What This Concept Is Saying
一異 (identity-and-difference, *ekatvānyatva*) is one of the central "two-horned" (*anta*, 邊) negational tools of Madhyamaka (中觀). It does not ask "are things one or are they many?" in order to pick a winner. It asks something sharper: *can either pole — pure identity or pure difference — be grounded in the things themselves?* And it concludes that neither can. The point is not a triumphant third option ("dialectical synthesis") but the exhaustion of the question itself.
In *Mūlamadhyamakakārikā* (《中論》, T1564), 一異 sits inside a family of paired analyses — 去法去者 (going and the goer), 染與染者 (the dyed and the dyer), 上中下 (above/middle/below), 邊無邊 (finite/infinite), 常無常 (permanent/impermanent), 成壞 (arising/ceasing) — that all use the same move: *show that each pole of the pair has no self-standing existence apart from the other, and therefore the relation itself is not independently real*. 一異 is the most abstract of these, and so it functions as a kind of master key.
The classical analysis runs through several distinct moves, and it matters to keep them distinct — none of them on its own does the whole work.
Move One: Identity with the agent/process fails. The MMK opens this thread with 「去法即去者，是事則不然；去法異去者，是事亦不然」 — "If the going is identical to the goer, that does not hold; if the going is different from the goer, that does not hold either." Identity collapses the structure (a process needs someone it happens to); difference collapses the relation (two wholly unrelated things cannot enter into one event). The question is posed directly: 一、异有何过 — what fault lies in saying identity or difference?
Move Two: Difference cannot inhere in the thing itself. 「是法則無異，異法亦無異」 — "In the thing itself there is no difference; in a different thing there is no difference." If "difference" were the very nature of a thing, then 老 (aging) would be self-causing (oldness would simply make itself old); if difference required a *separate* thing to confer it, then 壯 (youth, the term's contrary) would have to be what causes oldness — and that is absurd. The milk-and-cheese analogy (乳酪之喻) sharpens this: if difference were intrinsic, milk would already *be* cheese, no fermentation required; but milk and cheese have many actual differences. If a separate "different thing" had to confer difference, then apart from milk there would have to be something that *is* cheese in itself — but there isn't. So "是法不異、異法亦不異" — the thing is not intrinsically different, nor is it different by virtue of something else.
Move Three: Difference is purely relational. 「異因異有異，異離異無異」 — "Difference exists because of what it differs from; apart from what it differs from, difference has no basis." This is the 相待 (mutual conditioning) point: "difference" is not a property but a position in a relation. Take away the second term and the word loses its meaning. The MMK extends this to arising (眾緣所生): "The house exists because of beams and rafters; if the beams break, the house breaks." The house is not different from its causes, because it has no independent existence it could carry away from them. To posit a fixed "difference" outside the relation would let every thing differ from every other thing without needing any counterpart — which is to say, the word "difference" would mean nothing at all.
Move Four: The "mark of difference" (異相) is doubly unestablished. This is the MMK's quietest and most elegant move. It asks whether, *inside* a thing that is already called "different" (異法), one needs a further "mark" (相, *lakṣaṇa*) of difference. No — that would be redundant. Does a "non-different thing" (不異法) carry such a mark? No — because then it wouldn't be a non-different thing. So in neither location can the mark of difference be placed. With no place to stand, neither "this differs from that" nor the "joining" (合) that was supposed to ride on top of difference can be established. This is what the text means by "凡物雖因異而說有合，然作為合之根據的異相不可得，故合亦不可得."
Move Five: Identity and difference are co-conditioning in the same breath. 「因一故有異，因異故有一」 — "Because of oneness, otherness; because of otherness, oneness." If either had its own self-nature (*svabhāva*), it would not need the other to define it; the fact that each only arises in relation to the other shows that neither is self-standing. The same logic is then reapplied to 成壞 (arising/ceasing), 上中下 (above/middle/below), and finally, decisively, to 「如來於五陰中以一異等方式求皆不可得」 — the Tathāgata cannot be located as identical with or different from the five aggregates. The closing verse asks: 「何者為一異」 — "What, after all, is identity or difference?" — and places the pair alongside 邊無邊, 常無常 as a set of designations swept up in 一切法空 (the emptiness of all dharmas) and 滅一切戲論 (the cessation of all conceptual play).
**Move Six: The *Śataśāstra* (《百論》, T1569) extension — the three kinds of difference.** The *Śataśāstra* takes the same analysis and pushes it into cases the MMK does not formalize so explicitly. A defender of ordinary categories might try to save "difference" by saying it comes in kinds. The text identifies three:
- 合異 (combination-difference): the way substance relates to attribute, as in 陀羅驠 (*dravya*, substance) and 求那 (*guṇa*, attribute) — the bottle is a substance that "joins with" existence and oneness. - 別異 (distinct-difference): the way one person relates to another person — "this one / that one." - 變異 (transformation-difference): the way cow-dung becomes ash — one thing changing into another.
The text then breaks the rescue attempt: if the bottle "joins with" existence to become existent and "joins with" oneness to become one, and the bottle is also a bottle, then there would be *multiple bottles*. Likewise, when the defender says the bottle, existence, and oneness are each distinct (異各各無), then "bottle is not existent, bottle is not one" — the bottle dissolves. When the defender invokes "異合" (joining through difference), the analysis shows the joining requires the difference, but the difference cannot be independently established, so the joining fails too.
Move Seven: One and many, whole and parts. 「初數無故」 — without a "one," there can be no "many"; 「一無故多亦無」 — if oneness has no self-nature, plurality loses its footing too. And 「分有分若一若異，是過先已破」 — "the faults of taking the parts-and-whole as identical or different have already been refuted." If the whole has no independent being, the parts have none either. This is the same logic that, in the *Śataśāstra*, is used against the realist theory that a bottle is *made of* atomic particles.
So when the *Śataśāstra* closes with 「一非所執，異亦爾。一異不可得先已破」 — "Oneness is not what is grasped; otherness is the same. That identity and difference are unattainable has already been refuted [i.e., already refuted the positions of the disputants]" — it is not making a modest logical point. It is announcing the end of the line for one of the most basic operations of human cognition: the operation that says *this* is the same as *that*, or *this* is different from *that*.
走查：生活里的实际勘察
To walk through 一異 in everyday life without losing its precision, we have to keep asking, at each step: *which move in the classical analysis is this scene activating?* Here are some ordinary scenes where 一異 silently runs the show, and where it almost always gets mapped wrong.
Scene 1: "I am my body." Someone says, on a hard morning, *I am this tired body.* That is identity — claiming the person (personhood, self, *pudgala*) and the body (the five aggregates in their physical aspect) are one thing. Now apply Move Two: if "I" were identical to this body, then the body's properties would have to inhere in me intrinsically. But I had a different body ten years ago, and I don't say "I am that body." I am also going to have a different body in twenty years. So was *that* body me, too? If so, then *every* body is me — which means the very notion "my body" loses its bite, since there is no body that is uniquely mine. Identity slides into absurdity.
Now try difference: *I am not my body — I am the one who has it.* But this "me" that has the body — does it have any properties that are not, in fact, body-properties? Memories, moods, intentions? These are aggregate-properties too — the *Vedanā* (受, feeling-tone), *Saṃjñā* (想, perception), *Saṃskāra* (行, volitional formations), *Vijñāna* (識, basic awareness) of classical analysis. To say "I am different from all the aggregates" is to try to install a sixth thing behind the five. The MMK's Move One — going and goer, dyed and dyer — is designed exactly for this. The process needs a "someone" running through it, but that "someone" cannot be pinned down as either identical to or different from the process without collapse.
Where modern people map this wrong: they treat the question as a *philosophical* one, when in fact it is a *phenomenological* one. You do not need to win the argument; you need to notice how, even now, *I am this body* and *I am not just this body* are both running simultaneously, and how exhausting that double-track is.
Scene 2: "We've grown apart" / "We are one." In a long relationship, two phrases alternate: *"we used to be the same person"* and *"we have become different people."* Neither holds up under 一異 analysis.
Apply Move Five — *because of oneness, otherness; because of otherness, oneness.* The version of the partner who felt like "the same as me" only registered as same because there was some friction, disagreement, surprise to compare against. Pure identity would not even be perceptible as identity; it would not register. The version of the partner who now feels "completely different" only registers as different because there is still a baseline recognition ("this is still them") against which the difference appears. Pure difference would be — and is, in the case of strangers — simply non-recognition, not "difference from."
Apply Move Six to this scene: the partner seems to combine three different "異"s at different moments — 別異 (a distinct person, separate center of agency), 合異 (a bundle of qualities and roles that "join with" my life), and 變異 (a person changing over time). On a bad night, you only see 別異 ("a stranger"); on a warm night, you only see 合異 ("my partner, woven into me"); on a nostalgic night, you only see 變異 ("they have changed"). The 一異 analysis shows that none of these three is the "real" nature of the partner; each is the partner appearing through a conceptual lens, and each lens collapses if pressed.
Where modern people map this wrong: they pick one of the three "異"s and treat it as the truth of the relationship. *Actually, they are a separate person* — that's the Western individualist romance. *Actually, we are joined into one unit* — that's the romantic-fusion fantasy. *Actually, people just change* — that's the stoic resignation. The 一異 analysis does not say all three are wrong; it says all three are *ungrounded* — each only functions by leaning on the others.
Scene 3: "Is this still the same company?" A founder's company has been bought, half the staff replaced, the product pivoted three times, the original mission statement forgotten. A customer asks: *is this still the same company?* Apply Move Three: 異因異有異, 異離異無異 — "difference exists because of what it differs from." "Difference from what?" If you mean difference from the 2015 company, you have to posit a 2015 company that has an independent identity to differ from. But the 2015 company was also a process — different staff, different products, different revenues each year. By the time you fix the original, you have moved the target so far that "this company" has no continuous identity anyway. Conversely, if you say "it's the same company because the legal entity is unchanged," then the *legal* identity is one thin thread held by paperwork, and you are conflating legal continuity with substantial continuity — exactly the move that 一異 analysis blocks.
Where modern people map this wrong: they think 一異 is "everything changes, nothing stays the same" — Heraclitean flux. That is Move Three taken alone, which is not 一異. 一異 also refuses the other side: "the company is *just the same*" — that's the legal-fiction reification. Real 一異 holds both at once and refuses both as fixed natures.
Scene 4: "That's not the real me." A person catches themselves in a behavior they despise — a burst of anger, a craving, a capitulation — and says, *"that wasn't the real me."* This is the classical identity move: there is a "real me" who is *different from* the act. Apply Move One (染與染者, the dyed and the dyer): if the anger is "not really me," what is the "me" that the anger happened to? If the anger were wholly other to me, I could not own it at all, and the sentence *"that wasn't me"* would be unnecessary — I would not even feel implicated. The very feeling of betrayal by one's own act is what shows that identity and difference are tangled here, not separable. 一異 analysis does not say "the anger *is* the real you"; it also does not say "the anger is *not* the real you." It says the question *real-me-or-not-real-me* is itself unestablished.
Where modern people map this wrong: they think 一異 is letting them off the hook ("nothing is really me, so no responsibility"). It is doing the opposite. By refusing both identity and difference, it makes the simple evasion *"that wasn't me"* impossible — because if it weren't me in any sense, why is there a story to tell? 一異 sits between *rigid self-blame* and *identity-diffusion*, refusing both.
Scene 5: "The model and the data." Take a contemporary case. A language model is trained on a corpus. After training, the model contains — in some sense — traces of that data. A user asks: *is the model the same as the data, or different from it?* Apply Move Three: 異因異有異 — difference only appears against a background of what it differs from. Apply Move Four: there is no "mark of difference" we can point to that settles this. The model behaves in ways that resemble certain stretches of training data; it also produces outputs that are nowhere in the corpus. It is neither the data (no single datum survives in its original form) nor other than the data (no output is independent of the training distribution). It is precisely the kind of thing 一異 was designed to talk about: a phenomenon whose being is entirely relational and yet resists being reduced to either pole of identity or difference.
It would be wrong to say "Madhyamaka predicted language models." 一異 is a tool for examining how we constitute *any* phenomenon, and language models are a particularly clear contemporary case where our usual identity/difference intuitions visibly fail.
Why Contemporary People Need This
The 一異 analysis is not an abstract exercise because the modern self and the modern world are *built* out of identity-and-difference claims that have become untenable.
The demand to be "authentic" and "to grow." Contemporary identity discourse asks people simultaneously to be *the same as* their past selves (authenticity, "stay true to who you are") and *different from* their past selves (growth, evolution, "you've changed for the better"). These two demands cannot both be grounded in a self with fixed nature — exactly the situation 一異 describes. The exhaustion people feel from this double demand is not a personal failing. It is the symptom of a category error: trying to ground "I" in either pure identity or pure difference.
Brand identity and corporate identity. Marketing copy constantly asks whether a brand is *the same as* its origins or *different from* them. The 一異 analysis shows why every rebrand feels slightly false: because the "original" itself was never a fixed identity, only a frozen moment in a process. But it also shows why every rebrand that abandons all continuity feels empty: because the difference only registers against some baseline. A more honest framing — possible through 一異 — would stop asking "is this still the same?" and ask "what conditions are producing this appearance now?"
Identity politics, both progressive and reactive. Modern discourse is saturated with claims about who is "the same as" a historically oppressed group and who is "different from" it, who counts as "really" a member of a category and who is "performing" it. 一異 does not adjudicate these debates, but it makes visible why they cannot be settled: the categories being deployed rest on identity/difference claims that cannot be grounded in self-nature. The move is not "all categories are equally valid" (that is the error of relativism); it is "categories function relationally and conditionally, and any attempt to fix them in self-nature fails."
The dissolution of boundaries in digital life. Work and life, public and private, self and curated self, human and AI-assisted text — these used to come with reasonably stable identity-markers. Now they do not. One's "real" self is on display in the workplace via Slack; one's "professional" self appears in casual group chats; one's anonymous self leaves traces via writing style. The 一異 analysis is not a tool to *fix* this disorientation. But it can unstick the demand that *some* of these must be the "real" one, and *some* must be the "fake" one.
Why not just use the word "emptiness"? Many readers will know *Śūnyatā* (空, emptiness) and assume 一異 is a specialized sub-case of it. That is fair, but the practical value of 一異 specifically is that it names a *recognizable cognitive operation* — the operation of saying *this is the same as that* or *this is different from that* — and shows that this operation, when examined, does not have the foundation we assume. This is more directly useful in everyday self-construction than the more abstract term 空 (emptiness), because most suffering and confusion in modern life is built out of faulty identity/difference judgments rather than out of metaphysics.
Common Misreadings and Clarifications
Misreading 1: 一異 means "everything is one" or "everything is many." This is the most common Western reception: 一異 collapses into monism ("all is one") or into a vague pluralism ("many things, each its own"). Both readings invert the analysis. The MMK is not picking *one* pole over the other; it is showing that *neither pole, nor their combination, can be self-standing*. 「因一故有異，因異故有一」 is not a metaphysical statement; it is a diagnostic showing that each pole only functions by leaning on the other. To settle on "all is one" is to commit the same identity-fallacy 一異 was designed to expose.
Misreading 2: 一異 is a logical puzzle about nouns. Readers trained in analytic philosophy sometimes hear 一異 as a series of problems about reference and predication — is the bundle identical to its parts? — and treat it as a precursor to the bundle theory or to mereological nihilism. The *Śataśāstra* section on 有分/分 (whole and parts) does engage that territory. But 一異 is not a contribution to logic; it is a tool for the dissolution of *upādāna* (執取, grasping). The aim is not to find the right account of parts and wholes but to undermine the *grasp* on parts-and-wholes as fixed natures. Without that aim, 一異 becomes a sterile word-game.
Misreading 3: 一異 refutes difference and confirms sameness. Some readers, after seeing Move Two break the case for *intrinsic* difference, conclude that 一異 teaches "all is the same." This is the exact inversion. Move Two is symmetric — it also breaks the case for *intrinsic identity* (老 cannot self-cause). And Move Six — the three kinds of difference in the *Śataśāstra* — is not a denial that things appear different; it is an analysis of *how* difference appears (combination, distinctness, transformation) without that appearance needing to be grounded in self-nature. Phenomena continue to appear; what is refused is that any appearance has an underlying essence it carries around with it.
Misreading 4: 一異 is just "two sides of every coin." This is the lazy dialectical reading — Hegelian or pop-philosophical. It treats 一異 as a kind of cosmic balance: every identity has its difference, every difference its identity. This preserves both poles as real; it just says they go together. 一異 refuses precisely this. 「異因異有異，異離異無異」 is not saying "identity and difference coexist nicely." It is saying that *neither has independent being*, and the very way they appear together is part of what shows that. The "two sides" reading is a watered-down version that misses the *unestablishability* of both poles.
Misreading 5: 一異 is only relevant to deep meditation states, not to ordinary life. Some traditional commentaries emphasize 一異 in the context of advanced *bhūmi* (地, stage) practice or the path of seeing (見道, *darśana-mārga*). But the actual analyses in the MMK and *Śataśāstra* use everyday examples — the goer and the going, the bottle and its qualities, the house and its beams, cow dung becoming ash. These are not esoteric; they are phenomena anyone can examine. As the *Vimalakīrti* and *Prajñāpāramitā* traditions emphasize, the analysis of identity and difference is operational right now, in how one is constituting one's relationships, one's self-story, one's work.
Misreading 6: 一異 is solved by science or by neuroscience. It is tempting, in a contemporary idiom, to say "neuroscience shows the self is a constructed process, so 一異 is vindicated." This is a category mistake. 一異 does not say "the self is a constructed process and we should know that." It says that *neither identity nor difference can be self-established* — and that is a stronger claim than "the self is constructed." A constructed-process account can still ground a new identity-claim ("I am my constructed process"), and a neuroscientific account can still fix identity in a substrate ("the self is the brain"). 一異 is not finished until neither ground is accepted. AI analogies — as in Scene 5 — can *illustrate* the difficulty of identity/difference, but they cannot *prove* the analysis, because the analysis concerns the way appearances are constituted in experience, not the empirical behavior of any particular system.
Clarification in closing. 一異 is not a conclusion but an *operation*. It is the operation of repeatedly asking, in any situation where you have caught yourself saying *this is the same as that* or *this is different from that*: *on what ground?* — and watching the ground, every time, fail to hold. The closing verse of the MMK asks 「何者為一異」 — "what, after all, is identity or difference?" — not as a question expecting an answer but as the final dissolution of a question that, until that moment, had seemed to demand one. The instruction is to live inside that dissolution long enough that the demand itself softens — and then to see what the world looks like.