Madhyamaka Wiki

Catuṣkoṭi Must Negate All Four Extremes

15 min read

The claim

Gorampa’s central methodological commitment is that the Madhyamaka negation is constitutively fourfold. To count as Madhyamaka at all, the negation must apply to all four koṭis of the catuṣkoṭi — existence, non-existence, both, neither — without qualification. Two characteristic Tibetan errors fail this test in opposite directions:

  • Tsongkhapa narrows the negation to bden grub alone. This leaves the existence-koṭi (in its non-bden grub form), the non-existence-koṭi, the both-koṭi, and the neither-koṭi all untouched. On Gorampa’s reading () such a narrowing “renders three of the four koṭis pointless” and reduces the Madhyamaka critique to a “scholastic epiphenomenon.”
  • Dolpopa exempts the pariniṣpanna from the negative dialectic. The existence-koṭi survives intact at the level of the ultimate, leaving the negation incomplete in the opposite direction.

Only a fourfold negation that withholds nothing — neither a refined med dgag of bden grub (against Tsongkhapa) nor a truly existent gzhan stong basis (against Dolpopa) — qualifies as the mtha’ bral dbu ma the Indian sources teach.

The logical core: why a separately-negated second koṭi defeats the bivalent reading

The fourfold-negation requirement is not only a doctrinal norm. It carries a precise logical payload, and that payload is what defeats the nihilist (Wood above all). The nihilist reads the prasajya negation of the first koṭi — “not existent” — as delivering the second koṭi, “non-existent,” and comes to rest there. That inference is valid only if “exists” and “does not exist” jointly exhaust the options, so that denying the one affirms the other. The catuṣkoṭi’s separate, explicit negation of the second koṭi is precisely a denial of that exhaustiveness. In refusing “does not exist” in its own right (MMK 22:11; 15:10), Nāgārjuna asserts that the first two koṭis are not a bivalent partition of the field. The negation therefore cannot terminate in the very koṭi it independently removes. “Stopped one koṭi short” is not a rhetorical flourish but a structural fact about a four-valued, non-exhaustive scheme. This is the formal underpinning of the Wood refutation at nihilism-charge-refuted (Step 2): Wood’s “consistent” reading covertly re-collapses the four koṭis into a bivalent two, smuggling back the exhaustiveness the structure was built to deny.

The move is native to the literature, not reverse-engineered from the difficulty (astasahasrika-84000-2026 1.8–1.12, with n. 97). A standing worry about the non-exhaustiveness argument is that it looks built to order — a device produced by later readers to get Nāgārjuna out of a bind. The Aṣṭasāhasrikā Prajñāpāramitā shows the same move in the earliest Prajñāpāramitā stratum, and shows it dramatised rather than argued. Subhūti says the mind of awakening “is free of a mindset”; Śāriputra puts the ontological question, “does that mindset exist?”; and Subhūti declines to answer it. He asks instead whether, in that state, an existence or a non-existence is found or perceived, receives “definitely not”, and then asks whether the question was well formed as put. Only when Śāriputra reframes does he answer at all. Ratnākaraśānti’s gloss names the structure exactly: neither is perceived “because there is no proof of them” nor found “because there is disproof of them”, so “their negation perfectly transcends both existence (bhāva) and non-existence” — and Śāriputra’s question, he says, amounts to asking “does that non-existence (nāstitā) exist?”

Three limits on what this may be used for. It is framework-internal, so it cannot be added to the framework-external blocks below as a fourth route; the independence problem recorded under the objections is untouched by it. Ratnākaraśānti is a Yogācārin reading a sūtra as a Yogācārin, so the gloss is one commentator’s and not “the Indian commentators’” — though it is the more striking for being reached by a reader whose own construal of acitta is paryudāsa. And the note is quoted from the Tibetan of the Sārottamā rather than from Sanskrit. What the passage does establish is provenance: the refusal to let the negation terminate in the second koṭi is how this literature behaves at its origin, not a rescue applied to it afterwards.

Reconciliation with Tsongkhapa’s “prasajya presupposes the excluded middle.” Tsongkhapa holds that a non-affirming negation (མེད་དགག་, prasajyapratiṣedha) presupposes the law of excluded middle and can close an argument without reifying anything (Non-affirming Negation, on VV 26). This can look as though it concedes Wood’s bivalence. It does not. The excluded middle (every proposition is true or false) is left intact. What the fourfold negation denies is a further, material premise: that {exists, does-not-exist} is an exhaustive pair of worldly options, such that ¬exists entails non-existence-as-a-fact. One may hold the excluded middle over propositions and still deny that “x exists” and “x does not exist” exhaust the ways x might be characterised, which is precisely the four-koṭi structure. Tsongkhapa’s formal point and the anti-Wood exhaustiveness point are therefore compatible: both insist the negation is non-affirming, and the four-extremes requirement adds that the field it ranges over is not bivalent.

A parallel block from outside the tradition. gokhale-ways-of-argumentation-2008 stops Wood at the same step by a different route: at the limiting cases, where the tetralemmas have been iterated until nothing stands behind existence and non-existence, the denial is non-truth-functional (a performative negation after Ruegg 1986) and so falls outside the excluded middle and double negation. Wood needs the terminal negation to state something; on that reading it states nothing. Whether this is genuinely a second route or the same move under another description is unsettled — a reader who holds that a non-bivalent field is what makes the terminal utterance non-propositional would collapse the two.

A third block, and the one that concedes Wood the most ground (Galloway). galloway-logical-issues-1989 grants the excluded middle outright and denies that anything is left for it to range over. The Madhyamaka prasajya is non-implicative to the point of yielding a zero statement — applied to A(x) it licenses neither A(x) nor ¬A(x) — so each corner of a negated catuṣkoṭi is zeroed and their conjunction is zeroed too, with no contradiction and no residue (pp. 6–9, 18–19). At MMK 15.4 the argument is explicit: a bhāva would have to lie in the set of svabhāva or the set of parabhāva, both have been emptied at 15.1–3, and what is discarded is the bhāva, not the law. “Those who believe that Nāgārjuna rejects the law should show us where he so states” (p. 20), with Buddhapālita and Candrakīrti quoted in support. This is the most direct of the three routes because it engages Wood’s own premise rather than working around it: Wood needs the excluded middle to force “non-existent” out of “not existent”, and Galloway hands him the law while removing the subject — Ruegg’s semantic-well-formedness point (1977, p. 50) with the set theory made explicit.

Two cautions on the third route. Galloway quotes Candrakīrti’s bhāvo ‘pi nāstīty avadhāryatām (“it must be affirmed that a bhāva also is not”) approvingly at the same place, which reads as a statement with content and sits awkwardly beside the zero-statement doctrine; cite the structure of the move rather than that sentence. And the three routes are not certainly independent of one another — the possible collapse of the Gokhale route into the non-exhaustiveness route is recorded above, and Galloway’s subject-failure move may be what makes the field non-bivalent rather than a separate consideration. Independence should be established before presents them as three.

A related formulation for the third-position guardrail. Galloway’s n. 18 (p. 31) states the point the July peer review’s finding M5 wanted: ontologically the middle way “is merely the absence of error on one side together with its absence on the other; a twofold absence of deviation is not a third deviation.” He attaches Samādhirāja 9.27 — having avoided both extremes, “the wise takes up no middle position” — and the Kaccāyana material of MMK 15:7 and 15:10. The formulation answers the “then what occupies the space outside the four corners?” objection structurally, by denying that the question has an occupant to ask after, rather than by naming one.

Evidence for

  • **Primary support, gorampa-distinguishing-views-1469 ** (against Tsongkhapa): narrowing to bden grub alone “renders three of the four koṭis pointless.” The chapter unpacks this with extended primary citations from MMK 22:11 (the koṭis of empty / non-empty / both / neither should not be predicated), MMK 13:8 (in Cabezón and Dargyay’s translation, “Those who have a view of emptiness / Will accomplish nothing”), and the Saṃcayagāthā and Aṣṭasāhasrikā on signlessness.
  • Primary support, gorampa-distinguishing-views-1469 .2 (Gorampa’s positive system): the scriptural sources for the freedom-from-the-four-extremes (mtha’ bzhi’i spros bral) are catalogued as the trustworthy textual evidence for the position — including the Aṣṭasāhasrikā, the Vajracchedikā, the Lalitavistara, and the Saṃcaya. The four-extremes-negation is presented not as Gorampa’s innovation but as the standard scriptural presentation of the Madhyamaka view.
  • gorampa-distinguishing-views-1469 .1.2.2 (the real ultimate): the four extremes are eliminated simultaneously (cig char du) in āryan equipoise. The passage articulates the difference between stepwise negation (analytic, quasi-ultimate) and simultaneous negation (contemplative, real ultimate). The fourfold-negation requirement applies at both levels but is fully realised only at the second.
  • Convergent dissolution from other Tibetan voices. karmapa-feast-fortunate-1578’s “partial emptiness” (nyi tshe ba’i stong pa nyid) critique of Tsongkhapa makes the structurally same point in different language: refuting “true existence” instead of phenomena themselves leaves the existence-koṭi standing. On Mipham’s reading, the actual ultimate of shantarakshita-madhyamakalankara is reached only when the conceptually-cognised emptiness is left behind. This matches Gorampa’s requirement that the conceptual apprehension of emptiness be itself negated, completing the four-extremes negation. atisha-key-instructions and Atiśa’s broader corpus (apple-jewels-middle-way-2018) compress the same requirement into the contemplative register.
  • Nyingma primary witness on a Mahāyāna path-treatise: BCA 9.34 via kunzang-pelden-nectar-manjushri-2007. Mipham’s Norbu Ketaka reading, transmitted by Khenpo Kunzang Pelden, states the fourfold negation directly on Śāntideva’s verse: “when neither the thing (to be negated) nor the nonexistence of the thing… are present to the mind, no alternatives for true existence remain (in terms of being both existent and nonexistent or neither existent nor nonexistent). Consequently, the mind has no other object to fix on, no ideas like ‘It is empty’ or ‘It is not empty.‘” All four koṭis fall, and the conceptual apprehension of emptiness with them — the same requirement Gorampa derives from MMK 22:11. Confirms the cross-school convergence (Sakya / Kagyü / Atiśa / Nyingma) from a fourth text-base (the BCA rather than the MA/MMK).
  • The scriptural warrant Gorampa himself names is now in the wiki. Both passages above cite the Aṣṭasāhasrikā Prajñāpāramitā — at on signlessness alongside the Saṃcayagāthā, and at .2 among the trustworthy sources for freedom from the four extremes. The sūtra is at astasahasrika-84000-2026, and 1.8–1.12 shows why he could reach for it: the refusal to answer the existence question is enacted there rather than argued. Gorampa’s fourfold-negation requirement is on this point a report of what his scriptural sources do.
  • Indian textual root. MMK 22:11 is the principal Indian textual locus; the Vigrahavyāvartanī’s no-thesis defence is the structural rationale (a Madhyamaka thesis would single out one koṭi and so violate the requirement). The fourfold-negation is therefore not Tibetan but Indian-attested.
  • Pāli-canonical anticipation: Vacchagotta and the avyākata. The source is the Padmakara Translation Group’s Introduction to mipham-introduction-middle-way-2002, read in tenpa-personal-notes-2025 under “From Padmakara’s Madhyamakāvatāra introduction.” The fourteen unanswered questions Vacchagotta poses to the Buddha are themselves posed in tetralemma form: whether the universe has a beginning, or not, or both, or neither; whether it has an end, or not, or both, or neither; whether the Buddha exists after death, or not, or both, or neither; whether the self is identical with the body or different from it. The Buddha’s silence is the response Nāgārjuna later articulates as the negation of all four extremes. “The Madhyamaka can be understood as the exploration and systematic expression of the Buddha’s silence.” This pushes the catuṣkoṭi’s structural root back from MMK 22:11 to the Pāli avyākata corpus itself. It supplies a śrāvaka-canonical witness that the fourfold-negation requirement is neither Tibetan-scholastic nor Mahāyāna-only, but is implicit in the Buddha’s own pedagogical practice — a register that framework-removed readings (Kalupahana especially) cannot dismiss as later imposition.

Evidence against / objections

  • Tsongkhapa’s pre-emption (tsongkhapa-illuminating-intent-1418): the six synonyms for the object of negation (true existence, ultimate existence, absolute existence, existence by intrinsic nature, existence by intrinsic characteristic, intrinsic existence) span the relevant range; the Prāsaṅgika negates all six. So the Geluk position does not narrow the negation to “true existence” only; the polemical summary by Gorampa misses the breadth of the Geluk bden grub category.

  • Tsongkhapa’s MMK-anchored four-koṭi negation (tsongkhapa-ocean-of-reasoning-1408 Ch 15 on MMK 15:10–11). Ocean Ch 15 negates all four koṭis explicitly: existence (“to say ‘it exists’ is to reify,” under the essentially qualifier); non-existence (“to say ‘it does not exist’ is to adopt the view of nihilism,” under the essentially qualifier); the both-koṭi (Yogācāra’s three-natures position, which “reifies the other-dependent” while “deprecating the conventional existence of subject and object,” falling into both extremes at once); and the neither-koṭi (implicit in the rejection of any position carrying an essential mode of existence). The Tsongkhapa-side framing is that all four koṭis are negated under the essentially-qualifier and none under the merely-qualifier. Gorampa’s polemical summary (“Tsongkhapa narrows to the first koṭi only”) is therefore not quite right: Tsongkhapa negates all four, but at the substantive mode-of-existence level. The disagreement is over what is being negated when all four are negated (essential mode of existence versus phenomena themselves), not over whether all four are negated.

  • Pragmatic worry. A constitutively fourfold negation risks contemplative quietism: if every cognitive content is negated at all four koṭis, the result may be indistinguishable from Hva-shang’s no-thought meditation. Tsongkhapa’s diagnosis at Lhag mthong is precisely this worry. The four-extremes requirement may be philosophically clean but soteriologically dangerous.

  • The quasi-ultimate concedes the point. Gorampa himself acknowledges (.1.2.1) that conceptual analysis cannot eliminate the four extremes simultaneously, only stepwise. So the strict fourfold-negation requirement applies only at the level of āryan equipoise, where it is by stipulation contentless. At the analytic level where Tsongkhapa actually operates, the requirement is impossible to satisfy. The argument as stated may therefore be evaluating Tsongkhapa by a standard he never claimed to meet at his operating register.

  • The framework-external routes may not be three. The page now carries three blocks on Wood’s inference — non-exhaustiveness (in-tradition), non-propositionality (Gokhale), and subject-failure (Galloway) — and their mutual independence is asserted rather than shown. If the terminal utterance is non-propositional because the subject has failed, and the field is non-bivalent because of the same fact, then would be citing one argument three times and dressing a single move as a convergence. Gokhale presses a version of this from the other direction, holding that Galloway’s defence collapses once the tetralemmas are iterated to existence-and-non-existence (see galloway-logical-issues-1989 note 4, where the objection is assessed and not resolved). Until this is settled the three should be presented as one argument with three formulations, which is still useful and is a weaker claim than this wiki currently makes.

    This has since got worse rather than better. gokhale-essentialism-eternalism-1996 shows Gokhale himself using the shared-presupposition mechanism — Nāgārjuna denies a statement and its internal negation together by denying what they presuppose, on Russell’s internal / external negation distinction, with MMK 1.3 as the exhibit — and finding its limit in exactly the place he later presses against Galloway: a denial of language as a whole cannot be negated twice. So his 1996 route and Galloway’s are the same mechanism, and his 2008 route is what he reaches when that mechanism runs out. Three formulations of one move is now the likelier reading, and should not claim a convergence until someone shows otherwise.

  • Selection effect on convergence. The “Karmapa, Mipham, Atiśa, Gorampa converge” datum is selected for. All four are critics of Tsongkhapa or pre-date him, and all four operate in non-Geluk traditions. The convergence is weaker evidence than it looks if the underlying anti-Geluk sympathy is shared.

  • A framework-aligned modern who reads the tetralemma bivalently: Garfield. This is an honest tension, not a supporting witness. In garfield-fundamental-wisdom-1995 Garfield, despite being Prāsaṅgika-Geluk-aligned and explicitly anti-nihilist, glosses the fourth koṭi of MMK 18:8 (“neither real nor not real”) as “the law of the excluded middle,” and holds that the prasajya negation and the tetralemma honour both non-contradiction and the excluded middle. He distributes the apparent contradictions across the two truths (conventionally real, ultimately unreal) rather than across a non-bivalent four-valued scheme, and he does not engage Wood. So a reader can be framework-internal and anti-nihilist without adopting the non-exhaustive, separately-negated-second-koṭi reading this page’s logical core relies on. The reconciliation is available: Garfield’s “excluded middle” is the propositional excluded middle (every proposition is true or false), which is compatible with denying that {exists, does-not-exist} materially exhausts the ways a thing may be characterised (see the Reconciliation note above). But Garfield does not draw that distinction, and his text reads the four corners more bivalently than the Gorampa line requires. His tetralemma gloss should therefore not be conflated with the second-koṭi reading; his independent contribution lies in the emptiness-of-emptiness / self-consuming-negation line, which does not depend on the four-valued logic. The tension also shows that the non-bivalent reading is not a modern consensus even among framework-friendly readers — a point worth conceding rather than burying.

Linked pages

Graph View

Backlinks