明知而故犯
Knowing Better, Doing Otherwise
第七卷第二章一开口，就是一道他自己标出来的难处：
有人会问：一个人想得明明是对的，怎么还会不能自制。
难在对面站着谁。
这一段是两层，不是一层。头一层是"有人说"：一个人身上真有那门知识，还能有别的东西压过它，把它像个奴隶一样拖着走，这太怪了；苏格拉底当年就是这么看的。第二层才是他自己转述苏格拉底本人：苏格拉底整个跟这个说法作对，认为根本没有不能自制这回事；没有人是在心里那样认着的时候还偏要做违背最好的那件事，之所以那样做，是出于不知道。
接下来那一句是整章的分水岭。而且它是一整句，中间只有一个逗号，不许拆成两句读：
这套道理跟摆在眼前的事明摆着不合；而既然该就这件事查一查，那就要问，要是出于不知道，这个不知道是哪一种样子的不知道。
这一句要分两半读。
前半是一记重手。在他自己那套办法里，一套道理跟摆在眼前的事对不上，就是这套道理照原样立不住。他开卷讲办法的时候写得清清楚楚：疑难走完，那些话还站着，才算证够了。所以他确实判了：苏格拉底那句"根本没有不能自制这回事"，照原样不能留。
后半是他没有停在这一记上。不再问知不知道，改问那个不知道长什么样。
判了，然后改问法。最后救回来的是一层，不是整套。
第三章他就照这个问法往下拆，一连拆了四道。
第一道：知道这个词本来就有两种用法。一个人有那门知识而没在用它，和一个人正在用它，两个都叫知道。所以事情有分别：一个人做那件不该做的事，是拿着那门知识而没在照看它，还是正照看着。
第二道：前提有两种。两个都拿在手里也不妨碍一个人做出违背知识的事，只要他用的是那个普遍的，不是那个一件一件的。因为做出来的事都是一件一件的。
第三道：还有一种拿着的方式。有而不用的时候，那个状态本身就不一样，一个人可以又算拿着又算没拿着，比方睡着的那个，疯了的那个，喝醉了的那个。而正在那种情绪里的人，样子正是这样。
第四道最狠，也最好记。
说得出那些从知识来的话，这不算凭据。因为正在那种情绪里的人也背得出证明，也背得出恩培多克勒的诗句；刚学会的人也把话一句一句串得下来，可他们还不知道，那得长到一块儿去，长要时间。所以不能自制的人说话，得照台上演员那样看。
这一拆跟前三拆不一样，要留神。前三拆里那个人是持有着的，只是没在照看，或者只用了普遍那一半。这一拆里，那个人还没有持有：话串得下来，可它还没长到一块儿去。
四道拆完，他回头结账：苏格拉底要的那一层看来是成立的。因为那个状态发生的时候，在场的不是那门正经算得上知识的东西，被拖着走的也不是它，是管感觉的那一头。
还有一件他当场交了出去。那个人怎么从那个状态里出来，他说这跟喝醉的和睡着的是同一笔账，不是这一处特有的，得去听研究自然的人怎么说。
他要的只有一件：动手那一刻，那个知是怎么个在场法。
这一道跟上一本第五篇挨着，可两篇动的不是同一样东西。
柏拉图那一处，是让一个从港口上来的人当场作证：一个人身上确实有好几样各要各的，而且会当着他自己的面分出胜负。他要证的是人里头不止一个。
亚里士多德这一处不碰人里头有几样。他碰的是"知道"这个词。同一个词，可以指手里拿着，也可以指正在用；可以指那个普遍的到了手，也可以指那个一件一件的到了手；还可以指嘴里说得下来而心里还没长到一块儿。一个把人拆开，一个把词拆开。
而两处的用处也不一样。人拆开了，能说明打架怎么发生；词拆开了，能说明为什么问"他到底知不知道"问不出东西来。
他那一句里有"凭据"两个字。把这两个字拎出来，就到了这边的地界。
《第一批判》审幻相那一章，给冒领定了一个很干脆的判法：一项主张在使用中出示的凭证，只要有一项不在它经检验和复核所获准的条目里，就算冒领。说短一点，出示的越出了持有的。
而紧跟着的那一句，管的正是本篇这一道：这个判定是回溯的，不问动机的。一样东西呈现出来的时候本来就默认不带地址，所以冒领多半没有人在行骗；被诊断的是呈现的样子，不是呈现的人存的什么心。
这两句是同一种凭据形状：出示得全，不担保这一刻以获准的方式持有着。
真正相接的是第四拆那一层，不是第一拆。第一拆里那个人持有着，出示没有越出持有，越出的是使用；第四拆里那个人话串得下来而还没长到一块儿，出示才真的越出了持有。演员嘴里的话一个字不差，出示的是那一整套，手上持有的是另一回事。
而亚里士多德没有说那个人在骗人。他问的只是这句话在这一刻是以哪一种方式在场的。
同一章还写着一条纪律，本篇非把它抄进来不可：诊断点名栏位，不点名人。一纸合格的诊断写的是哪一种凭证被出示了，账上哪一栏没有这一条，出处在哪儿；它不写的是出示的人存的什么心，该当何罪。理由写在同一处：定罪会把人逼得藏起来，藏起来的那些说法就免掉了对账，而免掉对账正是那条越滚越大的反馈要的条件之一。用猎巫的方式追问，有反过来把幻相养大的风险。
这一条在办公室里验得着。
交上来的东西出了错，会上有两种问法。一种是问这是谁做的，他当时怎么想的。另一种是问这一条的依据在哪儿，出处填了没有。
第一种问完，下一回错的东西不会变少，只会藏得深一点。第二种问完，纸上多一栏。
他没有停在栏位上。
他分完那几种在场的方式，是要回过身去给那个人一个说法。最后那一句他给足了：那一刻在场的是管感觉的那一头，不是那门正经的知识。这是一句关于那个人当时身上出了什么事的话。
这边分完就停。诊断的产出不是把那条东西划掉，是放回它自己的地址：已经挣着的一分不没收，成色由复核那边另算。而下一句写死了不许往前走：点栏位，不点人。
一个要说清那一刻他身上出了什么事，一个明写只到栏位为止。
这一层在日常里的一个样子，是复盘。
一件事办砸了，坐下来复盘。当时该怎么做，说得出一条不落的时候不少见。说得清楚，说得有条理，连别人为什么也会做错都说得出来。
而当时他确实没那么做。
这时候要是问一句"你到底知不知道"，这一问当场就废了。他知道，眼前这一堆话就是证据；他不知道，事情办成那样也是证据。两个答案各有各的证据，所以这一问分不出事情来。
亚里士多德换掉的正是这一问。他不问知不知道，他问那一刻那句话是怎么在场的：是拿着而没在照看，还是那个普遍的在手上而那个一件一件的不在，还是话串得下来而还没长到一块儿去。
他那一头有一格，这边找不出对应的位置。
这边给每一笔判定都配一个四栏的地址：这一判管的是哪个对象，站在哪一层，走的是哪一支，靠着哪几条设下来的东西。条件和依赖另有登记。地址是这边最细的一样活计，越界不越界全看它。
可地址记的是这一判用上了什么。
他要的那一样不在这上头。他要的是这一判本来够得着而这一次没有用上的那些。够得着而没用上，这边的地址上没有这一栏。
这边离得最近的一条叫冒领今站：拿一个旧站上做出来的说法，不经重审就对今天发判。那一条管的是拿出来了而拿错了。该拿出来而没有拿，这边没有对应的一条。
有人会说，把这一栏添上不就完了。
不是添一栏的事。地址是一次判定留下的记录，它写得下的只有这一次实际经手的东西。够得着而没有伸手的那些压根没有经手过，所以它们进不了这一栏。不是这一栏太窄，是它们不在这一次里。
要记它们，得另立一张单子：这个人此刻手边摆着哪些东西。这张单子不是没法立。它可以限定到某一个人，某一刻，某一类事情，而且事后可以复盘着重建。
可它自己也是一次作业。它得有自己的地址，它也只记它自己经手过的，它照样会漏。所以这不是一句"补上就完了"，也不是一句"永远补不上"。是这边现在手上没有这张单子；而它一旦立起来，会带着跟地址同一种的短缺。
而这一格不是可有可无的。他那一整套答案就架在这一格上：两个前提都在手里，用的只有一个。
这边另有一条写得很硬：账不因遗忘而销。一个再没有人过问的问题，仍然在账上。
这一句管的是它没有消失。
它没有说，动手那一刻，人伸不伸得到它。
The Socratic Scandal
Weakness of will—akrasia—presents Aristotle with a conflict between argument and experience. Socrates had maintained that no one knowingly acts against what is best; wrongdoing must involve ignorance. To many readers, the claim appears to erase a familiar phenomenon. People smoke after understanding the risk, send the message they resolved not to send, and repeat conduct they can explain in advance as destructive.
Aristotle refuses to discard the phenomenon, but he also refuses the simple assertion that knowledge is overpowered like a captive dragged away. The decisive question is: in what sense does the person know?
Book VII develops several distinctions. One can possess knowledge without using it. One can know a universal proposition while failing at the particular premise that governs this action. One can speak from knowledge while in a condition analogous to sleep, madness, or drunkenness. And one can repeat newly learned words that have not yet become part of oneself.
The result is not a single theory of excuse. It is a grammar of partial presence.
Reciting Empedocles
Aristotle's most vivid example is linguistic. A person under the influence of passion may utter proofs or verses of Empedocles. A beginner can string together the sentences of a discipline before possessing it in the mature sense. Actors speak lines that do not express their own convictions. Verbal availability proves less than it seems.
This helps explain a common institutional error. A person passes the training, repeats the principle, and uses the approved vocabulary. The organization records knowledge as present. When conduct contradicts the lesson, the only remaining explanations appear to be hypocrisy or irresistible appetite.
Aristotle opens another possibility: the proposition is possessed but not active in the practical syllogism. “Excessive drinking is harmful” may remain available at the universal level while “this is another excessive drink, now” fails to appear with practical force. Appetite activates a competing description: this is pleasant.
The person does not act in complete ignorance. Nor does fully active knowledge of the relevant particular stand untouched and lose a fair contest. The practical state is fractured.
Credentials and Possession
SAE can translate part of the problem into the difference between display and authorized possession. A credential, citation, or fluent phrase does not guarantee that a judgment has been adopted under the speaker's own responsibility. It may be borrowed authority traveling without the grounds that originally supported it.
That is especially clear with newly learned moral language. Someone can say “boundaries,” “consent,” or “structural bias” accurately enough to be recognized by a group while using the terms as passwords. The vocabulary is not false. Its presence does not settle whether the person can locate the relevant judgment in a live situation or answer for it when challenged.
SAE's audit can ask which premises actually entered the decision, not merely which propositions were available somewhere in the person's memory. In that respect it aligns with Aristotle's attention to use.
But the alignment has a limit. Aristotle is describing the state of an agent: bodily passion, universal and particular beliefs, perception, habituation, and recovery. SAE's field records judgments and their grounds. It should not convert that diagnostic structure into an exhaustive psychology of the person. The ledger can show what was used; it cannot read everything that was possessed but inactive.
The Missing Inventory
One response would be to add an inventory: list all relevant knowledge available to the agent and compare it with what entered the action. Yet the inventory is another judgment. Which knowledge was genuinely accessible? Which should have been salient? How do we distinguish dormant possession from words stored without assimilation? The proposed solution recreates the original problem at a higher level.
This is where Aristotle's analogies are better than a database. Sleep, intoxication, and beginner recitation do not define one mechanism. They mark different ways in which “having knowledge” can fall short of knowing in action. His analysis preserves gradation without pretending to measure it from outside.
SAE can still contribute precision. After the act, the agent may construct a story: “I knew all along.” The audit can ask what evidence shows that the relevant premise was active, what description governed the particular, and whether the later account is importing a knowledge state into the earlier moment. Forgetting the ground of a judgment does not erase responsibility, but neither does later recollection prove that the ground was practically present.
Knowing Better Is Not One Thing
The phrase “you knew better” often performs too much work. It may mean that the person once received the information, could recite a rule, recognized the particular, had time to deliberate, or actively affirmed the better course at the moment of action. Those are different claims with different consequences for blame and repair.
Aristotle partly rescues Socrates by narrowing the target. Fully active knowledge is not simply dragged around by appetite. Akrasia involves a defect in the way knowledge is present, especially at the perceptual and particular level. Yet Aristotle also preserves ordinary experience: the agent acts willingly and with knowledge in some sense, and the failure remains attributable to the person.
The achievement is a refusal of two absolutions. We need not say, “The person knew, so knowledge has no practical force.” Nor must we say, “The person acted badly, so there was no knowledge at all.” Knowledge has modes of possession and activation.
SAE turns that lesson toward claims. What was displayed, what was adopted, what was used, and what address actually governed the act? Aristotle turns it toward the agent. What kind of knowing was present when appetite moved the body?
The familiar contradiction—knowing better and doing otherwise—does not disappear. It becomes more exact. “Knowing” was never one switch.