实的确定性；不构成停止与他人交流、信任、协作的理由；不构成忽视他人痛苦感受的许可证；至于自由意志、责任该如何最终安放，本文不下判决，明确悬置为**尚待研究**。本文这一篇，自己也是这场思考此刻生成的一段语言；它是指月之指；指可弃，月不可弃。 ## Abstract This paper traces one unified mechanism—**semantic genesis**—recurring across pronouns, space, time, and the defense of free will, each time conjuring a "subject beyond the body" out of nothing. Its prototype is plain: **"his thought" is originally only a disambiguation operator.** With two bodies present, the phrase does one thing—cutting the topic from "about B" to "about A"; whether "thought" needs a subject is never raised. Once body B recedes, leaving only A, saying "his thought" again lets imagination fill the grammatical subject-slot left behind—no new evidence, only grammar's felt need for completeness, whose closure is mistaken for correctness. We forget the phrase's task was done, and rationalize the new imagining unawares—behind this lies a false belief in semantic constancy independent of context. The paper shows this mechanism recur four times. First, **onto verbs and pronouns**: "I want to eat" and "he wants to eat" share one word, though the latter is, for me, a purely external report while the former is my lived experience; this sharing manufactures a false symmetry, true only from a third-party view. "I" is likewise: only when a third party poses a disambiguating question ("who wants to eat?" "I!") is "I" truly symmetrical; outside this, as in Cartesian soliloquy, "I" undergoes a covert deformation. Hence **"I think, therefore I am," read in its original semantics, is absurd**—the body's existence plainly precedes its thinking, and a third party never needs to argue for Descartes's existence; the sentence seems profound only through a covert substitution, from the body to an imagined soul. Second, **onto space**: we imagine what a color-blind person "sees" as an inner, isomorphic grayscale room, revealing that we've reimagined everyday sight itself as internal and cognitive; but this is derivative—originally, what is seen is external, independent of cognition. Counterfactuals and Plato's cave perform the same semantic derivation. Third, **onto the concrete "labeling" procedure**: we imagine another's "inner room" by copying an outer room and hanging a sign reading "inner" on its door; cloning that person's outer form and placing it inside yields a "soul"; hanging a further sign "blind" yields the philosophical zombie—**if this inner room can never be accessed, its very reason for being posited vanishes, the zombie concept's own paradox.** The same virtual-room device explains how we understand our own thinking (imagining having shut a door earlier): the virtual room and its virtual occupant prop each other up into seeming alive, though neither exists; a sense of constancy further misleads us into feeling the scene was conceived a minute ago, when it was only just now constructed. The metaphor's substance is merely some physical state in the brain; its spatial form lends it a constancy the physical state itself never had. Fourth, **onto time and choice**: a solution or answer only just constructed or confirmed gives the illusion of having always existed, merely unglanced-at; narratives of "having had options at the time" are proof that options are retroactively inserted—what actually occurred was one continuous, uncuttable physical process, artificially sliced into steps, each retrofitted with options. The same mechanism explains why free-will's defenders keep compressing freedom into an ever-narrower instant of choice: this compression has no natural stopping point, and can be reapplied indefinitely to any "free interval"; and when the unimaginability of a free brain becomes unavoidable, defenders quietly swap the choosing subject from brain to a soul beyond it—semantic genesis recurring in free-will apologetics. **That defenders keep shrinking the front line to a single instant is itself behavioral evidence that they concede everything outside it is unfree.** The paper concludes that a single sense of **constancy** underlies all four recurrences—mistaking what is only now manufactured, filled in, or confirmed for something that existed all along, independent of the moment—invisible precisely because it operates in language's and imagination's most basic grammatical parts. The paper guards several lines: this diagnosis does not touch first-person present certainty; it is no reason to cease communicating, trusting, or cooperating with others; it licenses nothing regarding disregard for another's suffering; how free will and responsibility ought ultimately to be settled, the paper renders no verdict on, leaving it expressly **still under study**. This very paper is itself a stretch of language this thinking has just now generated; it is a finger pointing at the moon; the finger may be discarded, the moon may not.

本文重写并推进一条论证，最终追到它自己的语言学根源。判断的起点：**若把他人的想法定义为他的内心独白，这就破解了意识的难问题——我们不再需要假设他人具有内心独白之外的"意识"。** 展开这一定义，须拆开"人有想法"捆着的三层意思——内部的、不可见的、似乎不确定的；不确定性被改写为信息只传给了当事人、没有传给旁人这一后果。本文收紧两处曾经的含混：**没有什么所谓"感觉到确定"这件事，当事人知道自己的内心独白，这句话本身就是确定性的完整说明；内心独白，就是主观感受——和他人看待我的客观视角相比，没有再"如何主观感受到内心独白"这一说。** 主观／客观标注的是同一件事被从两个位置观测的两种称呼，不是两种不同的东西。本文进而拆穿僵尸世界的自相矛盾：若僵尸能用内心独白真的描述环境，按已确立的等同，它便已在感受这个世界；若这份描述不是真的描述，僵尸赖以成立的"功能完全相同"这一前提便未被兑现。本文诊断难问题的真正病灶：**它从未证成它悄悄预设的前提——"他人具有与我的意识等同的东西"；这正是皇帝的新装。** 哲学僵尸，其实是唯我论的极端形式。本文将此接回一条更深的结构——**主体之外的主体，原因之外的原因**：灵魂，只是主体感、原因感、解离感这三份第一人称现象的名词化，恰如愤怒把体验隐喻成实体，这一隐喻仅对我自己成立。本文进而指出，我们据以相信他人有意识的真正根据，不是意识，是**语言**——我用同一套语义系统理解自己的内心独白与他人说的话，这份对称感受，才是相信他人有意识的真正原因。可这块地基同样不稳：**哲学僵尸问题，等价于中文房间问题**——除非认定他人正常的说话活动本身就证明了理解能力，否则没有任何证据表明他人理解语言；一具躯体的输出只是输入与结构的函数，读不出结构之外的任何东西。本文如实呈现由此产生的两个分岔（归纳猜想、彻底的唯我论），指出它们分属**两种互不相容的叙事风格**——物理／结构的叙事与灵魂论现代版的叙事，语法互斥，不可通约；本文此刻用后者向"你们"说话，只证明作者仍在使用这套风格，不证明"你们"具有它所预设的东西。本文最后追到这整套困惑的语言学根源：**"语义创世"。** "他的想法"这句话，原本只是一个**区分算子**——两具躯体同时在场时，它只做一件事：把这句话，从"关于乙的情况"切分成"关于甲的情况"，仅此而已，主语是否指着躯体之外的东西，这个问题此刻从未被激活。可一旦乙的躯体退场、只剩甲一具躯体，我们再度说"他的想法"时，语法留下的那个主语空位，被**想象**填了上去——这个填充，没有任何新证据支撑，只因语法结构"读起来该完整"，而这份闭环带来的顺畅感，被我们错当成了对应真实的正确感。我们遗忘了这句话原是区分算子、其历史任务在乙退场时已经完成，却将错就错，用新的方式解读旧语义，从而合理化了新的想象而不自知——这预设了话语具有脱离具体情境的恒常性，而这一预设，本身就是错的。本文进而指出第二重混淆：与躯体交流以获知**已知事实性质**的信息（如图书馆几点开门，预存于结构中）是合理的活动；可试图与躯体交流以获知意识难问题的答案，是不合理的——那答案根本不是已知事实性质的信息，从不预存于任何结构中，等待被检索。**试图通过与他人的交流活动获得意识难问题的答案，是在不可能的前提条件下进行的、不可能成功的认知活动类型。** 本文守数条须格外郑重划定的界线：不触及第一人称确定性；绝非停止与他人交流、信任、协作的理由，语言的社会实践不因这一诊断而停摆，本文自身能被写出、被期待被理解，正是明证；绝非忽视他人痛苦的许可证。他人是否理解语言、是否具有主观体验，该被如何最终对待，本文不下判决，明确悬置为**尚待研究**。本文这一篇，自己也是用灵魂论现代版叙事、向"你们"说出的一段内心独白；它是指月之指；指可弃，月不可弃。 ## Abstract This paper rewrites and advances an argument to its own linguistic root. Starting judgment: **defining another's thought as their inner monologue dissolves the hard problem of consciousness**—we no longer need to posit a "consciousness" beyond the monologue. Unpacking this requires prying apart the trio bundled into "having thoughts"—internal, invisible, seemingly uncertain; uncertainty is rewritten as the consequence of information reaching one party and not another. The paper tightens two prior ambiguities: **there is no "felt sense of certainty" separable from knowing—that the subject knows their inner monologue just is the complete account of certainty; the inner monologue just is the subjective experience, by comparison with another's objective view of me; there is no further question of how it is subjectively felt.** Subjective/objective name one event from two vantage points, not two kinds of thing. The paper then dismantles the zombie world's self-contradiction, diagnoses the hard problem's true ailment—it never argues for its presupposition that another possesses something equivalent to my consciousness, the emperor's new clothes—and connects this to a deeper structure, **a subject beyond the subject, a cause beyond the cause**: the soul is the nominalization of a sense of agency, causation, and dissociation, valid only for oneself. The paper argues what actually grounds belief in another's consciousness is not consciousness but **language**—the same semantic system handling my monologue and another's speech, a symmetry that grounds the belief. Yet this foundation is equally unstable: **the zombie problem equals the Chinese Room problem**; a body's output is merely a function of input and structure, yielding nothing beyond it. The paper presents, without adjudicating, two resulting forks, showing they belong to **two incompatible narrative styles**—physical-structural and modern soul-theory—grammatically exclusive, incommensurable; this paper's own address to "you" proves only the author's current habit of speech, not that "you" possess what it presupposes. The paper finally traces the whole confusion to its linguistic root: **semantic genesis.** "His thought" was originally only a **disambiguation operator**—with two bodies present, it merely marks "about A, not B"; whether the subject points beyond the body is never activated. Once body B recedes, leaving only A, the grammatical subject-slot left behind gets filled by **imagination**—no new evidence, only grammar's felt need for completeness, whose closure is mistaken for correctness. We forget the phrase's task was already done, and rationalize the new imagining unawares, presupposing semantic constancy independent of context—itself false. A second confusion follows: querying a body for **known-fact information** (library hours, pre-stored in structure) is legitimate; querying it for the hard problem's answer is not—that answer is never pre-stored, awaiting retrieval, in any structure. **Seeking the hard problem's answer through communication with others is an impossible type of cognitive activity, attempted under impossible preconditions.** The paper guards several lines: it does not touch first-person certainty; it is by no means a reason to cease communicating, trusting, or cooperating with others—the social practice of language continues regardless, and this very paper's being written and expected to be understood is proof; and none of this licenses disregarding another's suffering. How another's linguistic understanding and subjective experience ought ultimately be treated, the paper renders no verdict on, leaving it **still under study**. This very paper is itself an externalized inner monologue, addressed to "you" in the modern style of soul-theory; it is a finger pointing at the moon; the finger may be discarded, the moon may not.

本文重写并推进一条论证，把它从"意识"这个靶子，逼向一个更深、也更难被绕开的地基——**语言的理解**。判断的起点：**若把他人的想法定义为他的内心独白，这就破解了意识的难问题——我们不再需要假设他人具有内心独白之外的"意识"，内心独白可被视为一种类似说话的纯物理过程。** 展开这一定义，须先拆开"人有想法"捆着的三层意思——内部的、不可见的、似乎不确定的；内部性被继承，不可见性被还原成未发声的语言活动，不确定性被改写为信息只传给当事人、没传给旁人这一后果。本文进而收紧两处曾经的含混：**没有什么所谓"感觉到确定"这件事，与"知道"分离开来还需被单独解释——当事人知道自己的内心独白，这句话本身就是确定性的完整说明；内心独白，就是主观感受——和他人看待我的客观视角相比，没有再"如何主观感受到内心独白"这一说。** 主观／客观标注的不是两种不同的东西，是同一件事被从两个不同位置观测时的两种称呼。本文进而正面拆穿这条路径历史上最强的反驳——僵尸世界的可设想性：若僵尸能用内心独白真的描述环境，按已确立的等同，它便已在感受这个世界，"感觉起来什么都不是"便自我推翻；若这份描述不是真的描述，僵尸赖以成立的"功能完全相同"这一前提便未被兑现。本文由此诊断难问题的真正病灶：**它问"为什么他人能有意识"，却从未证成它悄悄预设的前提——"他人具有与我的意识等同的东西"；这正是皇帝的新装，先假装皇帝穿了衣服，再讨论这件衣服为何难以被看见。** 哲学僵尸的概念，其实是唯我论的极端形式——"意识"在其中的定义，只是"我的意识"的另一个名字。本文将此接回一条更根本的结构——**主体之外的主体，原因之外的原因**：难问题问的是，大脑／身体这个主体、这个原因之外，为何还要另有一个主体、另有一个原因，可根本没有另外的主体或原因；灵魂，只是主体感、原因感、解离感这三份真实的第一人称现象的名词化，恰如愤怒把体验隐喻成实体，这一隐喻仅对我自己成立。本文进而推进到本文最深的一步：**意识的难问题中的意识之定义，是唯我论式的，而又声称他人也有，这一设定本身就是自相矛盾的；真正该问的第一个问题，是为什么他人（从我的视角）没有意识，答案是视角问题；而我们据以相信他人有意识的真正根据，不是意识，是语言——我的内心独白，和我听到的他人所说的话，对我而言是串在一起的，我用同一套语义系统理解这三段内容，于是我对称地感觉别人似乎也能理解他自己说的话。** 语言，因此比意识更根本。可本文进而指出，这块新地基同样不稳：所有这些语言——我的内心独白与他人嘴上说的话——只是同等地被呈现给我，除非我认定他人（从我视角判断）正常的说话活动本身就证明了他的理解能力，否则这里完全没有证据表明他人理解任何语言；**哲学僵尸问题，等价于中文房间问题**，二者都可以对称地宣称除自己外无人理解任何语言，而这就是唯我论——一具躯体的输出只是输入与结构的函数，我们只能从输出反推结构，不可能从输出读出结构之外的任何新东西，这一结构，就是个中文房间，或哲学僵尸。本文如实呈现由此产生的两个分岔，不代为择定：其一，用同一套语义系统顺畅处理全部语言输入，故猜想他人也用同一套系统理解，是一个更简约、更合适的**猜想**；其二，把"理解"一词分两种情况理解——对自己与对他人——由此认为只有自己才真正理解，是**唯我论**，且这也应被还原到它最彻底的程度：只有自己才理解语言。本文最后指出，这两个分岔之所以看似互不相容、却各自自洽，是因为它们分属**两种互不兼容的叙事风格**：**其一，物理／结构的叙事**——在此风格下，"它"只能指那具躯体，它的理解只能指一种物理过程本身，其结构中并不存在任何可以充当"理解之主体"的东西；**其二，玄幻或灵异风格，即灵魂论的现代版本**——唯有在此风格下，"理解"才预设一个主体、才问得出"它是否真的懂了"。两种叙事对"理解"一词的语法安排互斥，不可通约，不是同一件事的两种描述深度。本文据此指出：**我们此刻正用第二种叙事风格与"你们"探讨这一切，这只说明说话者（作为哲学家）此刻仍在使用这套灵魂论的现代版本，丝毫不能反过来证明"你们"具有这套叙事所预设的那个理解主体——使用一套叙事，从不等于这套叙事所断言的东西为真。** 本文由此发展的推论是：他人的主观感受、乃至他人是否理解语言，其证据基础，皆比我们通常以为的要薄弱得多——不是因为它们已被证伪，是因为支撑这一整套信念的地基，一路追下去，最终落在了一种从未被独立证成、只是被言说习惯挪用的叙事风格上。本文守数条须格外郑重划定的界线：这一论证不触及第一人称的确定性；这一论证不构成停止与他人交流、信任、协作的理由——语言的社会实践，无论这一哲学诊断如何，仍完整地、必要地运转着，本文自身能被写出、被期待被读者理解，正是这一实践性事实的明证；且这一切绝不构成忽视他人报告之痛苦、感受或福祉的许可证。至于他人是否理解语言、是否具有主观体验，这一问题该被如何最终对待，本文不下判决，明确将其悬置为**尚待研究**。本文这一篇，自己也是用灵魂论现代版的叙事风格、向"你们"说出的一段内心独白的外化；它是指月之指；指可弃，月不可弃。 ## Abstract This paper rewrites and advances an argument, driving it from "consciousness" as its target toward a deeper, less avoidable foundation—**the understanding of language.** Its starting judgment: **if we define another's thought as their inner monologue, this dissolves the hard problem of consciousness—we no longer need to posit that another possesses a "consciousness" beyond the inner monologue.** Unpacking this requires prying apart the trio bundled into "a person has thoughts"—internal, invisible, seemingly uncertain—and tightening two prior points of ambiguity: **there is no "felt sense of certainty" separable from knowing itself; the inner monologue just is the subjective experience—by comparison with the objective view another takes of me; there is no further question of how the monologue is subjectively felt.** Subjective/objective mark not two kinds of thing but one event named differently by reference frame. The paper then squarely dismantles the strongest historical rebuttal—zombie conceivability: if a zombie's inner monologue genuinely describes its surroundings, then, by the established identity, it is already feeling that world, undoing "feels like nothing"; if the description is not genuine, the zombie's founding premise of functional identity goes unmet. The paper thereby diagnoses the hard problem's true ailment: **it asks why another can have consciousness while never arguing for what it presupposes—that another possesses something equivalent to my consciousness; this is the emperor's new clothes.** The concept of the philosophical zombie is, in truth, an extreme form of solipsism. The paper connects this to a deeper structure—**a subject beyond the subject, a cause beyond the cause**—and supplies an accounting of "soul" as the nominalization of a sense of agency, a sense of causation, a sense of dissociation, valid only for oneself, exactly as anger reifies a feeling into a solid. The paper then advances to its deepest move: **the definition of consciousness within the hard problem is solipsistic, yet claims another has it too—this setup is itself self-contradictory; the real first question is why another (from my viewpoint) lacks consciousness, and the answer is a matter of vantage point; and what we actually rely on to believe another has consciousness is not consciousness at all, but language—my inner monologue and the speech I hear from another are, for me, interlaced into one stream, understood through a single semantic system, and it is this symmetry that grounds my feeling that another likewise understands his own speech.** Language is thus more fundamental than consciousness. Yet the paper shows this new foundation is equally unstable: all this language—my inner monologue and another's spoken words—is presented to me on equal footing, and there is no evidence whatsoever that another understands any language, unless I take his (from my vantage, apparently normal) speech activity itself as proof of his comprehension. **The problem of the philosophical zombie is equivalent to the Chinese Room problem**—both can symmetrically claim that no one but oneself understands any language, and this just is solipsism; a body's output is merely a function of its input and structure, from which we can only infer structure, never anything beyond it—this structure just is a Chinese Room, or a zombie. The paper faithfully presents, without adjudicating, the two forks this yields: that positing another's comprehension, via the same semantic system handling all linguistic input smoothly, is the more parsimonious, apt guess; or that splitting "understanding" into two senses—for oneself and for another—yields solipsism, radicalized to its fullest extent: only oneself understands language at all. The paper argues these two forks appear incompatible yet each self-consistent because they belong to **two mutually incompatible narrative styles**: **the physical/structural narrative**, in which "it" can only denote the body, its "understanding" can only denote a physical process itself, and its structure contains nothing that could serve as a subject of understanding; and **the fantastical or supernatural style, the modern version of soul-theory**, in which alone "understanding" presupposes a subject, and one can ask whether it truly grasps. These two grammars for the word "understanding" are mutually exclusive, not two depths of description of one thing. The paper thereby concludes: **that we are presently conversing with "you" in the second narrative style shows only that the speaker, as a philosopher, is currently still using this modern soul-theory; it in no way proves that "you" possess the subject this narrative presupposes—using a narrative never entails that what it asserts is true.** The paper thus advances the inference that the evidential basis for another's subjective experience, and for another's linguistic understanding, is considerably weaker than ordinarily supposed—not because either has been disproven, but because the whole edifice, traced to its foundation, rests on a narrative style never independently established, only borrowed by habit of speech. The paper guards several lines with particular gravity: this argument does not touch first-person certainty; it is by no means a reason to cease communicating, trusting, or cooperating with others—the social practice of language continues to run, fully and necessarily, regardless of this philosophical diagnosis, and that this very paper can be written and expected to be understood by a reader is itself proof of that practical fact; and none of this licenses, in the slightest, disregarding another's reported suffering, feeling, or welfare. How the question of another's linguistic comprehension and subjective experience ought ultimately to be settled, the paper renders no verdict on, leaving it expressly suspended as **still under study**. This very paper is itself an externalization of an inner monologue, addressed to "you" in the modern style of soul-theory; it is a finger pointing at the moon; the finger may be discarded, the moon may not.

本文重写并收紧一次此前留有余地的论证：**若把他人的想法定义为他的内心独白，这就破解了意识的难问题——我们不再需要假设他人具有内心独白之外的"意识"，内心独白可被视为一种类似说话的纯物理过程。** 展开这一定义的机制须先拆开"人有想法"这句话原本捆着的三层意思：**内部的、不可见的、似乎不确定的**——本文指出它们是三件独立的事，只是在"想法"这个案例上恰好叠在了一起。内部性被内心独白这一定义直接继承；不可见性被还原成一件极普通的事——内心独白是未发出声音的语言活动，如同一封锁在抽屉里、只有写信人读过的信；不确定性被彻底改写来源——它不是想法自带的神秘属性，是信息只传给了当事人、没有传给旁人这一纯粹信息论事实的后果。本文进而收紧此前含混：**没有什么所谓"感觉到确定"这件事，与"知道"本身分离开来还需被单独解释——当事人知道自己的内心独白，这句话本身，就是确定性的完整说明。** 由此推进到最紧的一步：**内心独白，就是主观感受——和他人看待我的客观视角相比；不再有"如何主观感受到内心独白"这一说。** 主观／客观标注的不是两种不同的东西，是同一件事被从两个不同位置观测时的两种称呼；二者之间没有一道需要被跨越的鸿沟，只有一次视角的切换。本文进而正面处理这一路径最强的历史反驳——僵尸世界的可设想性——并指出其自相矛盾：假设一个世界，其中的僵尸能用内心独白描述周围环境，那么，若这份描述是真的描述，按已确立的等同（内心独白即主观感受），它便已经在感受着这个世界，"感觉起来什么都不是"这一设定便自我推翻；若这份描述不构成真的描述，则"功能与物理完全相同"这一僵尸设定的前提本身就没有兑现。而且，我们对僵尸第一人称视角的不可及，与我们对任何其他人第一人称视角的不可及，完全对称——若不可及本身足以让人怀疑僵尸没有感受，这份怀疑理应平等地落在一切他人身上，除非能给出一个非武断的区分标准。本文由此指出难问题真正的病灶：**意识的难问题，问的是"为什么他人能有意识"，却从未论证过它悄悄预设的那个前提——"他人具有与我的意识等同的东西"；这正是皇帝的新装：先假装皇帝穿了衣服，再郑重其事地讨论这件衣服为何难以被看见，而衣服从未存在过。** 他人明明只有外在的纯物理活动；哲学僵尸的概念，其实是唯我论的极端形式——"意识"在其中的定义，实则只是"我的意识"的另一个名字，从未真正给出过"他人的意识"这一独立定义。本文进而把这一诊断，接回一条更根本的结构：**意识的难问题，问的是大脑／身体这个主体、这个原因之外，为何还要另有一个主体、另有一个原因——可根本没有另外的主体或原因**；这与此前已被拆穿的"框外的额外操控者"、"宇宙状态之外的额外动力"，是同一个错误的第三次现身，只是这次换上了"灵魂／意识"的服装。本文进而给出"灵魂"这一名词的账目：它是**主体感、原因感、解离感**这三份真实的第一人称现象，被压缩进一个名词的结果——恰如"愤怒"这个名词，把一种体验隐喻成了一个可被容器盛放的实体，"灵魂"把主体感、原因感、解离感，隐喻成了一个可被身体这个容器盛放的实体；这一隐喻，**仅对我自己成立**——因为我确有那三份体验作为它的基础；把同一个词挪用到他人身上时，挪用的只是名字，没有挪用支撑这个名字的体验基础。本文对此仍守公允：这一诊断有力地拆穿了僵尸世界这一具体设定的自相矛盾，但真正的边界并未就此消失，只是退到了一处更深、也更专门的方法论问题——即便承认"僵尸这一具体设定自相矛盾"，解释鸿沟传统仍可追问：能否仅凭**先验的、概念上的分析**，不预设"内心独白即主观感受"这一等同本身，就从一份完整的物理／功能描述，推出必然存在主观体验；这一等同，此刻仍是被主张、被论证支持，而非被独立地、无循环地推导出来的——这与"是否能设想一个僵尸"已是两个不同层级的问题，本文如实呈现这一更深的分歧，不作裁定。在此基础上，本文发展一记更尖锐的推论：**他人的主观感受是否真的存在，在这一更收紧的新定义下，显得更加可疑。** 此前最强的证据——"他显然在思考"——已被证明即是内心独白本身、即是主观感受，被整个抽走；剩下的只是一条更弱的类比推理，而"他人的意识"这一概念本身，如今被诊断为一次从未被证成的、把"我的意识"这个名字挪用到他人身上的移植。本文守数条须格外郑重划定的界线：这一论证不触及第一人称的确定性；"更可疑"绝不等于"已被证伪"；且这一切绝不构成忽视他人报告之痛苦、感受或福祉的许可证。他人主观体验这一假设最终该被如何对待，本文不下判决，明确将其悬置为**尚待研究**。本文这一篇，自己也是由这场思考生成的一段语言、一段内心独白的外化；它是指月之指；指可弃，月不可弃。 ## Abstract This paper rewrites and tightens an argument that previously left room for slack. **If we define another's thought as their inner monologue, this dissolves the hard problem of consciousness—we no longer need to posit that another possesses a "consciousness" beyond the inner monologue; the inner monologue can be regarded as a purely physical process akin to speaking.** Unpacking this requires first prying apart the trio bundled into "a person has thoughts"—**internal, invisible, seemingly uncertain**—three independent matters that merely coincide in the case of thought. Internality is inherited; invisibility is reduced to something mundane—unvoiced linguistic activity, like a letter locked in a drawer read only by its writer; uncertainty is rewritten as the consequence of information reaching one party and not another. The paper tightens a prior ambiguity: **there is no "felt sense of certainty" separable from knowing itself—that the subject knows their own inner monologue just is the complete account of certainty.** It advances to its tightest claim: **the inner monologue just is the subjective experience—by comparison with the objective view another takes of me; there is no further question of how the inner monologue is subjectively felt.** Subjective/objective mark not two kinds of thing but one event named differently by reference frame; no bridge is needed, only a shift of vantage. The paper then squarely engages the strongest historical rebuttal to this path—the conceivability of philosophical zombies—and shows it self-contradicting: suppose a world of zombies able to describe their surroundings via inner monologue; if that description is genuine, then, by the established identity (inner monologue just is subjective experience), the zombie is already feeling that world, undoing the stipulation that it "feels like nothing"; if the description is not genuine, the zombie's founding stipulation of functional-physical identity is itself unmet. Moreover, our epistemic access to a zombie's first-person viewpoint and to any other person's are perfectly symmetrical; if inaccessibility alone licenses doubting the zombie's feeling, that doubt must fall equally on every other person, absent some non-arbitrary criterion of distinction. The paper thereby diagnoses the hard problem's true ailment: **it asks "why can another have consciousness" while never arguing for what it quietly presupposes—that another possesses something equivalent to my consciousness; this is the emperor's new clothes, pretending the emperor is dressed and then gravely debating why the garment is so hard to see, when no garment was ever there.** Another plainly has only external, purely physical activity; the concept of the philosophical zombie is, in truth, an extreme form of solipsism—"consciousness," within it, is merely another name for "my consciousness," and "another's consciousness" was never independently defined at all. The paper connects this to a deeper structure: the hard problem asks why, beyond the subject that is the brain/body, and beyond the cause that is the brain/body, there must be yet another subject, another cause—**but there is no such further subject or cause**; this is the same error, in its third guise, already dismantled elsewhere as the extra controller standing outside the frame and the extra impetus posited beyond the universe-state. The paper then supplies an accounting of the noun "soul": it is the nominalization of three real, first-personal phenomena—**a sense of agency, a sense of causation, a sense of dissociation**—compressed into a single noun, exactly as "anger" metaphorically reifies a felt experience into a container-held solid; this metaphor is **valid only for oneself**, since only I possess those three phenomena as its ground; transplanting the same word onto another borrows the name without borrowing the experiential ground that made the name apt. The paper here remains fair: this diagnosis forcefully dismantles the zombie world as a specific stipulation, but the true boundary has not vanished, only receded to a deeper, more technical methodological question—even granting the zombie's self-contradiction, the explanatory-gap tradition may still ask whether **a priori, conceptual analysis alone**, without presupposing the very identity "inner monologue = subjective experience," can derive that subjective experience must exist from a complete physical/functional description; this identity is, at present, asserted and argued for, not independently, non-circularly derived—a question one level deeper than zombie-conceivability, which the paper presents fairly and adjudicates on. On this foundation the paper advances a sharper inference: **whether another's subjective experience truly exists appears more doubtful under this tightened definition.** The strongest prior evidence—"he plainly thinks"—has now been shown to just be inner monologue, just be subjective experience, and is withdrawn entirely; what remains is a weaker analogy, and "another's consciousness" is now diagnosed as an unexamined transplant of a name whose only ground was ever first-personal. The paper guards several lines with particular gravity: this does not touch first-person certainty; "more doubtful" is by no means "disproven"; and none of this licenses, in the slightest, disregarding another's reported suffering, feeling, or welfare. How the hypothesis of another's subjective experience ought ultimately to be treated, the paper renders no verdict on, leaving it expressly suspended as **still under study**. This very paper is itself a stretch of language generated by this course of thinking, an externalization of an inner monologue; it is a finger pointing at the moon; the finger may be discarded, the moon may not.

本文处理一次概念上的改写，及它引出的一个更尖锐的后果。判断的第一步：如果把他人的想法定义为他的内心独白，那这就破解了意识的难问题——我们不再需要假设他人具有内心独白之外的”意识”，而内心独白可被视为一种类似说话的纯物理过程。 意识的难问题问的，从来不是”大脑如何处理信息、如何做出反应”（这些原则上可被物理科学解决），而是——为什么这些物理过程，会伴随一份主观的、“感觉起来是什么样子”的体验；这份体验，似乎是任何物理描述都触不到的东西，这正是”他人是否有意识”看似永远无法从外部验证的根源。本文指出，这一记洞见的招法，不是去解释这份”感觉起来”如何从物理中产生，而是替换被追问的对象：若”他人的想法”被严格定义为”他的内心独白”，而内心独白（体验作为输入生成出的语言）本身是纯物理的、和说话同一类的事件，那么”他人是否有想法”这一问题，便完全落在了物理科学的辖区之内，不再需要假设一个躲在独白之外、永远无法验证的”意识”。本文守一处诚实的限定：这改写了难问题的一个版本，不是解答了难问题原初所指——若难问题问的，正是”想法”这个词想指向、却始终没能指准的那份”感觉起来”的东西，本文的论证并未击倒它，只是把追问的对象，换成了它旁边一个更容易被物理描述覆盖的邻居。本文进而发展一记更尖锐的推论：他人的主观感受是否真的存在，在这一新定义下，显得更加可疑。 其论证如下：此前，“他显然在思考、在反思”，是我们相信他人拥有主观体验的主要证据来源之一——一个能思考、能反省自身处境的存在，直觉上强烈暗示着”里面有人在”。可一旦思考本身被证明可以完全由内心独白这一纯物理的生成过程解释——他所说的每一句”我在想”“我感觉”，都只是体验流经身心之后直接产出的语言输出，不需要一份额外的主观感受来伴随它——这一主要证据来源，便被整个抽走了：思考不再需要意识来解释，思考本身即是物理过程的完整产物。抽走这一证据之后，支持”他人有主观体验”这一信念的，只剩下一个更弱的类比推理——他在生理与行为上与我相似，故大概也和我一样有主观体验；而这一类比推理，本就弱于”他显然在思考、故必有体验”这条此前被默认的、更直接的推论。加上简约原则的压力：若一个人可被观察、可被报告的一切事实（包括他说出”我感觉到了”这句话本身），都能被一套不假设主观感受的纯物理机制完整解释，那么额外假设一层不承担任何解释功能的”感受”，便成了一个缺乏理由的多余假设。本文由此得出：主观感受这一假设，其证据基础，在这一新定义下，确实变得薄弱了。本文守数条须格外郑重划定的界线。其一，这一论证不触及第一人称的确定性——你此刻真实地感觉到些什么，这一点，不因任何关于他人的论证而受影响；本文所动摇的，只是”他人（第三人称）是否真有主观体验”这一信念的证据基础，从不涉及你对自己此刻体验的确定性。其二，“更可疑”绝不等于”已被证伪”——本文没有证明他人没有主观体验，只是指出，我们此前相信这一点的主要理由，已经不再成立，剩下的理由更弱；这是一个关于证据强度的论断，不是一个关于事实真假的断言。其三，也是最须郑重守住的一条：这一论证绝不构成忽视他人报告之痛苦、感受或福祉的许可证——一个人说他在痛苦，无论支撑”他确实有主观体验”这一信念的哲学证据有多薄弱，这句话在实践中该如何被对待、该给予怎样的回应，是一个完全独立的伦理与实践问题，本文的论证半分不能被拿来当作忽视、轻慢、或推卸对他人痛苦之回应的理由；把一个关于证据强度的哲学论断，直接兑现为一次具体的、关于该不该在意他人的实践裁决，是本文明确拒绝背书的一步误用。至于他人的主观体验这一假设最终该被如何对待、这对我们的实践与伦理该有什么后续影响——本文不下判决，明确将其悬置为尚待研究。本文这一篇，自己也是由这场思考生成的一段语言；它是指月之指；指可弃，月不可弃。 Abstract This paper treats a conceptual redefinition, and a sharper consequence it draws out. The first step of the judgment: if we define another’s thought as their inner monologue, this dissolves the hard problem of consciousness—we no longer need to posit that another possesses a “consciousness” beyond the inner monologue, and the inner monologue can be regarded as a purely physical process akin to speaking. The hard problem never asked “how does the brain process information, how does it produce responses” (matters that physical science can, in principle, resolve); it asked why these physical processes are accompanied by a subjective experience of “what it is like”—an experience that seems to elude any physical description, and this is precisely why “whether another has consciousness” appears forever unverifiable from outside. The paper argues that this insight’s maneuver is not to explain how “what it is like” arises from the physical, but to substitute the object being asked about: if “another’s thought” is strictly defined as “their inner monologue,” and the inner monologue (language generated with experience as input) is itself purely physical, an event of the same kind as speaking, then “does another have thoughts” falls entirely within the jurisdiction of physical science, with no need to posit a “consciousness” hiding beyond the monologue, forever unverifiable. The paper holds one honest qualification: this reframes one version of the hard problem; it does not answer what the hard problem originally meant—if the hard problem asks about that very “what it is like” that the word “thought” was trying, and failing, to point at, this paper’s argument has not knocked it down, only substituted, in its place, a neighboring object more amenable to physical description. The paper then develops a sharper inference: whether another’s subjective experience truly exists becomes more doubtful under this new definition of thought. The argument runs as follows. Previously, “he plainly thinks, plainly reflects” was one chief piece of evidence for believing another possesses subjective experience—a being capable of thought, of reflecting on its own situation, strongly suggests, on intuition, that “someone is in there.” But once thinking itself is shown to be fully explicable by the purely physical generative process of inner monologue—every “I think,” every “I feel” he utters being nothing but a linguistic output directly produced once experience has passed through the body and mind, needing no additional subjective feeling to accompany it—this chief piece of evidence is entirely withdrawn: thinking no longer requires consciousness to explain it; thinking itself is the complete product of a physical process. With this evidence withdrawn, what remains to support the belief that “another has subjective experience” is only a weaker analogical inference—he resembles me physiologically and behaviorally, so he probably has subjective experience as I do—and this analogical inference was always weaker than the more direct, previously assumed inference “he plainly thinks, therefore he must have experience.” Add the pressure of parsimony: if every observable, reportable fact about a person (including his very utterance “I feel it”) can be fully explained by a purely physical mechanism that posits no subjective feeling, then positing an additional layer of “feeling” that performs no explanatory work becomes a superfluous hypothesis lacking justification. The paper thus concludes that the hypothesis of subjective experience, under this new definition, does see its evidential basis weakened. The paper draws several lines that must be guarded with particular gravity. First, this argument does not touch first-person certainty—that you presently, really, feel something is unaffected by any argument about others; what this paper unsettles is only the evidential basis for the belief that “another (in the third person) truly has subjective experience,” never your certainty about your own present experience. Second, “more doubtful” is by no means “disproven”—the paper has not shown that others lack subjective experience, only that the chief reason we previously believed this no longer holds, and what remains is weaker; this is a claim about the strength of evidence, not a claim about the truth of the matter. Third, and most gravely to be guarded: this argument is by no means a license to disregard another’s reported pain, feeling, or welfare—that a person says he is suffering, however weak the philosophical evidence for “he truly has subjective experience” may be, how this utterance ought to be treated in practice, what response it warrants, is an entirely separate ethical and practical question; not one particle of this paper’s argument may be used as a reason to neglect, belittle, or evade a response to another’s suffering; converting a claim about evidential strength directly into a concrete practical verdict about whether another’s welfare matters is a misuse this paper expressly refuses to endorse. As to how the hypothesis of another’s subjective experience ought ultimately to be treated, and what consequences this should have for our practice and ethics—the paper renders no verdict, and expressly leaves this suspended as still under study. This very paper is itself a stretch of language generated by this course of thinking; it is a finger pointing at the moon; the finger may be discarded, the moon may not.

本文处理我们几乎从未怀疑过的最后一处栖身之地——那个此刻正在心里说话的声音。通常的图景是：先有体验，然后”我”，一个站在体验之外、能自由组织语言的说话者，对这份体验做出评述、反应、叙述；内心独白，被认为是这个自由说话者的作品。本文的判断是：体验，被用来生成内部语言，或者说内心独白可被视为由体验作为输入而生成。 这不是说独白由体验”引起”或”促成”这么温和的说法，是一次彻底的身份置换：没有一个站在体验之外、自由组织这段独白的说话者；那段独白本身，就是体验这一输入，流经某个过程后，直接产出的一个输出——和这具身体对一份刺激产生一次神经放电、一次心跳加速，是同一种性质的事。心里那个”声音”，不是自由的作者，是又一个下游产物。本文用一个可逆性的论证来支撑这一翻转：说”我在叙述我的体验”，预设了”我”是主动的叙述者，体验是被动的、待处理的素材；可反过来讲同样成立、同样讲得通——“体验主动地生成了这段内心独白，而那个说话的声音，不过是被动地、机械地把它吐出来的那个环节”。两种讲法，落在同一个物理过程上，一样对：没有一处刻着”是我在主动叙述”还是”是体验在主动生成叙述”。而这根箭头的朝向能够反转、且反转后同样通顺，恰恰证明它不在实情里，是我们讲述时习惯性地贴上去的——我们默认把”主动叙述者”这个身份，理所当然地派给了内心独白，这份默认本身，就是一次错置。本文进而将这一翻转，接回一条更根本的判断：思维本就是在被给予的诸时空局部之间连线的活动，这些线并不实存；而内心独白，正是这套连线操作变成了可被听见的语言——它不是先于思维、独立于思维的一个”我”在说话，它就是连线本身的语音化产物。说”我在想”，与说”这根线正在被连出来”，是同一件事的两种讲法。本文由此指出，内心独白，正是我们把体验拿去追问”我从中能学到什么、该如何应对、该如何规划”这一整套操作，得以运转的具体的、可听见的现场——那台把体验收编为原材料的机器，不是一个抽象的隐喻，它就是此刻正在心里说话的这个声音本身。本文进而把这一翻转，从第一人称推广到他人：支撑它的可逆性论证，从未预设只对第一人称有效，把”我”换成”他”，同样成立——这颠覆了我们感觉他人的想法是由他人自由产生的这一”常识”；我们对自己和对他人的心智，用的是同一套”自由发起者”模型，而这正是我们借以归责、归功、信任他人的地基。本文并就此做出一处方法论定位：这一整套论证，达到了和决定论同样的伦理学效果，而不必诉诸决定论为前提——它不依赖”宇宙下一状态是否被上一状态唯一决定”这一悬而未决的物理学赌注，只依赖一个可在每一具体案例中当场检验的观察（叙述方向可逆、故不实存），因此不会因物理学在决定论问题上的任何进展而受损。本文守一处分寸：说内心独白是生成的输出、不是独立自由的叙述者，绝不是说这段独白是假的、不该被信、思考本身不值得认真对待。那段独白，真在生成、真在发生，和体验本身一样真实——被拆的，只是”这声音是站在体验之外、自由发起叙述的独立主体”这一身份设定，不是”这声音真的响着、真的在说着这些话”这个实情。认清”内心独白是生成的、不是自由发起的”，与照常把它当作思考、规划、自我理解的重要工具去使用，是并行不悖的两件事；而将此推广到他人，也绝不构成”他人不必为自己的想法与行为负责”这一裁决，责任究竟该如何安放，仍明确悬置为尚待研究。本文这一篇，自己也是由写作这一体验生成的一段语言、一根指月之指；指可弃，月不可弃。 Abstract This paper treats the last refuge we have almost never thought to doubt: the voice presently speaking in one’s head. The usual picture holds that experience comes first, and then “I”—a speaker standing outside experience, free to compose language—comment on, react to, narrate that experience; the inner monologue is taken to be this free speaker’s work. The paper’s judgment: experience is used to generate internal language, or, put otherwise, the inner monologue may be regarded as generated with experience as its input. This is not the mild claim that the monologue is “caused” or “prompted” by experience; it is a thoroughgoing exchange of identity: there is no speaker standing outside experience, freely composing this monologue; the monologue itself is an output directly produced once experience, as input, has passed through some process—of the same kind as this body producing a burst of neural firing, an acceleration of the heartbeat, in response to a stimulus. The voice in one’s head is not a free author; it is one more downstream product. The paper supports this reversal with an argument from reversibility: to say “I narrate my experience” presupposes that “I” am the active narrator and experience the passive material awaiting processing; yet the reverse holds equally well and is equally coherent—“experience actively generates this inner monologue, and the speaking voice is merely the passive, mechanical link that expels it.” Both tellings fall on one and the same physical process and are equally correct: nothing marks which is “I actively narrating” and which is “experience actively generating a narration.” That this arrow’s direction can be reversed, and remains just as coherent when reversed, is precisely what shows it is not in the facts but pasted on by habit of telling—we default to assigning the identity of “active narrator” to the inner monologue as a matter of course, and this very default is itself a misplacement. The paper further connects this reversal to a more fundamental judgment: thinking just is the activity of drawing lines between the spatiotemporal locals of what is given, and these lines do not exist; the inner monologue is precisely this line-drawing operation turned audible—it is not some “I,” prior to and independent of thinking, doing the speaking; it just is the vocalized product of the connecting itself. To say “I am thinking” and to say “this line is presently being drawn” are two ways of saying one and the same thing. The paper thereby argues that the inner monologue is the concrete, audible site at which the whole operation of taking experience and pressing it into “what can I learn from this, how should I respond, how should I plan” actually runs—the machine that folds experience into raw material is no abstract metaphor; it is this very voice presently speaking in one’s head. The paper keeps one measure: to say the inner monologue is a generated output, not an independent free narrator, is by no means to say this monologue is false, unworthy of trust, or that thinking itself does not deserve to be taken seriously. That monologue really is generated, really does occur, and is every bit as real as the experience itself; what is dismantled is only the identity-setting that “this voice is an independent subject standing outside experience, freely initiating narration,” not the fact that “this voice really sounds, really is saying these words.” To recognize that “the inner monologue is generated, not freely initiated” and to go on using it, as usual, as an important tool for thinking, planning, and self-understanding, are two parallel-not-conflicting things. This very paper is itself a stretch of language generated by the experience of writing it, a finger pointing at the moon; the finger may be discarded, the moon may not.

本文追一条收编的链条，分三层，一层比一层收得更紧，直至收进这篇论文自身写作的这一刻。第一层：当我们计划未来时，或自以为在按照对未来的规划安排当下时，当下成为实现未来的工具，当下自己的身体也成了实现未来计划的工具，这时身体的感受、状态便被相对忽略——除非状态差到提升该状态本身成为一个跨时间计划的中心。 身体一旦被纳入”实现未来”这一功能位置，其性质便从”正被经历者”变成”正被使用者”；工具不需要被关照此刻的感受，只需要被关照还撑不撑得住功能，感受便因此被相对忽略。第二层，同一机制推广到他人：我们对他人的关心，同样也是如此——我们不是直接关注他人的状态和感受，而是把它们纳入成为一个系统里的变量：感受到对方的情绪后，关注的是这一情绪会把互动带向何方，从而思考自己该如何反应，而不是直接关注情绪本身，去体验、理解、共情它。本文格外强调这一层最锋利的一步：如果体验、理解、共情的目的仍是自己可以应对得更好，则这仍是工具化——工具化的分界，不在动作是否细致（哪怕最投入的共情，也可以是工具化流程里更精密的一道工序），而在这份体验完成之后，有没有被立即征用去服务别的目的；而头脑中的伦理道德系统——如何应对才符合道德、或避免被批评——正是驱动这层工具化最重要的引擎之一，道德关切非但没有豁免于工具化，反而常是它最隐蔽、最难被自我怀疑的入口。第三层，也是最深的一层，收窄到体验本身：当我们把自己所有的体验，都视为”我从中可以学到什么？理解宇宙的哪个性质？如何利用学到的来更好地规划？”，体验便又沦为工具，体验本身——除其工具性之外——被视为无意义、无价值、无重要性、无关痛痒。 这一层不再需要一个外部对象（未来、他人）来充当工具化的落点，它直接把正在发生的这份经历，就地转成了可供提炼的原材料。而本文进而指出，连从体验中总结”真正合理的道德准则应是什么”、或关注体验中所蕴含的道德是非，这些活动，仍是把体验工具化——而主流伦理学实践鼓励我们这么做：伦理学作为一门学科，其整套操作方式，正是系统性地教人将具体、活生生的体验，蒸馏成抽象、可迁移的道德准则；它不但不豁免于工具化，反而可能是工具化中最有声望、最难被质疑的一套建制。三层合看，是同一个结构在不同对象上的重演：把一个本该作为目的的东西，收编进一个面向别处的系统，充当那个系统的输入变量。本文守一处极不寻常的分寸，须以异常的诚实来说：这一记揭示，没有为”反思、提炼、求道”这类活动留出一个安全区——本文自身，正是从一次体验（这一记洞见被给出、被思考的过程）中，提炼出一个可携带、可复用的一般机制，并把它写成了一篇论文；这个写作过程本身，与第三层所描述的那台机器，并无二致。本文没有一个漂亮的办法，能让这一步豁免于它自己所揭示的收编；试图论证”但这一次的提炼是不同的、是被允许的”，本身也会是又一次把这处境收编进一个自我豁免的规则里，从而成为这台机器的又一次运转。故本文不假装已经解决这个问题，只把这一处境如实说出，让它留在那里：任何一次”看清工具化”的反思，只要仍然是为了得出些什么、用来调整些什么，便仍在这台机器的运转范围之内；这本身，或许不是一个等待被解决的技术问题，是这份揭示自己给自己留下的、无法被绕开的边界。本文这一篇，自己也是指月之指，也不例外；指可弃，月不可弃——只是这一次，连”弃”这个动作，若也是为了达成某种更干净的姿态，恐怕仍是同一台机器的又一转。 Abstract This paper traces a chain of reduction across three layers, each narrower than the last, ending in the very moment of this paper’s own writing. First layer: when we plan the future, or believe ourselves to be arranging the present according to a plan for the future, the present becomes a tool for realizing that future, and the present body of oneself likewise becomes a tool for realizing one’s own future plan; at that point the body’s sensations and state are relatively neglected—except when the state deteriorates to a point where improving it itself becomes the center of a new, cross-temporal plan. Once the body is folded into the functional slot of “realizing the future,” its character shifts from “presently being experienced” to “presently being used”; a tool need not be attended to for how it feels now, only for whether it still holds up functionally, and sensation is thereby relatively neglected. Second layer, the same mechanism generalized to others: our concern for other people is likewise so—we do not attend directly to another’s state and feelings, but fold them into a variable within a system: upon sensing another’s emotion, what we attend to is which direction this emotion will carry the interaction, so as to work out how we ourselves should respond, rather than attending directly to the emotion itself—experiencing, understanding, empathizing with it. The paper stresses, with particular sharpness, the finest point here: if the purpose of experiencing, understanding, or empathizing remains that one can thereby respond better, this too is instrumentalization—the dividing line of instrumentalization is not whether the act is finely done (even the most engaged empathy can be a more refined step within an instrumentalizing process), but whether, once that experience is complete, it is immediately conscripted to serve some further end; and the ethical system in our heads—how to respond so as to be moral, or to avoid criticism—is one chief engine driving this layer of instrumentalization; moral concern is not merely unexempt from instrumentalization but is often its most covert, least self-questionable point of entry. The third layer, and the deepest, narrows to experience itself: when we take all our own experience to be “what can I learn from this? which property of the universe does it help me understand? how can I use what I’ve learned to plan better?”, experience is once again reduced to a tool, and experience itself—apart from its instrumental value—is taken to be meaningless, worthless, of no importance, of no consequence. This layer no longer needs an external object (the future, another person) to serve as the site of instrumentalization; it directly converts the experience presently occurring into raw material fit for distillation. The paper further argues that even distilling from experience “what the truly reasonable moral principles ought to be,” or attending to the moral rights and wrongs embedded in what is experienced, remains an instrumentalization of experience—and mainstream ethical practice encourages us to do precisely this: ethics as a discipline, in its whole mode of operation, systematically teaches one to distill concrete, lived experience into abstract, transferable moral principles; far from being exempt from instrumentalization, it may be among the most prestigious, least questionable institutions of it. Taken together, the three layers are one and the same structure recurring on different objects: something that ought to stand as an end is folded into a system oriented elsewhere, serving as that system’s input variable. The paper keeps one measure of an unusual sort, which must be stated with unusual honesty: this revelation leaves no safe zone for activities such as reflection, distillation, or the pursuit of understanding—this very paper is itself distilled from an experience (the process by which this insight was given and thought through) into a portable, reusable general mechanism, written up as a paper; this act of writing is, in itself, no different from the very machine described in the third layer. The paper has no clean device by which this step could be exempted from the very reduction it reveals; to argue that “this particular instance of distillation is different, is permitted,” would itself be to fold this predicament into a self-exempting rule, and so would itself be one more turn of the same machine. So the paper does not pretend to have resolved this, and simply states the predicament as it is, letting it stand: any reflection that “sees through” instrumentalization, so long as it still aims at arriving at something, at being used to adjust something, remains within the operating range of this same machine; this, perhaps, is not a technical problem awaiting solution, but a boundary the revelation leaves for itself, one that cannot be stepped around. This very paper is itself a finger pointing at the moon, and is no exception; the finger may be discarded, the moon may not—except that, this time, even the act of “discarding,” if aimed at achieving some cleaner posture, is likely still one more turn of the same machine.

本文追一条链条，它从一件看似无可指摘的事出发——对道德的关注——却经过若干个环节，闭合成它自己的反面。本文的判断是：对于道德的关注，恰恰可能引发不道德的行为。 这条链条有六环。其一，人关注道德，想要避免做出不道德的行为。其二，要避免不道德的行为，首先需要判断什么行为是不道德的——这是行动之前必须先解决的问题。其三，这种判定本身是不可能的：没有一个稳定、可靠的程序，能给出”这个行为到底道不道德”的确定答案，因为道德色彩本身可以随叙述而任意游移。其四，判定不可能，人便退而求其次，在心中模拟众人最可能使用的判断标准——拿一个可以设想的东西（众人大概会怎么评判），去顶替一个够不着的东西（道德本身的实情）。其五，也是本文要格外精细拆解的一步：这层模拟，悄悄退化了——从”设想众人可能会怎么评判”，滑成了”单纯对众人批评的恐惧与避免”。这两者看似无缝衔接，实则已把关切的对象整个换掉：前者好歹还留着一层对”该怎样”的关切，哪怕只是猜众人的标准；后者已经不再问”这样做对不对”，只问”这样做会不会被骂”——而”对不对”与”会不会被骂”，是可以完全脱钩的两件事。其六，在这份恐惧的压力下产生的行为，必然带来诸多弊端：怯懦（该做、会招骂的事不敢做）、从众（谁在看就顺谁的意）、推诿（把责任先一步推给别人以自保）、形式主义（表演看得见的合规、忽视不显眼却要紧的善）、该出手时因怕被挑错而袖手——而这些弊端，在另一种视角下，恰恰就是不道德。 闭环由此完成：出发于对道德的关注，途经”判定不可能→退而模拟众人→模拟退化为恐惧→行为变形”，最终落地的，是另一套标准下的不道德行为。本文守两条须格外郑重划定的界线。其一，这绝不该被读成”故不该关注道德、关注道德必然有害、干脆别管”——这条链条之所以走到不道德，病灶恰恰在第三、五两步：误以为必须先拿到一个确定的判定才能行动，以及任由那层设想不知不觉退化成纯粹的恐惧；这是处理”判定不可能”这一困境的一种特定、且拙劣的方式，不是关注道德的必然后果——本文自身一贯采取的做法（面对不可判定，明确承认、悬置，不假装已有答案、也不用某个代理去冒充答案）恰是这条病态链条的一个反例，证明退化并非唯一出路，只是常见的一条歧路。其二，这绝非一把可以debunk任何道德关切或任何一次批评的万能钥匙——不能拿它去说”你对我的批评，不过是众人从众恐惧的产物，不是真的道德判断，所以我可以不理会”；本文诊断的是一种机制，不是一张可以扣在任何一次具体关切或批评头上、宣布其”其实只是恐惧、故不算数”的标签。至于”人在面对不可判定的道德困境时，究竟该如何行事，才不至于滑向这条退化的链条”——本文对此不下判决，明确将其悬置为尚待研究。本文这一篇，自己也是被这同一种关切驱动写就的，也不豁免于它自己所揭示的风险；它是指月之指；指可弃，月不可弃。 Abstract This paper traces a chain that starts from something seemingly beyond reproach—concern for morality—yet, through several links, closes into its own opposite. The paper’s judgment: concern for morality can, precisely because of that concern, give rise to immoral conduct. The chain has six links. First, a person cares about morality and wishes to avoid immoral acts. Second, to avoid immoral acts, one must first judge which acts are immoral—a question that must be settled before acting. Third, this judgment is itself impossible: there is no stable, reliable procedure yielding a determinate answer to “is this act immoral,” since moral color itself can be made to drift arbitrarily with narration. Fourth, the judgment being impossible, a person falls back on a second-best: simulating in the mind the standard of judgment the crowd would most likely use—substituting something conceivable (a guess at how the crowd would judge) for something unreachable (the fact of morality itself). Fifth, and this is the step the paper dissects with particular care: this simulation quietly degrades—sliding from “imagining how the crowd would likely judge” into “the mere fear of, and avoidance of, the crowd’s criticism.” The two appear seamlessly continuous, yet the object of concern has, in fact, been entirely swapped: the former at least retains a layer of concern with “what ought to be,” even if only guessing the crowd’s standard; the latter no longer asks “is this right or wrong” at all, only “will this get me criticized”—and “right or wrong” and “will get me criticized” can come apart completely. Sixth, conduct produced under the pressure of this fear inevitably brings a host of defects: cowardice (not daring to do what ought to be done, if it risks criticism), conformity (going along with whoever is watching), blame-shifting (pushing responsibility onto others first, for self-protection), formalism (performing visible compliance while neglecting an unobtrusive but weightier good), and failing to act when one should, for fear of being faulted—and these defects, under another lens, are themselves precisely immoral. The loop closes: starting from concern for morality, passing through “judgment impossible → falling back on simulating the crowd → the simulation degrading into fear → conduct deformed,” what lands, in the end, is immoral conduct by another standard. The paper guards two lines that must be drawn with particular gravity. First, this must by no means be read as “so one should not care about morality, since caring necessarily does harm, so best not to bother”—the pathology of this chain lies precisely in its third and fifth links: mistakenly assuming a determinate judgment must be secured before acting, and letting that imagined proxy degrade unchecked into pure fear; this is one particular, poor way of handling the predicament of “judgment being impossible,” not a necessary consequence of moral concern as such—this paper’s own consistent practice (facing the unjudgeable, openly acknowledging it, leaving it suspended, neither pretending an answer exists nor letting some proxy pose as one) is itself a counterexample to this pathological chain, proof that degradation is not the only path, only a common detour. Second, this is by no means a universal key for debunking any moral concern or any criticism whatsoever—one cannot use it to say “your criticism of me is merely the product of herd-fear, not real moral judgment, so I need not heed it”; the paper diagnoses a mechanism, not a label to be stamped on any concrete concern or criticism to declare it “really just fear, hence void.” As to how a person, facing an unjudgeable moral predicament, ought actually to act so as not to slide down this degrading chain—the paper renders no verdict on this, and expressly leaves it suspended as still under study. This very paper is itself written under the same kind of concern it describes, and is not exempt from the very risk it reveals; it too is a finger pointing at the moon; the finger may be discarded, the moon may not.

本文处理一个比”影响一旦被讲述、我们凭能否讲通去判定它存在”更上游的问题：在因果关系被讲述出来之前，是什么决定了它会不会被讲述？ 本文的判断是：因果关系的发现——其本质是发明，姑且仍称为发现——如果不能被涂抹上道德的色彩（好/坏），就不会引起我们的兴趣，会被搁置，直到某一天我们发明出涂抹道德色彩的方式，或者说，直到我们确定了它的道德色彩（黑/白）。 这不是说因果之线本身是连出来的这件事有多新——本文承认这一点，直接以”本质是发明”带过——而是要指出一层更隐蔽的筛选：在无穷多条本可以被连出来的因果线里，只有能被涂色的那些，才会真被连出来；涂不上色的，就沉在无数未被讲述的可能性里，仿佛它们不存在，直到某天，一种涂色的办法被发明出来，它们才会从这片沉默里显影。而一旦某条因果路径真被讲述、真被涂上了色，本文指出，涂色的方式极其精微，叙述上的毫厘之差，会让最终的道德性质失之千里。设想一条因果路径上，串起了某人的一串行为：说它们全是无意的，与说它们是有意为之，截然不同；说其中一个是有意的、其余无意，又与”全有意”或”全无意”两种极端都不同——这是涂色的第一重精微，意图的归派。而即使我们说这串行为全是无意的，只要进一步指出”他本可以通过某种方式避免”——譬如小学时若曾努力学习某种知识——这串行为的道德性质，便与我们未想到这一层时截然不同；这是第二重精微，反事实的插入：一条”本可以不这样”的想象岔路，不改变任何”实际发生了什么”的事实，却足以让整件事的道德分量彻底翻转；而这层追溯，理论上没有天然的止点，可以无限地往前追问。两重精微合起来，得出本文的落点：同一条客观发生过的因果路径，本质上，可以被讲成几乎任意一种道德色彩。 本文对此划定一条须格外郑重守住的界线，比以往任何一次揭示都要守得更紧：这只是揭示涂色机制本身的可操纵性，绝非推出以下两条中的任何一条。其一，绝非道德虚无主义——不是说”既然道德性质可以任意界定，一切道德判断便都是虚构、没有一个比另一个更对、故怎么判都行”；一串行为是否有意、他是否本可避免，仍然是关于实情的问题（哪怕认知上极难确定），不会因为判定的机制易被操纵，就使被判定的事实本身失去真假可言。其二，绝非一把可以现用的操纵工具——不是说”既然涂色可以任意，我便可以把自己的行为讲成全然无意且绝不可能避免，把他人的行为讲成全然有意且轻易可以避免”；本文揭示这套机制为何精微、为何易被推向任何方向，不为任何一次具体的涂色操作背书，不论那操作是指向他人的归咎，还是指向自身的开脱。至于因果的发现该如何摆脱道德兴趣的筛选、涂色的精微该如何被校准、道德判断究竟该如何安放——本文一步不踏，明确将其悬置为尚待研究。本文这一篇，自己也是一次被讲述出来的因果叙事、一根指月之指；指可弃，月不可弃。 Abstract This paper treats a question more upstream than “once an influence is told, we judge its existence by whether it can be told coherently”: before a causal relation is ever told, what determines whether it gets told at all? The paper’s judgment: the discovery of a causal relation—essentially an invention, though we may still call it discovery—if it cannot be painted with a moral color (good/bad), fails to interest us and is set aside, until the day we invent a way to paint it, or, put otherwise, until we settle its moral color (black/white). This is not merely to restate, once more, that causal lines are drawn rather than found—the paper grants this and moves past it—but to point to a subtler filter: among the innumerably many causal lines that could in principle be drawn, only those that can be painted are actually drawn; those that cannot are left submerged among the countless undrawn possibilities, as though they did not exist, until some day a method of painting is invented, and they surface from that silence. And once a causal path is actually told, actually painted, the paper argues, the manner of painting is exceedingly fine-grained, such that the slightest difference in how it is narrated sends its moral quality worlds apart. Consider a causal path stringing together a series of someone’s acts: to say they were all unintentional, and to say they were done on purpose, are utterly different; to say one was intentional and the rest were not differs again from either extreme—this is the first fineness, the attribution of intention. And even granting that the whole string was unintentional, merely pointing out that “he could have avoided it by some means”—say, by studying some subject diligently back in primary school—shifts the moral quality of the entire string from what it was before this layer was considered; this is the second fineness, the insertion of a counterfactual: an imagined side-path of “it could have been otherwise” changes nothing about what actually happened, yet suffices to flip the moral weight of the whole matter, and this regress, in principle, has no natural terminus—it can be pressed backward without limit. Together the two finenesses yield the paper’s landing: one and the same objectively occurred causal path can, in essence, be told in almost any moral color whatsoever. The paper draws one line here that must be guarded more firmly than any prior revelation in this series: this is only a revelation of the manipulability of the coloring mechanism itself, and by no means license for either of the following. First, it is by no means moral nihilism—it does not say that “since moral quality can be arbitrarily defined, all moral judgments are fictions, none more correct than another, so anything goes”; whether a string of acts was intentional, whether he could have avoided it, remain questions about the facts (however hard, cognitively, to settle), and do not lose their truth-value simply because the mechanism of judgment is manipulable. Second, it is by no means a ready-made tool for manipulation—it does not say that “since coloring can be arbitrary, I may as well narrate my own acts as wholly unintentional and utterly unavoidable, and another’s as wholly intentional and easily avoidable”; the paper reveals why this mechanism is so fine-grained, so easily pushed in any direction, without endorsing any single concrete act of coloring, whether aimed at blaming another or excusing oneself. As to how the discovery of causation might be freed from this filtering by moral interest, how the fineness of coloring ought to be calibrated, and how moral judgment ought actually to be situated—the paper takes not one step, and expressly leaves all of this suspended as still under study. This very paper is itself a causal narrative that has been told, a finger pointing at the moon; the finger may be discarded, the moon may not.

本文处理一个翻转，它比"认知边界常被误当成本体边界"这一诊断更进一步：**影响，只要能够被描述，就被当作存在；反过来，我们感觉存在的影响，也只是能够被我们描述的影响。** 这不是说我们有时把边界看错了地方，是说：我们判定"某处是否存在一份影响"的**唯一操作性标准**，从头到尾就是"能不能把它讲通"——可描述性不是发现实情的透镜，是我们认定实情的全部依据，没有第二条独立的通道。本文进而指出，这道"能否讲通"的门槛，远比我们通常以为的**容易跨过**，并给出两重机制。其一：**甲的影响并不需要有某种特定的甲独有的烙印，才算是甲的影响。** 甲只需推动一个通用的、无签名的中间变量——譬如乙体内某种化学物质达到某一水平——这个变量便会弥散地影响乙之后做的**所有**事，而不是某一件对它敏感的具体事，因为没有人能证明其中哪一件**没有**受它影响。这是一次举证责任的倒置：不必正面证明"甲影响了这件具体的事"，只需指出一个通用变量存在、且无法被排除，这本身便足以构成一个讲得通的影响叙事——影响因此比我们通常认为的，广泛、普遍得多。其二：这套机制还能从体内的化学水平，推广到整个宇宙的物理演化。设想我从一处环境经过，之后该处有人犯了罪；这犯罪承接着我经过时的那个事件——若我没经过，宇宙状态便会不同，不同的状态演化下去，也许就不出现那桩罪——这也讲得通：这是说不同的起点会导致不同的进程，而不需要明确描述出具体是如何导致的。本文进而指出这套机制在道德实践里一处贴身的、双重标准的缝隙：**心情会导致行为是常识**，我们对这条因果链毫无阻力地接受；可我们通常不会追究是什么"导致"了这份心情，尤其当那导致心情的前因，本身**道德上中性**（天气、一句无关的话、堵车五分钟）——不是这些中性事件真的没有因果作用，是我们选择不去追问，而这一止步，被我们误读成了"因果关系本来就在此断了"。本文对此揭示的机器，须格外郑重地守住一条界线，比以往任何一次揭示都要守得更紧：一段影响叙事若能被讲通，**绝不等于**它就该被采信、该被拿去实际分配或推卸责任。这台机器最危险之处，在于它可以指向任何方向——它既可以被用来说服他人相信"你的行为其实是我造成的"，也可以被用来说服自己相信"我做的事其实不该怪我，是某个说不清的前因导致的"；本文所做的，仅是揭示这台机器为何如此容易被启动，绝不为它的任何一次具体使用背书。能构造出一个讲得通的叙事，从来不等于这叙事该被采信，更不等于它该被拿去下判决；至于影响的判定标准该如何被重新校准、责任究竟该如何安放——这些问题，本文不touched其一步，明确将其悬置为**尚待研究**。本文这一篇，自己也是一段讲得通的叙事、一根指月之指；指可弃，月不可弃。 ## Abstract This paper treats a reversal that goes further than the earlier diagnosis "the cognitive boundary is often mistaken for the ontological boundary": **an influence, so long as it can be described, is thereby taken to exist; and conversely, the influence we feel to exist is only the influence we are able to describe.** This is not to say we sometimes locate the boundary in the wrong place; it is to say that the **sole operative criterion** by which we judge whether an influence exists somewhere is, from start to finish, "can it be told coherently"—describability is not a lens for discovering the facts, but the entire basis on which we affirm them, with no second, independent channel. The paper further argues that this threshold of "can it be told" is far **easier to clear** than we ordinarily suppose, and offers two mechanisms. First: **an influence of A's need not carry some signature exclusive to A in order to count as A's influence.** A need only move a generic, unsigned intermediate variable—say, some chemical substance in B reaching a certain level—and this variable will diffusely affect **everything** B does afterward, not merely some one event sensitive to it, because no one can prove that any given one of those events was **not** affected. This is an inversion of the burden of proof: one need not affirmatively prove "A affected this specific event," one need only point to a generic variable that exists and cannot be excluded—and that alone suffices to construct a tellable narrative of influence; influence is thus far more widespread and pervasive than we ordinarily suppose. Second: this same mechanism generalizes from a chemical level within the body to the physical evolution of the entire universe. Suppose I pass through a place, and afterward someone there commits a crime; that crime follows upon the event of my passing through—had I not passed through, the state of the universe would certainly have differed, and the subsequent evolution of a different state might not have produced that crime at all—this, too, is tellable: it says only that different starting points lead to different processes, without needing to specify concretely how the leading occurs. The paper further identifies, in moral practice, a bare and double-standard-ridden seam where this mechanism shows itself: **that mood drives behavior is common sense**, a causal link we accept without the least resistance; yet we ordinarily do not press the question of what "caused" that mood, especially when the antecedent that caused it is itself **morally neutral** (the weather, an unrelated remark, five minutes stuck in traffic)—not that these neutral events truly bear no causal role, but that we choose not to press the question, and this stopping is misread as "causation simply breaks off here." The paper holds, with particular gravity, a line more firmly guarded than any prior revelation in this series: that an influence-narrative can be told coherently is **by no means** grounds for it to be believed, still less to be actually used to apportion or excuse responsibility. What is most dangerous about this machine is that it can point in either direction—it can be used to persuade another that "your behavior was, in fact, caused by me," and equally to persuade oneself that "what I did should not, in fact, be blamed on me, since some unspecifiable antecedent caused it." This paper does no more than reveal why this machine is so easily set running; it endorses none of its concrete uses. That a narrative can be constructed coherently never means that narrative ought to be credited, still less adjudicated upon; as to how the criterion for judging influence ought to be recalibrated, and how responsibility ought actually to be situated—the paper does not take a single step toward either, and expressly leaves both suspended as **still under study**. This very paper is itself a coherently told narrative, a finger pointing at the moon; the finger may be discarded, the moon may not.

本文是一部统一框架的整全展开，把此前分散写就的诸篇论文，收拢成一条从最小公理一路走到认知边界的完整路径。全部论证的总纲是一句判断：**思维就是在被给予的诸时空局部间连线的活动，这些线并不实存。** 被给予的只是并置的时空局部；思维不停地在其间**连**（添因果与同一之箭）、**分**（切范畴、画边界）、**替换**（生成反事实），三种操作添出的都是笔迹，不是世界的关节。本文循此总纲，依次展开七个板块。其一，**具身模型**：把这条冷公理压成一句有体感的话——我们身处一场仍在进行的大爆炸，是它飞散的碎片；操控之线不实存，但"决定"仍有三重真用（环节、咬合之显、点亮），且爆炸内部本有真实的相对静止，须与纯粹的幻线分开、不可一锅端。其二，**范畴**：桌、云、猫这类日常范畴，是头脑用不同模块切出的浅范畴，各有其适用尺度；爆炸才是深范畴，在一切尺度成立——桌云猫更接近爆炸，指可弃，月不可弃。用日常概念思考这类深实情，是一种入戏式的遮蔽，恰如把生命游戏里滑翔机的条件性规律，误当成无条件的本质。其三，**操控意向的图景**：想象一个无窗房间时，脑中自动搭起一座操控剧场，把世界封成可操控的密盒、把物设为惰性客体、把自己设为能从外部撬动一切的全能施动者；可摆桌子的不是"你"，是爆炸；提问的、回答的，也都是爆炸。多放一个人，伦理学便立刻诞生——它是这幅操控图景复制为二、相遇争夺的产物；而关于过去的伦理论证，则是同一图景回溯地投射到已炸定的历史上，其两根支柱（操控者与反事实）皆是幻象，但那段过去真发生过的事、真被经历过的疼，不因此被否认。其四，**反事实的机制**：反事实是对叙事表征做最小替换而生成的，其空间被描述的颗粒度锁死；把这一命题推到极限便是一个操作性反证——你从未克隆出宇宙的一份完整备份、再逐一基本粒子地重写它以得到反事实宇宙；你手里从头到尾只有一句改过的话。其五，**动力的错置**：并无宇宙状态之外的额外动力，给定实际状态，发展只有一条路径；主动与被动的分工是可反转、故是贴上去的（读取者与被读物地位对等），而正是这层错置的转嫁，制造了"人是唯一责任者"的信念；把这套判法逐尺度复制叠加，便得到一场分形式的动力错置——从心像的整体，到其中的组件，到我们所居的超级宇宙。其六，**框与操控者**：心像所框的小宇宙很小，我必然在其外，这纯粹的位置事实被自动升级为"我是额外操控者"的能力断言；把这份操控者身份拿去实测，会发现一进门"原宇宙"便已不复存在；而"人操控环境"还是"环境操控人"，方向可以任意调换、换了照样通，说明这问题从一开始就问错了。其七，**认知边界与责任**：认了"甲影响了乙"，这条线便会无限延伸，直至任何人此刻的行为都是所有前人的总和——这正面反证了"源于自己"这一感觉的不可靠；而我们平时感觉不到，是因为绝大多数影响无法被具体描述、指认、界定，我们把这份认知上的"说不清"，错记成了本体上的"不存在"；同一机制在时间维上更加彻底——未来的未知因素，因没有具体结果的倒逼，连语言上的位置都未曾获得，比过去的未知因素被抹除得更干净，"唯一主宰感"正由此而生。全篇贯穿一条从未松动的纪律：所有这些揭示，拆的都只是覆于实情之上的重画（操控者、反事实、独家的主动性与责任），从不是那片真在发生的涌动、那些真被经历的体验；伦理与责任这一档，本文全程只做揭示，不下裁决，明确将其悬置为**尚待研究**。本文自身，连同它所使用的全部词汇，也是这片涌动在此处的一段，也是指月之指；指可弃，月不可弃。 ## Abstract This paper is the full unfolding of a unified framework, gathering what was previously written across many separate papers into one continuous path running from a minimal axiom to the boundary of cognition. The whole argument's trunk is a single judgment: **thinking is the activity of drawing lines between the spatiotemporal locals of what is given, and these lines do not exist.** What is given is only juxtaposed spatiotemporal locals; thinking ceaselessly **connects** (adding arrows of causation and identity), **divides** (cutting categories, drawing boundaries), and **substitutes** (generating counterfactuals) among them—three operations whose traces are never the world's joints. Following this trunk, the paper unfolds seven clusters. First, the **embodied model**: this cold axiom is compressed into a felt statement—we are within a still-ongoing Big Bang, its flying debris; the control-line does not exist, though "deciding" retains three genuine uses (link, a showing of interlock, lighting), and the explosion's interior really does contain relative rest, which must be told apart from pure phantom lines rather than lumped together. Second, **categories**: everyday categories like table, cloud, cat are shallow categories cut out by the mind's different modules, each holding only within its scale of applicability; the explosion is the deep category, holding at every scale—table, cloud, and cat are closer to the explosion; the finger may be discarded, the moon may not. Thinking such deep facts with everyday concepts is an in-character screening, just as mistaking the Game-of-Life glider's conditional rule of motion for an unconditional essence. Third, the **picture saturated with the intention to control**: imagining a windowless room, the mind automatically builds a theater of control, sealing the world into a controllable box, setting objects as inert, setting oneself as an omnipotent agent able to pry everything from outside; yet it is not "you" who arranges the table, but the explosion; the one posing and the one answering are likewise the explosion. Add one more person, and ethics is instantly born—the product of this control-picture doubled and set colliding; and ethical arguments about the past are the same picture projected backward onto an already-detonated history, its two pillars (the controller, the counterfactual) both phantoms, though what really happened and what pain was really undergone are in no way denied. Fourth, the **mechanism of the counterfactual**: a counterfactual is generated by minimal substitution upon a narrative representation, its space locked by the granularity of description; pushed to its limit this yields an operational reductio—you never cloned a full backup of the universe and rewrote it particle by particle to obtain a counterfactual universe; all along you hold only a sentence that has been altered. Fifth, the **misplacement of impetus**: there is no impetus outside the universe-state, so given the actual state, development has only one path; the division of active and passive is reversible and hence pasted on (reader and read are equal in status), and it is precisely this misplacement's transfer that manufactures the belief that "the person is the sole bearer of responsibility"; replicating and stacking this judgment across scales yields a fractal misplacement of impetus—from the whole of a mental image, to its components, to the super-universe we inhabit. Sixth, the **frame and the controller**: the small universe framed by a mental image is small enough that I am necessarily outside it, and this bare positional fact is automatically upgraded into the capability-claim "I am an extra controller"; putting this controller-identity to the test, one finds that upon entering, the "original universe" has already ceased to exist; and whether "the person controls the environment" or "the environment controls the person," the direction can be swapped at will and either way it works, showing the question was mis-posed from the start. Seventh, the **boundary of cognition and responsibility**: once one grants that "A influenced B," the line extends without limit, until any person's behavior right now is the sum of all prior people—a positive refutation of the reliability of the feeling "sourced from oneself"; and we ordinarily feel none of this because most influence cannot be specifically described, identified, or delimited, and we mistake this cognitive "cannot articulate" for an ontological "does not exist"; the same mechanism is sharper along the axis of time—the future's unknown factors, lacking the forcing of any actual outcome, never even acquire a verbal slot, and are erased more cleanly than the past's, whence the sense of sole mastery arises. One discipline runs unbroken through the whole: everything revealed here dismantles only the re-painting laid over the facts (the controller, the counterfactual, the exclusive claim to activity and responsibility), never the welling that really occurs, the experience really undergone. The register of ethics and responsibility, throughout, is treated by this paper strictly as revelation, never verdict, and is expressly left suspended as **still under study**. This paper itself, together with every word it uses, is a stretch of that same welling here, and a finger pointing at the moon; the finger may be discarded, the moon may not.

本文追一处日常态度里的不对称：我们在事后，能很自如地为自己的行为找出各种原因——包括外在的触发；可在事前，计划或担忧未来时，我们却不会把”外在的未知因素”纳入考虑。 这不对称的产生，本文指出，根子不在于我们否认了”未来会有我算不到的外部影响”这件事本身，而在于一个更隐蔽的机制：那些因素”未知”，于是我们没有任何具体的内容可以放进计划的图景里；而这份”没放进去”，被我们不自觉地，悄悄读成了”它们不存在”。这不是计划表上有一栏被空着——是计划表在被设计出来的那一刻，就没有“未知外部因素”这一栏；一栏填不了任何具体内容的表格，直接被从表格的设计里删掉了。于是我们对未来的整幅图景，便只剩下自己的意图、决定、行动——它显得完全是自制的，因为唯一会来打扰这份自制感的那个变量，从制图之初，就未被纳入。本文指出，“我将是自己人生的唯一主宰/掌控者/责任者”这一感觉，正是从这份图景的空缺里长出来的，而不是从”确实没有别的力量会介入”这一事实里长出来的——地图上没画的地方，不代表那里没有地形，只代表制图的人没法把它画上去；而我们习惯性地把”画不出来”读成了”不存在”。本文进而指出一个更精细的不对称：过去与未来，对着同一种”说不清楚、但很可能存在”的外部力量，处理方式截然不同。事后，我们至少在口头上会承认这类力量的存在——“肯定有些我没意识到的原因”，因为已经发生的那个具体的结果，倒逼着我们去承认”总有个原因”；可对着未来，我们连这份口头的承认都不必给出——因为什么都还没发生，没有任何具体的东西，逼着我们去承认”总会有些我算不到的力量”。没有那个”结果”的倒逼，“未知外部因素”这一栏，便直接从我们对未来的图景设计里消失了，比过去的未知因素被抹除得还要彻底。这也精确地解释了，为什么”唯一主宰感”格外容易发生在我们对未来、而非对过去的态度里：回顾时，这份感觉，好歹要与”我确实说得出一些外部原因”这个残留的事实打个照面；前瞻时，它没有任何东西要打照面，因为那个本该提醒我们”还有别的力量”的位置，从一开始，就被”未知”这个理由，最彻底地清空了。本文守一处分寸，严格只做揭示：说”唯一主宰感源于图景设计上的空缺”，绝不是说计划本身有缺陷、该被修正为强行纳入某些具体的未知因素——恰恰相反，不去虚构一个具体的、你根本不知道的”未知因素”填进计划表，是认知的诚实，不是疏漏；硬填一个具体内容进这一栏，反而是自欺。被点破的，只是”表格里没有这一栏”与”这一栏对应的东西不存在”之间，那一步悄悄发生、却从未被论证过的滑动。至于”人在多大程度上真是自己人生的主宰、这份责任感究竟该如何安放”——本文对此不下任何新判决，明确将其悬置为尚待研究，不在此落子。本文这一篇，自己也是被写下之前，被无数说不清的未知因素所塑造、此刻才被填入这张纸面的一段文字；它也是指月之指；指可弃，月不可弃。 Abstract This paper traces an asymmetry in everyday attitude: in hindsight we can quite freely find various causes for our own behavior—including external triggers; yet beforehand, when planning or worrying about the future, we do not factor in “unknown external factors.” This paper argues that the root of this asymmetry lies not in our denying the very fact that “the future will hold external influences I cannot foresee,” but in a more covert mechanism: those factors are “unknown,” and so we have no specific content to place into the picture of our plan; and this “not having been placed in” gets, without our noticing, quietly read as “they do not exist.” This is not a blank cell left on the planning table—it is that, at the very moment the table was designed, it never had a column for “unknown external factors” at all; a column into which no specific content can be entered is simply deleted from the table’s design. So the whole picture we hold of the future ends up composed only of our own intentions, decisions, actions—it appears entirely self-authored, because the one variable that would disturb this sense of self-authorship was never included from the start of the mapmaking. This paper argues that the feeling “I will be the sole master/controller/one responsible for my own life” grows precisely out of this blank in the picture, not out of the fact “there truly is no other force that will intervene”—what the map leaves blank is not thereby terrain that does not exist, only terrain the mapmaker could not draw; and we habitually read “could not draw” as “does not exist.” The paper further points out a finer asymmetry: past and future, facing the very same kind of “hard to specify, yet very likely present” external force, are treated quite differently. In hindsight, we at least verbally acknowledge the existence of such a force—“there must have been causes I wasn’t aware of”—because the actual, already-occurred outcome forces us to concede “there must have been some cause”; but facing the future, we need not offer even this verbal acknowledgment—because nothing has happened yet, and there is no specific thing forcing us to concede “there will surely be forces I cannot foresee.” Without that forcing by an outcome, the column “unknown external factors” simply vanishes from the design of our picture of the future—erased even more thoroughly than the unknown factors of the past. This also precisely explains why the sense of sole mastery arises more readily in our attitude toward the future than toward the past: in looking back, this feeling must at least face off against the residual fact “I can, after all, name some external causes”; in looking forward, it faces off against nothing at all, because the very slot that ought to remind us “there are other forces too” was, from the outset, most thoroughly emptied by the reason “unknown.” The paper keeps one measure, strictly limited to revelation: to say “the sense of sole mastery springs from a blank in the picture’s design” is by no means to say the plan itself is flawed and ought to be revised to forcibly include some specific unknown factor—on the contrary, not fabricating a specific “unknown factor” you simply do not know and inserting it into the plan is the honesty of cognition, not an oversight; forcibly filling that column with specific content would itself be a form of self-deception. What is punctured is only the step—quietly taken, never argued for—sliding from “there is no such column in the table” to “there is nothing corresponding to such a column.” As to “to what extent a person truly is the master of his own life, and how this sense of responsibility ought actually to be situated”—the paper renders no new verdict on this, expressly leaving it suspended as still under study, and sets no stone here. This very paper is itself a stretch of text, shaped before it was written by countless unspecifiable unknown factors, only now entered onto this page; it too is a finger pointing at the moon; the finger may be discarded, the moon may not.

本文追一条论证，从一句几乎无可指摘的日常判断出发，一路推到它自己收不住的边界，再把这份"收不住"，正面地用作对"一个人的行为源于他自己"这一感觉的反证，最后揭示我们为何察觉不到这份反证——问题出在认知的边界，被我们悄悄错记成了本体的边界。第一步（归谬本身）：甲乙互动，我们既可以认为乙的行为/状态完全或部分受甲的操控/影响，也可以认为乙完全自主。若认了前一种——乙确实受了甲的影响——这条线并不会在乙这里停下：乙之后在另一场景与丙互动，丙的行为/状态，便也（部分）承接了这份影响；丙之后遇上丁，同样的道理再套用一次……这条线可以不断、无限地延伸下去，以至于**任何人此刻的行为/状态，都是所有前人操控/影响的总和**。这不是量上的夸张，是结构上的必然：因为甲、乙、丙、丁之间，从来不是几颗靠着一条条独立影响线相连的珠子，是**同一片连续的涌动**，在不同的时空点上无缝相连着——你想在其间画一刀"从甲到乙"的线，会发现这刀口天然地接着往下延伸，因为那从来不是可离散计数的几条线，是一整片切不断的连续状态。第二步（装上责任的引擎）：把这条归谬，接上两条我们几乎不假思索便会同意的常识信念——**人对自己身体的行为应负责**，以及**一个人若影响了另一个人的状态，这份影响者也应负责**——认真、一以贯之地把这两条信念结合起来，就推出：一个人此刻的状态/行为，应由**所有有关的前人**负责。这个"荒谬"的结论，本文不将其当作需要去质疑两条前提的反证，而是正面地拿来用：它照出的，不是这套责任直觉本身出了错，是**"一个人的行为源于他自己"这一感觉，本就不可靠**——这是一次不需要引入任何新概念、仅凭我们自己认可的责任逻辑一路兑现到底，便能自我拆穿"源于自己"这份感觉的证明。第三步（为何我们感觉不到）：既然如此，为何我们平时对此毫无所感？本文指出，这牵涉一个认知论上的机制——**描述能力的边界，被我们悄悄错记成了实在有无的边界**。在特殊情况下，我们确实能具体指认某种影响，比如一门技能的师承：这类影响路径清楚、内容具体，于是它被我们承认、被计入账本。可绝大多数人际互动的影响，根本没有这种清晰的可指认性——一句随口的话、一个眼神、一次擦肩，可能在某人心里种下极细微的偏转，这偏转又在他后续无数决定里、以我们无法追踪的方式弥散地起着作用；这种影响无法被具体描述、指认、界定，于是它便显得"似乎不存在"了。但它其实肯定存在，只是**超出了我们认知能力的边界**——"说不清楚"，被我们错当成了"没有发生"；技能传承之所以被承认，不是因为它比其他影响更真实，只是因为它比其他影响更**好描述**——认知上的方便，被错当成了本体上的有无。同一机制，也解释了"环境为何总被感觉成死的"：被我们视为死物的环境对被视为活物的人的影响，尤其是其后续、长期、传递性的部分，通常也格外难以界定、表达，于是同样被忽略、被视为不存在——环境之所以显得没有动力，不是它真的惰性，是它的动力，恰好落在了我们认知能力最够不着的那种弥散、长期、传递性的影响形态上。本文守住一条须格外严守的界线：以上全部，只是**揭示**——揭示归谬本身的结构必然性，揭示"源于自己"的感觉建在一个可被自我拆穿的直觉逻辑上，揭示我们对"影响是否存在"的判断，被认知能力系统性地扭曲了。本文**没有、也不**由此推出"所以没人该为自己的行为负责，责任都该按无限链条摊派给说不清楚的前人"——这样的新判决，同样架在一个未经论证的预设上：**责任是一种可以被计量、追溯、按份摊派的东西**；而这个预设，恰恰可能和归谬本身揭示的实情（那从来不是离散可数的一条条线，是一整片连续、无天然切口的涌动）相冲突——若真没有可切的刻度，"该摊派多少给谁"这个问题，或许根本无解，而非另有一个更公平的新答案。故"责任究竟该如何安放"，本文仍将其明确悬置为**尚待研究**，不在此落子。本文这一篇，自己也是这条无限链条上，被前人以无数说不清的方式塑造出的一个环节，也是一根指月之指；指可弃，月不可弃。 ## Abstract This paper traces an argument from a nearly unimpeachable everyday judgment, all the way to the boundary it cannot itself hold, and then puts this "cannot hold" to positive use as a refutation of the feeling that "a person's behavior is sourced from himself," before finally revealing why we fail to notice this refutation—the trouble lying in a boundary of cognition quietly mistaken for a boundary of being. First (the reductio itself): in an interaction between A and B, one may equally hold that B's behavior/state is wholly or partly controlled/influenced by A's, or that B's behavior/state is wholly autonomous. If one grants the former—that B really was influenced by A—this line does not stop at B: B, later interacting with C in another scene, passes on this influence, so that C's behavior/state (partly) inherits it too; C later meets D, and the same reasoning applies once more… This line can extend without limit, so that **any person's behavior/state, right now, is the sum of all prior people's control/influence.** This is not a matter of exaggerated degree but of structural necessity: A, B, C, D are never beads strung together by discrete, independent lines of influence; they are **one continuous welling**, seamlessly joined at different points in space and time—try to cut a single line "from A to B," and one finds the cut naturally continues onward, for these were never discretely countable lines but one uncuttable, continuous stretch of state. Second (fitting the reductio with an engine of responsibility): attach this reductio to two commonsense beliefs we scarcely think twice about affirming—**a person is responsible for the acts of his own body**, and **if a person influences another's state, the influencer, too, bears responsibility**—and combining these two beliefs honestly and consistently yields: a person's state/behavior right now ought to be answered for by **all the prior people concerned**. This paper does not treat this "absurd" conclusion as a reductio meant to cast doubt on the two premises; rather, it is put to positive use: what it reveals is not that our responsibility intuition itself is mistaken, but that **the feeling "a person's behavior is sourced from himself" is, from the start, unreliable**—a proof requiring no new concept, arrived at simply by honoring our own accepted responsibility-logic all the way to its end, thereby refuting, from within, the felt sense of "sourced from oneself." Third (why we fail to feel this): why, then, do we ordinarily feel nothing of this? The paper points to a mechanism in the theory of cognition—**the boundary of our descriptive capacity has been quietly mistaken for the boundary of what exists.** In special cases we can indeed specifically identify an influence—say, the transmission of a skill from master to apprentice: such influence has a clear path, concrete content, and so it is acknowledged, entered into the ledger. But most interpersonal influence has no such clean identifiability—an offhand remark, a glance, a passing encounter may plant an infinitesimal deflection in someone's mind, a deflection that then operates diffusely, untraceably, across his countless later decisions; such influence cannot be specifically described, identified, or delimited, and so it comes to seem "as if it did not exist." But it certainly does exist; it merely **lies beyond the boundary of our cognitive capacity**—"cannot be articulated" gets mistaken for "did not happen"; the transmitted skill is acknowledged not because it is more real than other influences, but because it is more **describable**—a convenience of cognition mistaken for a fact of being. The same mechanism explains why "the environment always feels dead": the influence of the environment (taken as dead matter) on the person (taken as living), especially its later, long-term, transmissive portion, is likewise usually too hard to delimit and express, and so it too is overlooked, taken as nonexistent—the environment seems to lack force not because it truly is inert, but because its force happens to fall precisely in that diffuse, long-term, transmissive form our cognition is least able to reach. The paper keeps one line that must be guarded with special care: all of the above is only **revelation**—revealing the structural inevitability of the reductio itself, revealing that the feeling "sourced from oneself" rests on a self-refuting intuitive logic, revealing that our judgment of whether an influence exists is systematically distorted by cognitive capacity. The paper does **not**, and does not intend to, derive from this "so no one need answer for his own acts, and responsibility should be apportioned along the infinite chain to prior people who cannot be specified"—such a new verdict would itself rest on an unexamined premise, that **responsibility is a thing that can be measured, traced, and apportioned in shares**; and this premise may well conflict with what the reductio itself has revealed (that this was never discrete, countable lines, but one continuous stretch with no natural seam to cut)—if there really is no notch to apportion along, the question "how much is owed to whom" may simply have no answer, rather than admitting some fairer new one. So "how responsibility is actually to be situated" the paper expressly leaves suspended, **still under study**, and sets no stone here. This very paper is itself one link on this infinite chain, shaped by countless prior people in ways that cannot be articulated; it too is a finger pointing at the moon; the finger may be discarded, the moon may not.

本文追一个日常对话里几乎从不被察觉的画法，并指出它的破绽恰恰暴露在它最不设防的地方——方向可以随便换，换了照样说得通。 设想我正和一个人说话。我丝毫不觉得，这整个情境，底层其实是同一片连续的涌动；我也不觉得，对方所处的环境——房间的气氛、此刻的场合、旁人的目光——对他此刻说的话、做的事，有什么真正的影响。于是我心里默认的画法，是把”这个人”和”他的环境”劈成两半：一半会说话、会决定、会推动（那个人），一半死气沉沉地摆在那儿（环境）。而这一记洞见点出最要害的一步：这个”谁主动谁被动”的方向，居然可以随便换，换了照样通。我可以说”他是他那片环境的额外操控者”——他站在那儿说话、动作，仿佛在主导、支配着这片空间；我也一样可以说”环境是他的额外操控者”——房间的气氛、这场合的性质、旁人的目光，其实在暗中拿捏着他说什么、怎么说。这两种讲法，我心里对哪种都不觉得别扭。本文指出：这份”随便换、换了都通”，正是破绽所在——一个关系若真是实在里的关系，方向该是定死的，反了就该显得荒谬；可这里反了照样顺，说明两个方向，指的根本不是同一片实在里画着的箭头，是我们讲述时随手贴的。 不是”操控者是他还是环境”这个问题有两个可能答案；是这个问题从一开始就问错了——它偷偷预设了”人”和”环境”是两个东西，其中一个在支配另一个；可两个方向都通，恰恰揭穿了这个预设本身是假的：不是两个东西谁支配谁，是同一片涌动，我们习惯地把它劈成了”人”和”环境”两半，再因为”总得有个说法”，随手往哪个方向都贴一个操控之箭——箭往哪指不重要，因为压根没有两个各自独立、需要一个箭去连接的东西。这台”制造操控者”的机器，还有一层格外值得注意：这一次，我根本不在这个关系里——我只是在旁观两个人（或一个人与他的环境），我不是那个人，也不是他的环境；可我心里照样自动地、不由分说地，把”人”和”环境”劈成了两半，还给它们分派了一个操控方向。这说明这台机器不只在”我站在框外、面对我要处置的东西”时启动，它是一台更普遍的机器——只要眼前有个整体，它就会自动把整体劈成两半，再给两半之间随手安一个”谁支配谁”的箭头，跟”我”本人在不在这个关系里，没有关系。本文守一处分寸，纯属描述性：说”人与环境谁操控谁的方向是贴的、这问题问错了”，绝不是说”人”和”环境”这两个词毫无用处、该弃——用这两个词去理解一场对话、去判断一个人此刻的状态受了什么影响，极为有用，日常的社交判断、共情、劝说几乎全靠它。被点破的，只是”这两半之间存在一个真实的、定死方向的操控之箭”这一层设定；那片实情，从来只是一整片连续的、相互牵连的涌动，没有被真正劈开过。认清”人与环境的划分是我随手劈的、方向是我随手贴的”，与照常在日常里这样划分、这样理解一场对话，是并行不悖的两件事。本文这一篇，自己也是被劈成了”作者”与”论证内容”两半的一段涌动；它也是指月之指；指可弃，月不可弃。 Abstract This paper traces a way of picturing things that goes almost entirely unnoticed in everyday conversation, and shows that its flaw is exposed precisely where it seems least guarded—the direction can be swapped at will, and either way it works. Suppose I am talking with someone. I do not in the least feel that this whole situation is, underneath, one continuous welling; nor do I feel that the environment the other person is in—the mood of the room, the nature of the occasion, the eyes of onlookers—has any real bearing on what he says or does just now. So the picture I default to splits “this person” and “his environment” in two: one half that speaks, decides, and drives (the person), one half that sits there inertly (the environment). And the sharpest point of this insight is this: the direction of “who is active, who passive” can be swapped at will, and either way it works just as well. I can say “he is an extra controller of his environment”—he stands there speaking, acting, as if presiding over, governing this space; and I can equally say “the environment is an extra controller of him”—the mood of the room, the nature of the occasion, the eyes of onlookers are, in fact, covertly dictating what he says and how. Neither telling strikes me as the least bit awkward. This paper argues: this “either way works” is exactly the flaw—if a relation were truly a relation in reality, its direction should be fixed, and reversing it should look absurd; but here reversal works just as smoothly, which shows that the two directions do not point to one and the same arrow drawn in reality, but are pasted on however we happen to be telling it. The question is not “is the controller him or the environment, with two possible answers”; the question was mis-posed from the start—it quietly presupposes that “person” and “environment” are two things, one governing the other; but that both directions work exposes this very presupposition as false: it is not two things, one dominating the other, but one continuous welling, which we habitually split into “person” and “environment,” and then, because “there must be some account of it,” paste an arrow of control onto whichever direction happens to occur to us—it does not matter which way the arrow points, because there were never two independent things needing an arrow to connect them in the first place. This machine that manufactures controllers has a further point worth noting: this time, I am not even within this relation—I am merely watching two people (or a person and his environment); I am not that person, nor his environment; yet my mind still automatically, unhesitatingly, splits “person” and “environment” in two, and assigns them a direction of control. This shows the machine is not activated only when “I stand outside a frame, facing something I must deal with”; it is a more general machine—wherever there is a whole before one’s eyes, it will automatically split that whole in two, and casually fix an arrow of “who governs whom” between the two halves, regardless of whether “I” myself am within that relation or not. The paper keeps one measure, purely descriptive: to say “the direction of who controls whom between person and environment is pasted on, and the question is mis-posed” is by no means to say the words “person” and “environment” are useless and should be discarded—using these two words to understand a conversation, to judge what has influenced a person’s present state, is immensely useful, and daily social judgment, empathy, and persuasion run on it almost entirely. What is punctured is only the setup that “there exists, between these two halves, a real, direction-fixed arrow of control”; the reality has only ever been one continuous, mutually implicated welling, never truly split apart. To recognize that “the division of person and environment is one I casually cut, and the direction is one I casually pasted on,” and to go on, in daily life, dividing and understanding a conversation this way as usual, are two parallel-not-conflicting things. This very paper is itself a stretch of welling split into “author” and “argued content”; it too is a finger pointing at the moon; the finger may be discarded, the moon may not.