一、这个概念在说什么
Plain definition, in one breath
自宗 is the position you yourself uphold — and therefore have to live with the logical consequences of. It stands against 他宗 (tāzōng, "the other's thesis"), which is whatever your opponent in a debate is committed to. The moment you say something, you have a 自宗 on that point: a claim you own, defend, and must not contradict in your next breath. You can call it "my view," "my thesis," "what I take to be the case," "what I accept" — in the source texts it shows up under several near-synonyms: 自許 (zìxǔ, "self-assertion"), 自受 (zìshòu, "self-acceptance"), 所宗 (suǒzōng, "that which is taken as thesis"), 所立宗 (suǒlìzōng, "the thesis one establishes"), 所許 (suǒxǔ, "what one accepts"), 受許 (shòuxǔ, "accepted and asserted"). These are not six different things; they are six angles on the same act: *standing behind a claim*.
This sounds almost trivial — of course I stand behind what I say. But the classical material is built around a deeper worry: *most people don't actually know what their 自宗 is*. They argue as if they had a thesis, but if you trace their actual commitments, those commitments are vague, evasive, or self-undermining. 自宗 is the discipline of making your commitments visible, defensible, and — crucially — survivable under your own arguments.
Where it sits in the path of liberation
Buddhist debate logic (因明 / *pramāṇa*) is not an academic sport. It is a tool for cutting through the deep, half-noticed commitments that bind a being to *saṃsāra* — the round of confusion and suffering. Liberation is not just feeling peaceful; it is the removal of those hidden positions ("the self exists ultimately," "things exist by their own nature," "my view is the only correct one") that operate as the load-bearing walls of confusion.
Two things follow:
1. You cannot remove what you cannot see. If you don't know your own thesis, you cannot examine it. 2. You cannot remove it just by attacking others. If the opponent's position crumbles but yours was never clearly stated, then nothing has actually been dissolved.
自宗 is therefore the spine of the whole practice of intellectual purification (view, meditation, conduct). It is the practice of *knowing what you hold* so that what you hold can be held lightly, examined, and — if it turns out to be a wrong view — let go.
The anatomy of 自宗: what counts as your thesis
This is where many readers get tripped up. They think a thesis is something you explicitly state ("I assert that X"). But the classical material makes clear that 自宗 has two layers, and both count.
The explicit layer. When a treatise or a debater writes "I accept" (我許), "my position is" (我宗), "this is how it is" (此為如是), those are clear markers. The 《菩提道次第廣論》 (Lamrim Chenmo) lists exactly this kind of marker — phrases like 「不許名言，我等不能說」 ("if I do not accept conventional language, we cannot speak"), 「於我宗一切名言無斷滅失」 ("in my position there is no fault of cutting off all conventional phenomena"), and the summary 「已說四宗」 ("the four theses have been presented"). Each is an explicit act of standing behind a claim.
The implicit layer. But a thesis is also identified *by what you actually build*. 《菩提道次第廣論》 is explicit: "the criterion is not only whether the text plainly says 'I assert' or 'my position'; any passage where the treatise decides 'this is so,' 'this is not so,' 'this exists,' 'this does not exist' — all of that can be taken as what the author accepts." (Those judgments are *de facto* commitments even when unflagged.) So if a Madhyamaka (Mādhyamika, proponent of emptiness) treatise spends three chapters showing that things lack inherent existence (自性, *svabhāva*), and never once writes the words "I assert this," the absence of those words does not give the author a free pass. The whole treatise is a standing-behind.
This is a quietly devastating observation, and we will see in the next section why it is so important for contemporary life.
The crucial move: not "no thesis," but "no thesis of inherent existence"
Now we get to the heart of the classical debate. The Madhyamaka school is famous for slogans like 「無宗」 ("no thesis"), 「我無所宗」 ("I have no thesis"), 「無宗無所諍」 ("no thesis, no dispute"). A naïve reader concludes: *the Madhyamikas claim to hold nothing.* This is wrong, and getting it right is the entire point.
The 《入中論善顯密意疏》 (Commentary Clarifying the Intent of the Madhyamakāvatāra) handles this with surgical care. The opponent throws an accusation: "If you do not establish your own thesis and only refute others', you are a 破法人 (*pòfǎrén*, 'Dharma-destroyer,' one who only tears down without building)." The commentary preserves the charge in its exact technical form — 「若不立自宗唯破他宗，名破法人」 — and then answers it not by denying the slogan but by re-cutting what the slogan actually excludes.
The core verse is a marvel of precision:
「誰定有宗乃有過，我無此宗故無失。」 *"Whoever fixes on a thesis-with-inherent-existence has the fault. I have no such thesis, therefore no fault."*
The critical word is *定* — "fixes on," "determines as," "holds as definitively established with its own nature." The Madhyamaka does not claim to have no thesis whatsoever; it claims to have no thesis that is **有自性 (yǒu zìxìng, possessed of inherent existence, *svabhāva*)**. This is a tiny distinction that changes everything.
To make it concrete, the same commentary gives the contrasting pair:
「他計因果實有自相，自許如幻都無自性。」 *"They (the opponents) take cause and effect as having true self-characteristics; I assert [cause and effect as] like an illusion, completely without inherent existence."*
Both sides have a 自宗 on the matter of cause and effect. The difference is the *manner* of holding. One side grips cause and effect as having self-existing reality (實有自相). The other side accepts cause and effect — they do not deny cause and effect — but holds them "like an illusion" (如幻). Cause and effect are conventionally present; they function; they can be discussed; they are not dismissed. What is refused is the claim that they exist by their own nature from their own side.
This is why the same source can say both:
- *On the level of ultimate analysis (勝義, *paramārtha*)*: the Madhyamaka does not establish a thesis (no宗of自性). The refuter does not refute by self-nature; the refuted is not destroyed by self-nature. - *On the level of conventional discourse (名言, *vyavahāra*)*: the Madhyamaka fully accepts refuter and refuted, and accepts that he refutes the opponent's refuted-object.
Two layers. Two registers. One person. Two sets of commitments, held together without collapse. This is the Madhyamaka's actual 自宗: not the absence of a thesis, but a thesis that does not freeze.
Two ways of arguing: 應成 (yìngchéng) vs. 自續 (zìxù)
Closely tied to the above is a distinction the 《菩提道次第廣論》 uses to pin down what kind of thesis the Madhyamaka is willing to *positively argue* for. The two terms come from Sanskrit: prasaṅga (consequence-only reasoning) and svatantra (autonomous inference).
- 應成 (prasaṅga) — "consequence reasoning," "reductio." You take the opponent's *own* premises and show that those premises lead to absurd or unwanted conclusions in their own system. You do not need to assert your own positive premise. You let the opponent's position collapse under its own weight. The famous Candrakīrti style. - 自續 (svatantra) — "autonomous syllogism." You build your own three-membered argument (thesis, reason, example) from your own premises. This is what most classical Indian philosophers (and modern academics) do: state your view, give a reason, give an example.
The puzzle: which one does Madhyamaka actually use? Some interpreters, taking the "no thesis" slogan to its logical conclusion, claim that Madhyamaka uses *only* 應成 and never 自續 — they never positively assert anything; they only show that opponents' views entail contradictions. The 《菩提道次第廣論》 registers this position carefully, then dismantles it.
The dismantling has several parts. First: even the slogan "I have no thesis" is itself an assertion. To say 「無許」 ("no assertion") already exhibits a kind of assertion. Second: the position that "everything is established only via the other" (唯就他立) — i.e., you only have positions because your opponent does — *still* requires you to recognize that there is an "other" doing the asserting, a "self" doing the recognizing, and a fact of the matter being established via that recognition. You cannot avoid the 自宗 of *having a thesis about how thesis-establishment works* by hiding behind the rhetoric of having no thesis. Third: if you refuse to accept that 應成 reasoning itself counts as your own thesis, you cannot even coherently say you are a proponent of 應成.
The 《菩提道次第廣論》's positive conclusion: Madhyamaka has a 自宗 — but it is a *離過之宗 (líguò zhī zōng, "faultless thesis")*. In worldly matters not under ultimate analysis, it accepts that things exist. In the matter of its own thesis, it expresses that thesis "like an illusion" (如幻) — present, functional, but not fixed by its own nature. The Madhyamaka accepts that it speaks, asserts, refutes, and teaches — and it accepts that none of these acts is self-grounding.
「汝語唯壞汝自宗」 — the moment a refutation backfires
Perhaps the most quoted phrase in the entire source material, and certainly the most practically useful, comes from the verse:
「唯能壞汝自宗」 「汝語唯壞汝自宗」 *"It can only damage your own thesis" / "Your words only damage your own thesis."*
The context is the structure of debate. Suppose an opponent raises what looks like a refutation against your position. The first question is not "did the refutation succeed against me?" but "did the refutation *also* succeed against the opponent's own position, or *only* against mine?" A refutation that, when applied symmetrically, also undoes the very person who uttered it — that refutation has no force. It is a 似能破 (sìnéngpò, "seeming refutation," a refutation that only appears to work).
The principle generalizes: *the strength of any refutation is measured by whether it could survive being aimed back at its own speaker*. If not, the refutation is hollow.
This is the most portable single lesson in the entire classical material. It does not require any Buddhist metaphysics to apply. It is the principle of *symmetric accountability* in argument.
二、生活里的走查
The classical material is technical; its insights are not. Let us walk through five concrete contemporary scenes, mapping each move carefully back to the classical definitions.
Scene 1: The online argument that only refutes itself
You are scrolling through a thread where two people are arguing about corporate layoffs. Person A says, "Companies that lay off workers during record profits are moral failures." Person B replies, "And you, who probably own stock in those very companies through your retirement fund, are doing the moral failing yourself — you have no standing to criticize."
Now apply the lens.
What is B's 自宗 on the moral question? At least partly this: "those who benefit from a system have no standing to criticize it." Apply B's own principle to B themselves. B presumably also holds retirement investments, buys from large corporations, etc. If B's thesis is "beneficiaries have no moral standing to criticize," then B is also a beneficiary of countless systems B does not criticize. B's words only damage B's own thesis (汝語唯壞汝自宗). The refutation is symmetric: it applies to the critic's own position. By the classical rule, this refutation is a 似能破 — it only looks like a refutation; in fact it has no force.
Where do most people map this wrong? They treat the refutation as *ad hominem* and therefore dismiss it as automatically illegitimate. The classical move is subtler: the refutation is not illegitimate; it is *self-defeating*. It can only be valid if the critic is somehow exempt from their own principle. If they are not, the refutation collapses *together with* the critic's own position. To rescue their refutation, the critic must either narrow their principle (so it doesn't apply to them) or expand their own life (so they're consistent with it). Neither move happens automatically.
Scene 2: The "I'm just asking questions" person
Someone in a workplace meeting says, every week, "I'm not saying X is wrong, I just want to understand — why do we assume X is true?" They never state a position. They never propose an alternative. They never commit. But by week three, the team is paralyzed, the original proposal is gutted, and no new proposal exists.
Read this against 「若不立自宗唯破他宗，名破法人」 — "if one does not establish one's own thesis and only refutes others', one is called a Dharma-destroyer." The classical critique is precise. You can do enormous damage to a real position (the team's plan) without ever exposing yourself to the discipline of stating your own. The damage is real; the responsibility is hidden.
But the classical material does not let you off the hook either. The 《菩提道次第廣論》 notes that the act of *not* stating a thesis is itself a thesis — namely, "I prefer to operate without commitments." Apply that to the meeting-room person: their implicit 自宗 is "I can productively participate in collective deliberation while remaining uncommitted." Examine that thesis under scrutiny — does it actually work? Does the team benefit from someone who will not state a position? If no, then the no-thesis posture is itself a poor thesis, hidden from view. The first lesson is to *surface* the implicit thesis. The second is to ask whether, once surfaced, it survives examination.
Where do most people map this wrong? They assume "I have no position" is automatically a more enlightened stance than "I have a position." In the classical analysis, "no position" is a *claim about reality* (reality does not warrant any commitment), and that claim has to be defended. The Madhyamaka, famously the most "no-position" school in Buddhist philosophy, turns out — once you read 《菩提道次第廣論》 carefully — to be the school that has the most carefully articulated *position about why and how it appears to have no position*. Even the deepest "no-thesis" stance has its own thesis: 離過之宗, 如幻.
Scene 3: The partner who only diagnoses
A couple in long-term therapy. One partner, let's call them J, has become the household's resident "observer." J can name every dynamic in the relationship: "you're being avoidant," "this is a control pattern," "you're projecting your father onto me." J reads the relationship fluently, in therapist-speak. But J rarely says *what J wants*. When pressed, J says, "I don't want to be in a position of wanting things; I just want us both to be free."
Apply the lens. J's explicit 自宗: "I want both of us to be free." J's implicit 自宗, shown by the *way J engages* (always diagnosing, rarely stating desire): "the highest form of relating is interpretive, not wanting." Apply this back to J. Does J actually *not want things*? No — J wants to be the interpreter; J wants not to be vulnerable; J wants the relationship to remain in a register J controls. The "I don't want to want" is a desire *not to have desires*, which is itself a desire. J's interpretive stance only damages J's own commitment to be a non-wanting partner (汝語唯壞汝自宗).
Here the classical move from 《菩提道次第廣論》 is exactly what is needed: the explicit/implicit distinction. J's thesis on desire is not what J *says* about desire; it is what J's *entire pattern of engagement* demonstrates about desire. J's actual 自宗 is "interpreting is safer than wanting." Once surfaced, J can examine it. Is it true? Is it good? Does it serve the relationship J claims to want?
The 離過之宗 ("faultless thesis") does not tell J to start issuing demands. It tells J that any thesis J holds, including the thesis "I hold no thesis about what I want," must be held *like an illusion* — present, examined, not crushed by self-nature, not gripped as ultimate. J can say, "I want X, and I know that wanting X is not the foundation of who I am." That is a 如幻 stance. It is more vulnerable than "I'm just an observer," and it is also more honest, and it has a self that can take responsibility for the wanting.
Scene 4: The activist whose means contradict their end
A climate activist publicly shames individuals for flying while refusing to discuss the much larger emissions produced by the activist's own organization — including the international conferences the activist flies to in order to give speeches shaming flyers. The activist's explicit 自宗: "flying is morally indefensible given the climate crisis." The activist's implicit 自宗, visible in action: "the moral importance of my message justifies emissions that would be inexcusable in others." Apply the rule. The activist's shaming refutation — "you are flying; you are therefore complicit" — applies symmetrically to the activist's own flights. The activist's words only damage the activist's own thesis.
The classical 《入中論善顯密意疏》 was not written for climate activism. But its logic is the same: a 似能破 is a refutation that survives only by exempting its speaker. If the speaker will not exempt themselves, the refutation is empty. If the speaker exempts themselves, the speaker's position becomes incoherent (universal rules with personal exemptions are not universal rules).
Where do people map this wrong? They assume this is hypocrisy-calling, and they retreat to "well, no one is perfect." That misses the classical point entirely. The point is not "you are a bad person." The point is "your refutation does not actually establish what you think it establishes; it has no logical force against your target." The fix is not guilt; the fix is to either broaden the principle (and live by it) or narrow the principle (so it no longer claims universality).
Scene 5: The AI assistant that "has no opinion"
You ask an AI chatbot, "Should I take this job offer?" It replies, "I'm just an AI; I don't have opinions. Here are some considerations to help you think through it." Now apply the lens. The assistant has stated a 自宗: "I, an AI, do not have opinions." That is a thesis. It is also factually questionable — the assistant's very *selection* of which considerations to list, in which order, with which framing, is a series of opinionated acts. The thesis "I have no opinions" is undermined by the thesis-laden character of the response. The response's claim to opinionlessness only damages its own claim to opinionlessness.
This is not a bug. It is the structural shape of any "no-thesis" stance: as the 《菩提道次第廣論》 notes, 「無許」 itself exhibits a kind of assertion. To say "I assert nothing" is to assert something. The Madhyamaka insight, translated into a contemporary register, is this: the attempt to have no thesis is itself a thesis; examine it. This is a hugely practical principle for the age of algorithmic mediation. Every system that mediates information is making a claim about the world by what it includes, excludes, ranks, and frames. Pretending it is not making a claim is the worst kind of unexamined thesis — the kind that cannot be argued with because it refuses to surface.
(Note the limit of the analogy: AI assistants do not have *intentional* theses; they are statistical systems. So the analogy illuminates the *shape* of the no-thesis problem, not the metaphysics of agency. Classical Buddhist analysis applies cleanly to intentional speakers; its extension to non-intentional systems is suggestive, not conclusive.)
三、当代人为什么需要它
The post-truth condition is a 自宗 crisis
When public discourse feels unmoored — when every claim meets the reply "well, that's just your opinion" — the problem is not that people have too many theses. The problem is that people have stopped *owning* their theses. They float claims without standing behind them; they demand refutation without offering anything that can be refuted; they perform skepticism as a substitute for thought. This is precisely the disease the 《入中論善顯密意疏》 diagnosed in the "Dharma-destroyer": 「若不立自宗唯破他宗」 — no own thesis, only attack on others.
The classical remedy is not "have a thesis at all costs." It is: *know what your thesis is, on every level where it operates — explicit and implicit* — and then examine whether you can stand behind it under your own arguments. If you can, you have a defensible 自宗, and you can enter real conversation. If you cannot, you have either a hidden thesis (examine it) or a confused one (clarify it) or a wrong one (let it go).
AI and the algorithmization of speech
Large language models, recommendation systems, search engines — all of these are systems that produce speech-shaped output without (currently) having the kind of intentionality that classical Buddhist analysis assumes. But their *users* still have 自宗. When a person shares a model output as "my view," they are taking ownership of a thesis that may have been generated by a system whose commitments they do not understand. The classic discipline — what does this claim actually commit me to? what happens if I apply my own arguments to it? — is exactly the discipline that the AI era makes newly urgent. Without it, people end up holding theses they did not choose and cannot defend.
The "all views are equally valid" trap
A culture that treats every view as equally valid is, ironically, a culture in which no view is treated as binding. This is the failure mode the 《菩提道次第廣論》 warns about: the slogan "no thesis" taken so far that no thesis can bind anyone to anything. The classical Madhyamaka answer is not "all views are equally valid" but "any view can be held, but if you hold it, you are responsible for it; hold it like an illusion, not like a self-existent fact." That is a much more demanding position than either dogmatism ("my view is absolutely true") or relativism ("no view is true"). It is also the only position that actually survives sustained self-examination.
Decision-making under uncertainty
Contemporary life is full of decisions where the evidence is incomplete, the stakes are high, and pretending to certainty is dishonest. The 如幻 stance is built for this. You form a thesis ("I will take this job"), you commit to it enough to act, but you do not pretend the thesis is self-grounded in the nature of things. You hold it as a working commitment that can be revised. You do not collapse into paralysis (the no-thesis trap) and you do not collapse into rigidity (the 自性 thesis). This is the離過之宗, the faultless thesis, applied to life.
Why the symmetry rule is the highest gift of this teaching
Of all the components of 自宗, the principle 「汝語唯壞汝自宗」 is the one with the largest leverage in everyday life. It costs nothing to apply; it does not require metaphysical commitments; it does not even require Buddhism. *Before you issue any refutation, ask: does my refutation, applied to me, damage my own position?* If yes, your refutation has no force. Either fix your position, fix your refutation, or drop the whole exchange. This is the rule that, internalized, would change most public discourse overnight.
四、常见误读与澄清
Misreading 1: "Madhyamaka has no thesis, so I should have no thesis either."
This is the dominant popular misreading. It treats 「無宗」 as a complete statement. As the 《入中論善顯密意疏》 makes clear, 「無宗」 is a *qualified* statement: no thesis *with inherent existence*. The qualifying phrase changes everything. To copy the slogan without the qualification is to take the worst possible lesson: that all commitments are equally unreal, that one need not stand behind anything. The classical Madhyamaka is the *opposite* of this. It is the school with the most disciplined commitment to examining what it actually asserts, in what register, with what scope. Its freedom is the freedom of someone who has examined their commitments thoroughly, not the freedom of someone who has none.
Misreading 2: "Having no position is more enlightened than having a position."
This is the spiritual-but-not-religious cousin of Misreading 1. The classical analysis exposes it: 「無許」 is itself an assertion. The very act of claiming "I have no thesis" performs a thesis. To be honest about this is to either drop the claim or to upgrade it into something defensible ("I have no thesis *of self-nature* about X, while affirming Y in the conventional register"). The latter is the Madhyamaka move. The former is a posture that cannot survive its own scrutiny.
Misreading 3: "Refutation = denial."
The technical vocabulary of 能破 (néngpò, "the refuting") and 所破 (suǒpò, "the refuted") gets flattened in popular reading into "destroy / destroy." This is wrong on the classical view. Madhyamaka refutation does not annihilate its target. It shows that the target does not exist in the way the opponent took it to exist. Cause and effect are not denied; they are re-described as lacking self-nature. The refuter does not establish *that* the refuted is gone; the refuter shows *that the way the refuted was held* was unsupported. This is a much gentler operation than denial, and it preserves the world.
Misreading 4: "If I hold a thesis like an illusion, I am being wishy-washy."
This treats 如幻 as half-commitment. It is the opposite. To hold a thesis *like an illusion* is to hold it fully — to act on it, to teach it, to defend it, to stake your life on it where required — *while* knowing that it is not self-grounded. Compare: a doctor who performs surgery with full commitment, knowing that the diagnosis is a working model, not a self-existent fact. The doctor is not "wishy-washy" about the surgery; the doctor is appropriately committed to a model that has limits. This is the 如幻 stance. It is the strongest available posture for someone who knows that all knowledge is provisional but who still has to act.
Misreading 5: "The symmetry rule is just calling people hypocrites."
The 「汝語唯壞汝自宗」 principle is often mistaken for a sniping tool. It is not. Its purpose is *not* to win arguments by exposing inconsistency in opponents. Its purpose is to *check your own refutations* before you issue them. Used as a self-discipline, it is one of the most powerful intellectual hygiene tools ever devised. Used as a weapon against others, it is just one more way of "only refuting the other and never establishing one's own" — exactly the disease the principle is designed to cure. As the 《菩提道次第廣論》 warns: if you only use the symmetry rule against others, you have not understood it.
Misreading 6: "自宗 = sect identity, so I should be sectarian."
In some uses, 自宗 refers to a school or sect's commitments (the material itself notes this with the example of the 犢子部等正量部 — the Vātsīputrīya and Saṃmitīya schools — who "take the self to be the inner aggregates"). This is a use of the term, but it is not the liberating use. The liberating use is the *first*-person discipline: what is *my* thesis, on this matter, in this conversation, that I am willing to take responsibility for? Confusing the liberating use with sectarian identity is a common way of mistaking group affiliation for personal integrity.
A closing observation
What is most striking about the classical material on 自宗, and what most rewards contemporary attention, is its quiet insistence on a hard double requirement. You must *have* a thesis — explicit, defended, visible. And you must *not* mistake that thesis for a self-existent fact. Both halves are required. Drop the first half (as in the "no-thesis" misreading) and you have nothing to examine, nothing to stand behind, nothing that can be wrong. Drop the second half (as in dogmatism) and you have a thesis so rigid that it cannot survive contact with its own implications. The Madhyamaka's actual 自宗 — the 離過之宗, held 如幻 — is the place where both halves hold. It is not a comfortable place. It is not a posture. It is a working practice, and the most useful gift the tradition has to offer to anyone who has ever found themselves arguing without knowing what they actually believe.