When 不正知 is running, the eyes are open, the senses are firing, the world is right there — and yet the meaning that gets registered is off-key. It is not that nothing is seen. It is that *something* is seen, confidently, and that something is wrong.
The Yogācāra school's 《成唯識論》 (Chéng Wéishí Lùn, "Treatise on Establishing Consciousness-Only," T1585) gives it a deceptively short definition:
於所觀境，謬解為性；能障正知、毀犯為業。
"At the object being observed, [it takes] mistaken understanding as its self-nature; it obstructs right knowing (正知, *samprajanya*) and brings about violations (毀犯) as its function."
This compactness hides a deep well. Let me walk through it.
一、这个概念在说什么：不正知的位置与结构
1.1 它在唯识学的哪里
不正知 is one of the 二十二隨煩惱 (twenty-two *secondary* or *consequent* afflictions, *pratyaya-kleśa*). In the standard enumeration:
- 小隨煩惱 (minor): 10 — 忿 (anger-flash), 恨 (enmity), 惱 (vexation), 覆 (concealment), 誑 (deceit), 諂 (flattery), 憍 (arrogance), 害 (harmfulness), 嫉 (envy), 慳 (stinginess). - 中隨煩惱 (middle): 2 — 無慚 (lack of conscience), 無愧 (lack of shame). - 大隨煩惱 (great): 8 — 惛沈 (mental torpor), 掉舉 (agitation), 不信 (lack of faith), 懈怠 (laziness), 放逸 (heedlessness), 忘念 (forgetfulness), 散亂 (distraction), and 不正知.
二十種 here is a tidy classical number. But the 《成唯識論》 uses the rhetorical device of the "與、并、及" three words (roughly, "and," "also," "plus") to argue that the category 隨煩惱 doesn't stop at twenty: 《雜事》 and other texts speak of 貪、瞋、癡 as 隨煩惱 too, because 「隨煩惱名，亦攝煩惱，是前煩惱等流性故」 — "the name *secondary affliction* itself *includes* the original afflictions, because they are the flowing-forth (等流) of the original ones." In this broader framework, 不正知 is a typical 等流性 隨煩惱, an outflow of the root afflictions (根本煩惱, *mūla-kleśa*).
A well-known verse in the text sums up the last four members of the 大八 in one breath: 「放逸及失念、散亂不正知」(放逸 = heedlessness; 失念 = forgetfulness, the same as 忘念; 散亂 = distraction; 不正知 = wrong knowledge).
1.2 自性：謬解為性
The 自性 (self-nature) of 不正知 is 謬解 (miù-jiě) — *mistaken* understanding. The phrase 於所觀境 specifies *what* is mistaken: it's the object currently being observed, not an abstract proposition, not a distant memory, not an absent thing. Whatever the mind is right now turning toward, 不正知 bends the perception of it.
This is the crux: 不正知 is not ignorance in the sense of blankness. It is *mis-reading*. There is contact (觸), there is feeling (受), there is some kind of mental presentation — and then, almost invisibly, a wedge is driven in: the meaning that gets taken is *the* meaning, and it's wrong. Sometimes subtly, sometimes grossly, but always: wrong.
1.3 業用：能障正知，毀犯為業
What does it *do*? Two things, tightly coupled:
- 能障正知 — it obstructs 正知 (right knowing, *samprajanya*). In classical usage, 正知 is the moment-to-moment clear comprehension of what one is doing — for instance, in the 大念處經 (Mahāsatipaṭṭhāna Sutta, DN 22 / MN 10), the section on 覺知 / 明覺 (sampajañña) describes a continuous matching: when walking, knowing "I am walking"; when bending, knowing "I am bending." This isn't philosophical knowledge; it's an unbroken thread of present-tense attending precisely matched to what is happening. 不正知 is the moment that thread slips, even by a hair. - 毀犯 — it generates violations, oversights, breaches. Once you have misread what is in front of you, action proceeds *from* the misreading, and the action goes astray: a wrong word, a wrong move, a missed cue, a promise broken without realizing it.
The text then says: 「謂不正知者，多所毀犯故」 — "The one with 不正知 has many violations." This isn't a one-time error. It's a structural feature of afflicted cognition that produces *repeated* misses — small at first, compounding over time.
1.4 三說之諍：它究竟是慧、是癡，還是兩者的合體？
The 《成唯識論》 does something unusual here: it presents three views on what 不正知 ultimately *is* (its underlying substrate) and declines to choose. This is not laziness; it is a precise acknowledgment that the canonical texts speak differently.
**First view: 慧一分攝 (a portion of afflicted wisdom, *kliṣṭa-prajñā*).** Reason: the 《瑜伽師地論》 (Yogācāra-bhūmi) speaks of 不正知 as wisdom "corresponding to" (相應) the afflictions. In other words, it's a discriminating cognition (簡擇) that has itself gone astray. To mis-recognize requires some cognitive function — you cannot make a mistake about an object unless there is *something doing* the recognizing.
**Second view: 癡一分攝 (a portion of delusion, *moha*).** Reason: 「《瑜伽》說此是癡分故。『令知不正』故名不正知」 — the text says it is a portion of 癡, and the name itself (不正 = "non-correct") tells us what 知 (knowing) is being distorted. Under this view, 不正知 is not really "a mistake made by wisdom" but ignorance operating *as* a mode of seeming-knowledge.
Third view: 癡、慧俱一分攝 (both portions, by mutual implication). Reason: 「由前二文影略互顯故，且論復說此遍染心故」 — the previous two passages *imply* each other (影略互顯), and the texts also say 不正知 pervades all afflicted minds (遍染心), which works best if it includes both wisdom-aspect and delusion-aspect.
Why this debate matters practically: the 《成唯識論》 then sharpens it with a pointed sub-question — the 「念慧為性之難」 (the difficulty of taking 念/慧 as the nature):
若忘念、不正知，念、慧為性者，不遍染心，非諸染心皆緣曾受、有簡擇故；若以無明為自性者，遍染心起，由前說故。
Translated: "If 忘念 (forgetfulness) and 不正知 have 念 (memory) and 慧 (discriminating wisdom) as their nature, they don't pervade all afflicted minds — because not every afflicted mind operates on previously-experienced objects (緣曾受) or engages in discrimination (有簡擇). But if 無明 (fundamental ignorance) is their nature, they pervade all afflicted minds, as previously stated."
This is a subtle but devastating observation. It says: not every afflicted cognition is *cognitive* in the rich sense. A flash of anger, a surge of lust, a sudden panic — these don't necessarily involve memory or judgment; they're raw, pre-judgmental surges. If 不正知 required 念 (memory) and 慧 (discrimination) to operate, it could not accompany those moments. But if it has 無明 at its root, it can ride along with every afflicted moment of the mind.
The 《成唯識論》 leaves this tension standing, without a one-sided verdict. The practical takeaway: 不正知 is at once the *brightest* form of misreading (it looks like understanding, often sophisticated understanding) and something rooted in the *darkest* ignorance (it can fire in raw, pre-reflective moments too).
1.5 大八遍諸染心：它无处不在
The text gives 不正知 a singular place: it is one of the 大八隨煩惱 (the "great eight"), and all eight of them — 惛沈、掉舉、不信、懈怠、放逸、忘念、散亂、不正知 — *deterministically* accompany every afflicted mind:
故染污心決定皆與八隨煩惱相應而生，謂惛沈、掉舉、不信、懈怠、放逸、忘念、散亂、不正知。
This is a strong claim. Any time the mind is in an afflicted (染污) state — gripped by 貪、瞋、癡 — these eight are guaranteed to be there as well, like the smell that comes with fire. 不正知 in particular is the *epistemic* companion of every afflicted moment. Whenever there's root affliction in the room, there's also 不正知 in the room, ensuring you mis-see what's happening.
1.6 諸識相應：它跟著谁走
The classical analysis also asks which consciousnesses can carry 不正知:
- 第六識 (mano-vijñāna, the conceptual / deliberative mind): 容有一切 — can have *all* the secondary afflictions in their full form. This is the mind that names, judges, plans, narrates — prime territory for 不正知. - 第七識 (manas proper, the "affective self-grasping" consciousness): also carries the 大八 including 不正知 in every moment of its operation. This is the deeper mind that constantly feels "I" — and with "I" comes the mis-recognition of phenomena as mine, against me, for me, about me. - 五識 (the five sense consciousnesses): 不正知 is 容有 (permitted) here too, *because* it belongs to the 大八 (which by definition accompanies all afflicted mind, including afflicted sense-contact). What is *not* permitted in the five senses is the 小十 (the minor ten), because they are too coarse (麁猛); that's why 「小十麁猛，五識中無」. 不正知 is finer-grained; it can ride along with a mis-seen face, a mis-heard tone, a mis-felt touch.
This last point matters: 不正知 is not just a "thought-mistake." It can be a *perceptual* mis-take at the level of bare sense-contact.
二、生活里的走查：把《成唯識論》的定義套到當代場景里
Let me walk through some scenes. For each, I'll try to map strictly to the classical definition: what is the 所觀境 (object being observed), what is the 謬解 (mistaken understanding that gets added), and what 毀犯 (violations) result. I'll also flag where 大众最容易映射错.
2.1 場景一：通勤路上，一個人一直插隊
所觀境: a stranger in the subway repeatedly pushes past you to board first.
不正知發生的瞬間: the perception ("a body moving forward past me") gets loaded, almost instantly, with a story: "He's doing this *to* me. He thinks he's better. He doesn't care about anyone." The reading feels *instantaneous* and *obvious*.
Mapping to the definition: the perception itself — "a body moving forward" — is not the same as the inference about motive and character. The *gap* between them is exactly where 謬解 lives. There could be fifty reasons for the behavior (late for a meeting, distracted, carrying something heavy, trying to catch a specific car) — most of them have nothing to do with you. But 不正知 delivers a tight, confident reading that seals the gap and presents itself as *the* reading.
Resulting 毀犯: you glare, mutter, push back. A small inner war begins over a phantom. By the time you arrive at work, your body is in fight-or-flight; attention is partially pre-occupied. You snap at a colleague. The actual transgressions (mild ones, but real) propagate outward from the initial misreading.
大众最容易映射错的地方: many people will say "that's just 瞋 (anger)" and miss the 不正知 underneath. Anger is downstream — 不正知 is the upstream delivery mechanism that hands anger a story it can grip. The 瞋 may surface and be seen; the 不正知 is what wrote the script.
2.2 場景二：深夜刷手機，一條延遲回覆的消息
所觀境: a friend or partner replies to your message four hours later, with a short message.
不正知發生的瞬間: the four-hour delay is loaded as "they don't care"; the brevity is loaded as "they're annoyed"; together: "I'm being deprioritized / dismissed."
Mapping: the object here is a piece of communication stripped of context. The mind, starved of cues, fills the gap with 謬解. It isn't doing nothing — it is doing the most cognitively expensive thing: imagining the other person's inner state from three data points. That work is precisely the kind of discriminating activity (簡擇) where 不正知 loves to ride. The 慧 is operating; what it produces is wrong.
Resulting 毀犯: a passive-aggressive reply, or no reply at all, or an internal rehearsal of confrontation. The relationship becomes a small battlefield over a message that, in its actual context, may have been "I was in a meeting and had two seconds — hi."
大众最容易映射错的地方: people will say "I was just anxious" or "I was just overthinking." But anxiety and overthinking are themselves *effects* of 不正知 — they are downstream. The classical diagnosis goes one level up: the perception was wrong *before* the worry could even start.
2.3 場景三：加班後的夫妻對話
所觀境: you come home late; your partner is on the couch, eyes on a screen, says "Hi" without looking up.
不正知發生的瞬間: the flat "Hi" is read as "you're not important"; the lack of eye contact is read as "you're failing at the relationship"; together: "My partner doesn't love me anymore / doesn't respect my work."
Mapping: the actual object is a tired person on a couch. The 謬解 is a romance-novel interpretation forced onto a banal scene. The mind is doing plenty of interpretation (簡擇); 不正知 is the bias in the interpretation engine.
Resulting 毀犯: a defensive tone, an unkind comment about the screen, an emotional withdrawal. These violations are exactly what the text means by 「多所毀犯」 — many breaches, small and accumulating. Over months and years, the small misreadings *build* the actual distance the misreading feared in the first place. A self-fulfilling loop.
大众最容易映射错的地方: people say "I'm just responding to the message they sent me." But the message they "sent" was *manufactured* by 不正知; the actual message — a tired "Hi" — was almost uncoupled from the response it triggered.
2.4 場景四：與父母的電話
所觀境: a parent calls and immediately asks about your salary / marriage / weight / life plan.
不正知發生的瞬間: the familiar question is read as "they're criticizing me / they're dissatisfied with me / they want to control me."
Mapping: the perception (a parent asking a question) is loaded with an entire biographical narrative of unworthiness. 不正知 doesn't need fresh evidence; it has a *template* it can apply in one shot. This is its efficiency and its danger.
Resulting 毀犯: you snap, then feel guilty, then hang up. The actual intent — which might have been "I miss you, how are you doing?" — is never engaged with. The relationship is harmed by the misreading more than by any question ever asked.
2.5 場景五：工作面試 / 績效評估
所觀境: the interviewer's face is neutral during your answer. Or the boss pauses before responding.
不正知發生的瞬間: neutrality is read as disapproval; pause is read as a hidden verdict.
Mapping: this is 不正知 in its most refined form. The classical 慧 (discriminating function) is doing a lot of work — reading micro-expressions, weighing prosody, predicting outcomes — but the ground under it (癡, the deeper not-knowing) is making the model wrong. Notice how the *content* of the misreading is almost invariably pessimistic: 不正知, when it runs, has a direction. It bends toward threat, threat, threat. The bias is built-in.
2.6 場景六：孩子的頂嘴
所觀境: your ten-year-old answers back when you ask them to put down the phone.
不正知發生的瞬間: the act of answering back is loaded as "disrespect" or "rebellion." The misreading fuses into a fixed character judgment: "this kid is ungrateful."
Mapping: the perception (a young person vocalizing disagreement) is interpreted through a moral-categorical grid that doesn't match the developmental moment. 不正知 writes a "character" narrative onto what is actually a developmental event.
Resulting 毀犯: a punitive reaction disproportionate to the moment; relationship damage that compounds over years. The 《成唯識論》's 「多所毀犯」 is most poignant in long-term relationships: the small misreadings *become* the actual wounds.
大众最容易映射错的地方: parents will say "they were rude, of course I reacted." But the child's tone is not the same as their character, and not the same as a verdict on the family. The "of course I reacted" is the very pattern 不正知 establishes; the certainty of the "of course" is its signature.
2.7 對照：正知 (samprajanya) 的樣子
Just so we don't lose the contrast. In the 大念處經, the practice of 明覺 (sampajañña, often rendered 正知 in Chinese) is described with remarkable concreteness:
When going forward or backward, he clearly comprehends (sampajānāti); when looking ahead or to the side, he clearly comprehends; when bending or stretching, he clearly comprehends...
This is not philosophical observation. It is an unbroken thread of *present-tense knowing* matched to the actual activity. 不正知 is precisely what breaks that thread. The 謬解 is the moment when, instead of "I am bending," the mind is already elsewhere — commenting, judging, narrating, projecting. The body is bending; the mind has skipped ahead to "what does this bending *mean* for my project / my status / my future." The act is unfelt; the meaning is invented.
三、當代人为什么需要這個概念
3.1 當代環境把「所觀境」抽空了
Most of human history, the 所觀境 was a *whole situation* — the texture of a face, the tone of a voice, the scene unfolding. There were cues. Our contemporary environment, by contrast, is a parade of *decontextualized fragments*: a text message stripped of expression, a Slack notification without body language, a three-line news headline, a thumbnail of a video, a one-sentence email reply.
The cognitive system is built to *fill* missing context. It cannot leave a gap unfilled; it has to land on a reading. 不正知 is the name for what fills those gaps: confident, fast, usually wrong. The more context-stripped the input, the more 不正知 has to manufacture. We live in an era with *exceptional* conditions for it.
A suggestive parallel (without claiming equivalence): cognitive psychology's "dual-process" model distinguishes fast, automatic "Type 1" cognition from slow, deliberative "Type 2" cognition (Kahneman's framing). Much of 不正知 runs in the Type 1 layer — fast pattern completion on minimal data. The parallel is illuminating, but it is *not* the same claim: 不正知 is not "bias" in the cognitive-science sense; it is a feature of *afflicted* (染污) mind. The Buddhist diagnosis is older, structural, and tied to the claim about 無明 that the cognitive-science literature does not make.
3.2 注意力經濟的副作用
The attention economy is built on capturing, distorting, and reselling misreadings. Outrage cycles depend on 不正知 at scale — a headline is a 所觀境; the reader's misreading is the product; the engagement is the harvest. Understanding 不正知 at the personal level is a small inoculation against the industrial version: once you can see the mechanism in your own phone-checking, the macro version loses some of its fog.
3.3 不正知是「覺」得以介入的入口
Here is the practical point. You cannot directly transform the root afflictions (貪、瞋、癡) — they are deep. But you *can* work with the moment immediately *before* a root affliction takes hold, and that moment is exactly where 不正知 lives. The classical chain is: 觸 (contact) → 受 (feeling) → 想 (perception / recognition) → 思 (intention) → 愛 (craving). The 謬解 sits at 想 — it turns an ambiguous appearance into a fixed, biased object that 思 and 愛 then chase. Catch 不正知 at 想, and the chain is interrupted. Miss it, and the chain runs.
In cognitive-behavioral language (used as analogy, not equivalence): 不正知 is the cognitive-distortion layer. CBT works precisely because it identifies *named* distortions (catastrophizing, mind-reading, fortune-telling, etc.) and exposes them. The Buddhist diagnosis is older and more radical: the named distortions are *forms* of 不正知, and the underlying issue is not a list of biases but a structural feature of the afflicted mind. The treatment is the cultivation of 正知 (samprajanya) — moment-by-moment clear knowing — which is precisely what the 念處 (satipaṭṭhāna) practice is recommending.
3.4 「照見」的微觀版本
The Heart Sūtra's famous line — 「照見五蘊皆空，度一切苦厄」(the five aggregates seen to be empty, all suffering crossed over) — is sometimes read as a sweeping metaphysical claim. At the level of 不正知, the work is much smaller: *照見* (illuminating-seeing) the moment a single appearance is being misread, and seeing it *as misreading*. That micro-照見 is what weakens the 大八's grip — not a single dramatic awakening, but thousands of small unseens of the misreading.
四、常見誤讀與澄清
4.1 「不正知 = 我不知道」
Wrong. 不正知 is not absence of knowledge; it is *mis-*knowledge. The classical definition — 於所觀境，謬解為性 — is explicit: there is *some* understanding being formed, and it is wrong. The most insidious 不正知 is the kind that feels like *correct* understanding. Calling it "I didn't know" actually protects it from being seen, because "I didn't know" feels innocent. The classical first view (慧一分攝) emphasizes this: 不正知 is *discriminating cognition* running in the wrong direction.
4.2 「不正知 = 邪見」
Mostly wrong, with overlap. 邪見 (*mithyā-dṛṣṭi*, wrong view) is a *fixed doctrinal conviction* — e.g., a denial of karma, a denial of the three jewels, a nihilistic stance. It's a static position, a settled opinion. 不正知 is *dynamic* — it's a misreading happening right now. They reinforce each other (a wrong view will color every perception), but they are not the same. You can have 不正知 without any formal 邪見 (your moment-to-moment misreadings can be running without your holding any grand wrong doctrine), and you can hold a 邪見 without much 不正知 moment-to-moment (rare, but conceptually possible). The classical placement matters: 不正知 is a 隨煩惱 (consequent affliction), not a 根本煩惱 (root affliction). It is *等流* of the roots, not itself a root.
4.3 「不正知 = 無明 / 癡」
Wrong, but related. 無明 (*avidyā*, fundamental ignorance) is the root, the most fundamental affliction. 不正知, in the second of the three views, is a *portion* (分) of 無明, and in the third view is a joint portion of 慧 and 癡. They are not identical. 無明 is the deepest: not-knowing that doesn't even know itself. 不正知 is the brighter cousin: mis-knowing that looks like knowing. The 念慧之難 in the 《成唯識論》 is precisely the argument that 不正知, *as 念慧*, would not pervade all afflicted minds, but *as 無明-derived* it can. The distinction is technical but operationally important: when we work with 不正知, we are working with the surface where the mis-recognizing is *happening*, not with the deepest ground.
4.4 「既然大八遍諸染心，那我無時無刻都在不正知」
Right at the technical level; dangerous as a misunderstanding. Yes: at every afflicted moment, 不正知 is deterministically present. This is the Yogācāra's sober account of how thoroughly afflicted cognition is structured. But the practical implication is *not* despair — it is the basis for the practice of 正知 (samprajanya) and 念 (smṛti). Knowing that 不正知 is *always* there when the mind is afflicted is not a reason to give up; it is the reason to cultivate the antidote at every moment. The 念處 practice is precisely the practice of weaving 正知 through all activities, on the assumption that 不正知 is the default. The Satipaṭṭhāna sections on 明覺 (clear comprehension) and 念 (mindfulness) are the practical response to this teaching, not a contradiction of it.
4.5 「不正知是別人的毛病」
The most common trap. Because 不正知 happens fast and confident, it almost always *feels* like accurate perception when we are inside it. The contemporary habit of pointing to other people's "bias," "projection," "lack of awareness" — without naming the same dynamic in ourselves — is itself a sign of 不正知 at work. The "they're biased, I'm just seeing clearly" reading is precisely the kind of misreading the concept describes. The classical placement of 不正知 among the 大八 that pervade *all* afflicted minds is a structural reminder: this is not a problem others have. It is a feature of how afflicted minds work, including yours, including mine.
4.6 「我只要更努力地思考就好」
Wrong direction. 不正知 is not solved by more thinking. It is, in many of its forms, *caused by* more thinking — by the discriminating function (簡擇) running on insufficient grounds. The classical antidote is not better analysis but better *attending*: 念 (mindfulness, present-moment remembering of the object) and 正知 (clear comprehension of what's actually being done). The 大念處經 explicitly pairs 明覺 (sampajañña) with 念 (smṛti): mindfulness provides the holding, clear comprehension provides the matching. Together they make 不正知 harder to install.
4.7 「AI/算法只是不正知的延伸」
A contemporary parallel worth noting carefully, with limits. Algorithms trained on human-generated data do, in a sense, institutionalize the patterns of 不正知 — they predict based on the same biased shortcuts, and they often make confidently wrong calls on thin context, exactly as 不正知 does. Using them as a mirror can be illuminating: when an algorithm surfaces its prediction of what you'd click / buy / feel, the gap between *its* misreading and *your* misreading can be diagnostic. However — and this matters — algorithms are *not* afflicted minds. They don't have 染污心 in the technical sense; they don't generate 謬解 from 癡. They're statistical approximations. The parallel is suggestive, not equivalent. Do not claim "neuroscience has proven 不正知" or "machine learning has confirmed the Yogācāra." What you *can* say is: the contemporary computational paradigm has independently encountered something like the *pattern* 不正知 describes, and that convergence is a useful invitation to look at the human version more carefully.
結語：作為動詞的不正知
There is something quietly instructive about the 《成唯識論》's willingness to leave the three views of 不正知's nature standing, without resolving them into a single one. The 论's refusal to fix 不正知 to a single substrate is, perhaps, itself a quiet antidote: the mind that *understands* 不正知 well is also the mind that holds its own definitions lightly. 不正知, properly seen, is not a *thing* in the mind but an *act* — the act of taking the object before one and misreading it into a story. To name it precisely is to keep catching it in the act.
That, too, is part of the practice: not to defeat 不正知 once and for all, but to keep shortening the gap between the misreading and the noticing of the misreading — until the gap itself becomes a door, and through the door, the cultivation of 正知 — *samprajanya* — begins.