这个概念在说什么
Classical entry says only:「對分別作用所作的三種類別」— "the three categories made regarding the function of discrimination." That's a deliberately spare definition, because in the Yogācāra (唯識, *Consciousness-Only*) tradition the analytical work has already been done in larger texts like the *Yogācārabhūmi Śāstra* (《瑜伽師地論》) and Xuanzang's *Chéng Wéishì Lùn* (《成唯識論》, *Treatise on the Establishment of Consciousness-Only*). What the entry is pointing to is an anatomy of one verb: 分別 (vikalpa) — the activity of discriminating, marking off, differentiating — broken into three distinguishable modes.
The reason this matters: most people think "thinking" is one thing, roughly equivalent to the inner voice. The Yogācāra analysis says no — what we casually call "thinking" is actually three layered activities, each with its own logic, and most of our suffering is tangled up in *all three*, not just the obvious one. To recover agency, you have to see all three.
Here are the three, with their classical definitions kept intact.
1. 自性分別 (Zìxìng Fēnbié / *svabhāva-vikalpa* — discrimination of self-nature)
This is the most immediate layer. The classical formulation: 於一切時、隨境而起，於自境界、任運而轉，分別自性 — "at every moment, arising along with its object, in its own sphere turning spontaneously, discriminating [the dharma's] self-nature."
In plainer words: this is the discriminative activity that *simply registers* what's appearing — the raw "this-ing" of experience, before you've named it, judged it, or remembered anything. When light hits your eye and the seeing-consciousness distinguishes "color" from "shape," that distinction itself is 自性分別. When your skin touches a surface and the body-consciousness marks "warm" vs. "rough," that marking is 自性分別.
The key classical points:
- It is 任運 (*rèn yùn*, spontaneous, automatic) — not willed, not chosen. It runs by itself with every contact between sense organ, object, and consciousness. - It is already a form of 分別 — already a "marking off," already a subtle conceptualizing. It is *not* pristine non-conceptual awareness (那是另一回事). The Yogācāra tradition is precise here: even at the most basic level of perception, consciousness is doing a differentiating work. - It is the ground layer on which the other two can operate.
People sometimes want to make 自性分別 sound mystical or special. It is special only in the sense that it is the hardest to detect — because it is always already running whenever you are conscious at all.
2. 隨念分別 (Suíniàn Fēnbié / *smṛti-vikalpa* — discrimination of recollection)
Classical formulation: 於曾受境、隨念而起，分別曾受 — "with respect to previously-experienced objects, arising through following-thought, discriminating what was previously experienced."
The Chinese 隨念 is two characters doing real work: 隨 means "to follow / along with," 念 means "thought, memory, recollection." So 隨念分別 is the discriminative activity that follows the trace of past experience into present discrimination. When the smell of a particular dish triggers the recognition "this is cumin," memory is doing its discriminative work. When you see someone across a room and recognize them as your colleague, that recognition — that act of *identifying* — is 隨念分別.
The key classical points:
- It operates on 曾受境 (*zéng shòu jìng*) — "previously-experienced objects." It needs a stored trace to fire. Without prior experience, 隨念分別 has nothing to follow. - It is the bridge between past and present discrimination. Without it, every moment would be a disconnected atom. With it, the world has continuity, familiarity, the texture of "this is like that thing from before." - It is active, not passive. 隨念分別 is not "memory sitting in a drawer." It is the *work* memory does — comparing, matching, recognizing, tagging the present through the lens of the past.
This is also why it can become a source of trouble: when memory's recognition habit is too rigid, you start filtering everything new through what is already familiar.
3. 計度分別 (Jìdù Fēnbié / *kalpanā-vikalpa* — discrimination of calculation/deliberation)
Classical formulation: 於非曾受境、計度而起，分別非曾受 — "with respect to not-yet-experienced objects, arising through calculation, discriminating what is not yet experienced."
The character 計 means "to count, measure, calculate"; 度 means "to estimate, weigh, gauge." So 計度分別 is the discriminative activity of measuring and weighing — reasoning, inferring, hypothesizing, planning, building arguments, spinning narratives. This is the one most people recognize as "thinking."
The key classical points:
- Critically, it can operate on 非曾受境 (*fēi zéng shòu jìng*) — "not-yet-experienced objects." This is what gives it an almost unlimited reach: imagination, future-planning, counterfactual thinking, speculation, fiction, worry about things that may never happen. *This is the only one of the three that can generate content beyond what has been directly experienced.* - It is the most gross (*最粗*) layer — easiest to detect, easiest to narrate, easiest to get lost in. - It is the layer where language does most of its work. Where 隨念分別 recognizes, 計度分別 constructs whole sentences, scenarios, justifications, theories.
Notice that the three are not three separate "minds." They are three modes of one continuous activity. 自性分別 happens *first* (the immediate mark), 隨念分別 *follows* (memory fills in), 計度分別 *extends* (reasoning projects forward and sideways). In any given moment of experience, all three may be running, layered, almost seamlessly.
Where does this sit in the path?
三分別 sits inside the Yogācāra analysis of how the phenomenal world (the world of experience) gets constructed. The broader claim is: what we take as "external objects" are *presentations* of consciousness, and these presentations are built up precisely through the work of vikalpa in its three modes. So the practical path of *seeing this* (唯識觀, *consciousness-only contemplation*) is not to suppress the three — that's neither possible nor what the teaching asks. It's to *see how they operate*, until the construction becomes transparent. In the larger 唯識 map, this is groundwork for understanding how attachment (執) and the seventh consciousness (末那識, *manas*) take these discriminations as self and cling.
A note on metaphors: unlike 五蘊 (the five aggregates), which come with the famous images of foam, bubbles, mirage, banana, and magic show, 三分別 doesn't have a matching set of poetic similes. The three are more anatomical than imagistic. So instead of compressing into one image, the practice is to *hold them apart as three* — which is itself the work the concept is asking of you.
生活里的走查
Let's walk this through with three scenes a contemporary reader will recognize.
Scene 1: The subway commute, eight in the morning
You're packed into a rush-hour train. Try to slow down and notice what's happening:
- 自性分別 is already running. Before you've formed a single thought, your eye-consciousness is discriminating "bright light above" from "shadow below," your ear-consciousness is registering the low rumble of the train, your body-consciousness is marking "pressure against shoulder" from "warmth from a backpack against your back." None of this is yet a thought. It's the immediate, automatic, 任運 work of discrimination. You're not choosing it. It's the ground layer.
- 隨念分別 kicks in as memory tags the present. "That smell is that bakery near the second stop." "That voice through the speaker sounds like the same conductor as last week." "That face in the newspaper over someone's shoulder is the politician from the news cycle last month." Memory is *following* stored traces into the present, *recognizing*, identifying. The world becomes familiar.
- 計度分別 lights up. "If I miss this stop I will be late. If I'm late I will have to apologize. Why does this train always slow down here? Did they change the schedule? Should I leave earlier tomorrow? Is it worth switching lines?" Now you are *calculating*, measuring, projecting forward into scenarios that may never happen. The inner voice is spinning.
The most common mapping mistake here: thinking only the third one — 計度分別 — counts as "thinking," and therefore believing that "quieting the mind" means just quieting the inner voice. But 自性分別 will still be running even in deep meditation (some Yogācāra commentators argue it cannot be stopped in this lifetime); 隨念分別 will still be filling the present with recognition and familiarity. The classical categories tell you that *all three are at work*, and a path that only addresses one is missing most of what's actually happening.
Scene 2: Late-night scrolling
You're in bed, phone in hand, and a feed is updating.
- 自性分別: the screen's light, the motion, the colors — registered immediately by seeing-consciousness. The phone is *appearing* to you, distinguished from the room around it, before you have any thought about it.
- 隨念分別: "Oh, this is the same person who posted yesterday." "That thumbnail is from that channel I usually skip." "I know this image — I saw it three weeks ago." Recognition, tagging, the sense of "this is familiar / this is new / this is who I follow." Memory is doing its discriminating work, fast and quiet.
- 計度分別: "I should sleep. But this might be interesting. Why am I still awake? Tomorrow I'll regret this. But also — does this person actually believe what they just said? Let me read the comments. Why are they arguing about that? I used to think that too, but now I'm not sure..." A whole discursive storm that can run for an hour, *most of it about content you have never experienced and never will experience again*.
The mapping mistake here: thinking the issue is the *content* ("I shouldn't be looking at this stuff"). The Yogācāra analysis says the issue is the *operation of vikalpa itself* — three layered activities that, once engaged, will continue generating experience. You can change the content. The structure of the engagement is what carries over.
Scene 3: An argument with someone you love
You and your partner (or close friend, or parent) are in a fight.
- 自性分別: your body-consciousness registers the heat in your chest, the tension in your jaw, the higher pitch of your own voice, the specific look on the other person's face. These are immediate, raw, 任運 markings of the situation.
- 隨念分別: "This is exactly like the fight we had in March." "She always does this." "That tone of voice is the same one his mother used." Memory is following past traces into the present and tagging it: familiar, known, *this kind of moment again*. Recognition happens so fast you don't notice it as a separate step.
- 計度分別: "She's doing this because of her father. If I bring up the kitchen incident she'll explode. I should stay calm. No, I shouldn't let her get away with this. What about what *I* need? Why am I always the one who has to... ?" The full deliberative machinery — past patterns being applied to present, future scenarios being rehearsed, justifications being built.
The mapping mistake here: *and this is important* — believing that what you are feeling is the situation itself, rather than the layered activity of the three vikalpas. The other person did not deliver the full content of "she always does this" — your 隨念分別 did. The other person did not lay out the elaborate scenario of "if I bring up X then Y then Z" — your 計度分別 did. The body-sensation is closer to what was actually delivered; the rest is constructed.
当代人为什么需要它
Three reasons this classical anatomy is useful right now.
1. The modern world loads each layer differently than premodern environments did.
- 自性分別 is being exploited at industrial scale. The attention economy — every app, every interface, every notification — is engineered to hijack the most immediate discriminative layer. You don't even need to think about it; your eye-consciousness discriminates the moving color, your body-consciousness discriminates the vibration, and you're already pulled in. Understanding that 自性分別 is *already a form of vikalpa* — already a subtle conceptual marking, not pristine seeing — is useful precisely because so much of the modern environment is calibrated to that layer.
- 隨念分別 is being saturated. Social media is essentially a 隨念分別 amplifier: every moment tagged, archived, made instantly recallable and comparable to past moments. The classical insight that 隨念分別 operates on 曾受境 (*previously-experienced objects*) becomes poignant in an era where the "previously-experienced" is now a searchable database of every photo, message, and reaction you have ever produced.
- 計度分別 is being overworked without being named. Most contemporary advice about stress, productivity, and anxiety addresses the inner voice — the planning, worrying, narrating mind. But because the popular framework treats this as the *whole* of thinking, the other two layers keep operating underneath, and people feel like they've "tried to stop thinking" and failed. The Yogācāra framework tells you: you were only addressing one-third of the activity.
2. Mindfulness practice, as it has spread in contemporary culture, is implicitly a practice with the three layers — but rarely names them as such.
When someone practices "noticing" without judging, what they are often doing is bringing awareness to the *output* of 自性分別 before 隨念分別 and 計度分別 have fully taken over. When someone practices "name it to tame it," they are usually intervening at the 計度分別 level — naming the inner narrative to loosen its grip. When someone practices "remembering" — bringing a past situation into present awareness to learn from it — they are working with 隨念分別. Naming the three layers makes it possible to *choose where to intervene*, rather than assuming the only place to work is the loudest layer.
A note of caution: contemporary neuroscience has rich things to say about perception, memory consolidation, and default-mode-network activity (the "inner voice"). Some commentators draw parallels between 計度分別 and the default mode network, between 自性分別 and early sensory processing. These parallels can be suggestive. They are not claims that neuroscience has *proven* the Yogācāra analysis, and the Yogācāra analysis is making different kinds of claims (about the structure of experience, not about brain tissue). Useful analogy; not equivalent. The classical category is finer-grained and more philosophically loaded than any single neuroscientific construct.
3. AI is, in a real sense, a 計度分別 machine — which is why understanding the three layers clarifies what AI is not.
Large language models can calculate, weigh, combine, generate content on objects they have never experienced. By the classical definition, this is the *exclusive domain* of 計度分別. They do not, in the Yogācāra sense, have 自性分別 that arises *with* an object in real-time perceptual contact. They do not, in any straightforward sense, have 隨念分別 as a continuous bridge of recognition flowing into present discrimination. The interesting question for contemporary readers is not "is AI conscious?" — it's "what does it mean that we have built a system that operates at one layer of vikalpa but lacks the other two?" That question becomes available only if you have the three-layer framework.
常见误读与澄清
Misreading 1: "三分別 = thinking." The most common collapse. People translate 分別 as "thinking" and assume the three are three kinds of thoughts. They are three kinds of *discriminative activity*, and the most immediate one (自性分別) is not a thought in any ordinary sense — it is the running, automatic marking that is the precondition of having any thoughts at all. Reducing 三分別 to "thinking" misses two-thirds of the concept.
Misreading 2: "自性分別 is non-conceptual / pure awareness / original mind." A seductive misreading. 自性分別 is subtle and immediate, but it is *still* a form of 分別 — still a marking, still a discriminating. The Yogācāra tradition is explicit that even here, consciousness is *doing* something conceptually. Treating 自性分別 as a synonym for original mind or Buddha-nature flattens the analysis and skips over the entire point of the Yogācāra critique: experience is constructed *all the way down*, not just at the level of discursive thoughts.
Misreading 3: "The three can be turned off." They are structural features of how ordinary consciousness operates. The path is not to "turn them off" by force. It is to *see* them clearly — to recognize 自性分別 as 自性分別, 隨念分別 as 隨念分別, 計度分別 as 計度分別 — and to see, through that seeing, that what they construct is not what it appears to be. This is the *唯識* move: the cure is not the abolition of vikalpa, it is the transparency of vikalpa.
Misreading 4: "隨念分別 = memory = past." Memory is the *material* 隨念分別 works on, but 隨念分別 is not identical to memory-as-storage. It is the *active discriminative use* of past experience to shape present discrimination. When you recognize, identify, tag, or "place" something as familiar — that *placing* is 隨念分別 at work, even if you wouldn't call the underlying storage "memory" in the technical sense. Treating 隨念分別 as just "remembering things" misses its function as a continuous present-tense activity.
Misreading 5: "計度分別 is bad / wrong / unspiritual." The classical categories are descriptive, not moral. 計度分別 is what allows you to plan, write, deliberate, reason ethically, understand a teaching. The 唯識 tradition is not against it; it is against being *unseen* by it. The trouble is not the calculation itself; the trouble is when the calculation takes itself to be the whole of experience and is not recognized as one of three layered activities.
Misreading 6: "三分別 is the same as 三受 / 三毒 / 三學 / 三慧 / 三無漏學." Different "threes" in the Buddhist taxonomy are doing different jobs. 三受 (three feelings: 苦 / 樂 / 不苦不樂) classifies the *felt tone* of experience. 三毒 (three poisons: 貪 / 瞋 / 癡) classifies the *afflictive roots*. 三學 (three trainings: 戒 / 定 / 慧) classifies the *path*. 三分別 classifies the *structure of discriminative cognition*. Mixing them up is easy because they share a "three" shape, but they answer different questions.
Misreading 7: "If I have understood 三分別 intellectually, that is the seeing." Knowing the names is not the practice. The practice is to actually notice, in real time, which of the three is firing — the immediate marking, the memory-tag, or the discursive spin — and to notice that they are operating *together*, *most of the time*, in any moment of experience. This is the work, and it is the work that the conceptual framework is in service of.
If you take only one thing from this: most of what you call "thinking" is actually three layered activities you haven't separated, and the part that gives you the most trouble may not be the part you think it is. The loudest one — the inner voice, the spin, 計度分別 — gets all the attention, but the quietest one — the immediate 任運 marking, 自性分別 — is running all the time underneath, including when you believe you have "stopped thinking." The classical analysis hands you a more accurate map of your own experience. What you do with the map is the next step.