What this matrix compares
This matrix sets side by side the principal interpreters who have read Nāgārjuna’s MMK as collapsing into nihilism — or, in a narrower band of cases, into absolutism, positivism, deflationary pragmatism, or outright invalidity. It compares them along five axes: the verdict reached; the specific inferential move that yields that verdict; the framework status of the interpreter relative to the Mahāyāna hermeneutical apparatus (Two Truths as pedagogy, Provisional and Definitive, Three Turnings); the methodological mode of the engagement; and the primary text in which the position is set out.
The matrix is constructive and descriptive. Its adjudicative companion is the argument page framework-absence-yields-nihilism, which proposes that the convergent pattern shown here is not coincidental: framework-removal predictably produces nihilism in most cases, with absolutism and positivism as the secondary failure modes of partial removal under a Western overlay. This page presents the data; that page interprets it.
The principal source for the survey is westerhoff-nihilist-interpretation-2016, whose JIP article covers all three historical constituencies — non-Buddhist Indian critics, Buddhist critics, and modern Buddhologists. The matrix extends Westerhoff’s catalogue with two further bodies of material. The first is the Della Santina diagnosis (della-santina-madhyamaka-western-1986) of Kantian and Wittgensteinian overlay distortions (Murti, Streng, Gudmunsen). These produce non-nihilist failure modes (absolutism, positivism), yet they instantiate the same underlying pattern of partial-framework engagement. The second is kalupahana-mmk-1986, whose deflationary-pragmatic reading reaches a different verdict — not nihilism but Early-Buddhist empiricism — by the same framework-removal move. The third, added in August 2026, is hayes-nagarjunas-appeal-1994, which reaches a fifth verdict again — that the arguments are simply invalid — and which independently constructs three of the four distortion types catalogued here. Where this matrix and Westerhoff diverge on a given interpreter, I follow Westerhoff’s reading and note the divergence in the relevant cell.
The matrix
| Interpreter (period / school) | Verdict reached | Inferential move / reason | Framework status (re framework-absence-yields-nihilism) | Methodological mode | Primary text |
|---|---|---|---|---|---|
| Uddyotakara (6th–7th c., Nyāya) | Nihilism (sarvābhāvavāda) | Four epistemic-instrument arguments: (a) pramāṇas must exist to assert nihilism; (b) the sentence “all is non-existent” must itself exist with its referent; (c) speaker and listener must exist; (d) if “exists” and “does not exist” differ in meaning, some fact must ground the difference, which the nihilist also wants to deny | External — no access to framework; treats niḥsvabhāva = sarvābhāva | Polemical sūtra commentary | Nyāyavārttika on Nyāyasūtra 4.1.37 (per westerhoff-nihilist-interpretation-2016 pp. 340–342) |
| Kumārila (c. 7th c., Mīmāṃsā) | Nihilism via universal extension of emptiness | Mādhyamikas go beyond Yogācāra by denying both knowable objects (jñeya) and cognition (jñāna) — the verb “to know” grammatically requires an object, so if the representata are empty so must be the representations | External — no access; bāhyārtha-śūnyatva read straightforwardly as non-existence | Mīmāṃsā exegesis; grammatical-semantic argument | Mīmāṃsāślokavārttika 5.3.14–16 with Sucaritamiśra’s ṭīkā |
| Śaṅkara (8th c., Advaita Vedānta) | Nihilism via denial of both conventional and ultimate | Mādhyamikas deny lokavyavahāra on top of paramārtha — together = denial of everything. Even if there is no pot, the cognition of a pot is supported by every pramāṇa and cannot be denied; emptiness of everything would have to deny this too | External — no access; vyavahāra treated as a foundationalist commitment Madhyamaka would have to share to avoid collapse | Vedānta bhāṣya-exegesis | Brahmasūtra-bhāṣya 2.2.31; Bṛhadāraṇyaka-Upaniṣad-bhāṣya 4.3.7 |
| Rāmānuja (11th–12th c., Viśiṣṭādvaita) | Nihilism (tucchatva, “nothingness”) | (i) Ordinary talk of bhāva / abhāva concerns local states (a pot exists in the house, not the kitchen); universal non-existence is unintelligible in this grammar. (ii) The pramāṇa by which emptiness is established must itself be real — if not, “everything would be real” (recapitulating Uddyotakara) | External — no access | Vedānta bhāṣya-exegesis | Śrībhāṣya on Brahmasūtra 2.2.30 |
| Madhva (13th c., Dvaita Vedānta) | Nihilism | Causality from a non-existent (asat) thing is unobserved; asat lacks all viśeṣa like a sky-flower; Madhyamaka’s claim that empty causes produce effects contradicts everyday experience | External — no access; reads śūnya as asat | Vedānta bhāṣya-exegesis | Brahmasūtra-bhāṣya 2.2.26 with Jayatīrtha’s Nyāyasudhā |
| Mādhava (14th c., doxographer) | Nihilism (sarvasya asattvam) | (i) Madhyamaka reasoning is fully general — applies indifferently to superimposed object, basis, relation, seer, seen; (ii) since these pairs are mutually existentially dependent, establishing non-existence of one entails non-existence of all → universal emptiness as universal non-existence | External — doxographic survey | Doxography | Sarvadarśanasaṅgraha §“Mādhyamika-darśana” |
| Asaṅga (4th c., Yogācāra) | pradhāna nāstika (“the most extreme kind of nihilist”) | “All is mere designation” (prajñaptimātram eva sarvam) requires a non-designated vastumātra basis on which the designation is made. Without basis (āśraya), neither tattva nor prajñapti is reasonable. This is durgṛhīta-śūnyatā (“wrongly grasped emptiness”). Direct ancestor of the modern Williams–Burton argument | Pre-Prāsaṅgika — Yogācāra-internal framework; pre-Candrakīrti’s hermeneutical apparatus | Yogācāra exegesis | Bodhisattvabhūmi, Tattvārtha chapter |
| Vasubandhu (4th–5th c., Yogācāra → Sautrāntika) | sarvanāstitā (universal non-existence) | Mādhyamikas “deny the existence of all the dharmas” and therefore prevent liberation. Reasoning not spelled out but inherited from Asaṅga (sole reference to Madhyamaka in entire corpus; Yaśomitra’s Sphuṭārthā-vyākhyā confirms the identification) | Pre-Prāsaṅgika; shares Asaṅga’s no-foundation worry | Abhidharma exegesis | Abhidharmakośa-bhāṣya ad 5.19 |
| de la Vallée Poussin (1908–1926) | Dharmas “do not exist at all, either in reality or apparently” | Like the beauty of the daughter of a barren woman: “the object described, the description and the person describing are all similarly nonexistent.” Paramārthasatya = unqualified negation of appearances. (Later reversed his view, per de Jong 1972.) | Full methodological removal (initially); the high-water mark of early Western Buddhology’s consensus reading (Burnouf 1844, Kern 1896, Walleser 1911, Keith 1923, Wach 1924, Stcherbatsky 1927 all converge here) | Encyclopaedia / historical-philological survey | ”Nihilism (Buddhist)” in Hastings, Encyclopedia of Religion and Ethics |
| Murti (1955) | Absolutism (not nihilism) — distinct distortion | Reads Two Truths through Kant’s noumenal / phenomenal split: paramārtha as a transcendent absolute analogous to the Kantian noumenon. The pedagogical framework is retained in name but its content is replaced by Kantian ontology | Partial retention + Western (Kantian) ontological overlay; the Della Santina diagnosis of della-santina-madhyamaka-western-1986 | Cross-tradition philosophical comparison | Central Philosophy of Buddhism |
| Streng (1967) | Pragmatic positivism (not nihilism) — distinct distortion | Reads Two Truths through later-Wittgensteinian ordinary-language theory: emptiness as a therapeutic dissolution of metaphysics; “leave everything as it is” | Partial retention + Western (Wittgensteinian) semantic overlay | Comparative philosophy | Emptiness: A Study in Religious Meaning |
| Gudmunsen (1977) | Positivism (ordinary language as paradigm of truth) | Same Wittgensteinian overlay as Streng, pressed further: conventional truth elevated to the paradigm of truth simpliciter | Partial retention + Wittgensteinian overlay | Comparative philosophy | Wittgenstein and Buddhism |
| Kalupahana (1976 / 1986) | Deflationary pragmatic empiricism (not nihilism) | Reads MMK as a commentary on the Kaccāyanagotta Sutta, returning Nāgārjuna to “Early Buddhism.” Emptiness becomes “absence of metaphysics, not special insight”; the Two Truths become flat and non-hierarchical | Methodological removal of the Mahāyāna framework via return to Early Buddhism; self-undermining (apophatic about MMK while hyperkataphatic about “what Nāgārjuna really meant”) | Sri Lankan academic, Theravāda-leaning | kalupahana-mmk-1986 (Nāgārjuna: The Philosophy of the Middle Way) |
| Wood (1994) | Nihilism: “sheer, unqualified, absolute nothingness” (sarvam sarvena nāsti) | Consistent prasajya reading of the Catuṣkoṭi: relevant-logic (Routley–Meyer), speech-act (Matilal), Buridan–Prior, and Ruegg’s FPC readings all dismissed; the only consistent reading terminates in absolute nothingness. Even appearances (vijñapti) do not exist | Argumentative rejection of the framework, not methodological removal: engages Two Truths and rejects them as “a literal or philosophical distinction”; engages and rejects Jayatilleke’s pragmatic avyākata reading as a category mistake. Cleanest exhibit of framework rejection on its own terms | Formal logic + Indology | wood-nagarjunian-disputations-1994 (Chs II §–4; IV n. 11; V §, , ) |
| Oetke (1991) | “On the level of the highest truth there is nothing of any kind” | Treats the paramārtha-level by analogy with the theoretical level in physical science (lightning “really is” an electric discharge). The MMK 24:18 / upādāya prajñapti derivation runs the same regress-from-universal-prajñaptimātra inference as Burton, in a different idiom | Framework-as-formalism without framework-as-pedagogy — retains the Two Truths as a sentential-operator scaffolding but empties it of pedagogical content (no neyārtha / nītārtha, no Three Turnings, no soteriological function). Distinctive sub-pattern | Analytic philosophy + German Indology | oetke-remarks-interpretation-1991 |
| Williams (1998 ff.) | Partial nihilism: pain has svabhāva; universal niḥsvabhāva judged “incoherent … equivalent to metaphysical nihilism” | Williams–Burton regress: all things being prajñaptisat would mean conceptual constructs without an unconstructed basis or unconstructed constructor → ontological nihilism. Williams holds with Abhidharma that pains have svabhāva (“found under analysis”; “it hurts; its existence is its hurting”) | Partial removal — accepts Two Truths as a level-distinction and engages the Indo-Tibetan commentators by name, but defects on the universal-emptiness extension. Localised collapse on the specific item (pain), per siderits-reality-altruism-2000 fn. 10 | Analytic philosophy + Buddhological scholarship | ”Altruism and Reality” (1998); Buddhist Thought (2000) |
| Burton (1999, 2001) | Full ontological nihilism | Williams–Burton regress, full strength: “conceptually constructed things need an unconstructed basis out of which they are constructed [and] an agent of the construction which is not himself a conceptual construction.” Same point Asaṅga makes in the Bodhisattvabhūmi | Full methodological removal — explicitly disengages from commentarial tradition (“careful not to import what are actually later Mādhyamika concepts”); acknowledges the dGe lugs pa solution as “the most ingenious attempted solution” but excludes it by methodological choice (Ch. 4 n. 23) | Analytic philosophy | burton-emptiness-appraised-1999; “Is Madhyamaka Buddhism Really the Middle Way?” (2001) |
| Hayes (1994) | The arguments are invalid (not nihilism) — a fifth, distinct verdict. Nāgārjuna is “a metaphysician par excellence” and a rationalist whose conclusions are unsupported | The informal fallacy of equivocation: svabhāva₁ (identity) slides into svabhāva₂ (causal independence), parabhāva₁ (difference) into parabhāva₂ (dependence), bhāva(P) into bhāva(L), and “exists” into “exists as”. Each verse of MMK 1:5 and MMK 15:1–5 is true on one reading, but the conclusions do not follow from the premises. Nāgārjuna’s logic is standard, not deviant: he states non-contradiction and the excluded middle and depends on deriving contradictions | Full methodological removal, in its purest form: not a rejection of the framework but its total absence from the field of view. No Two Truths anywhere in eighty pages, no saṃvṛti / paramārtha, no neyārtha / nītārtha, no Three Turnings, no Tibetan material; MMK 24:8–10 never treated; Candrakīrti dismissed as asserting strongly in the absence of an argument. He nonetheless rejects three of this matrix’s four other verdict types by name (Absolutist, Positivist, Deconstructionist) as anachronistic projection | Sanskrit philology (Pāṇini, the Kāśikā-vṛtti) + analytic logic | hayes-nagarjunas-appeal-1994 (§–4) |
| Westerhoff (2009, 2016) | Sophisticated nihilism defended as compatible with Madhyamaka — emptiness as conventional corrective, truth as warranted assertibility | Constructs a consistent nihilism from eliminativism about the dependent + non-foundationalism; argues this is distinct from the five forms of nihilism Madhyamaka explicitly rejects. The “equilibrium principle”: opponent-relative interpretation, with the nihilist reading sharpest against today’s naturalistic realists | Framework-engaged — extensive engagement with Candrakīrti’s Prasannapadā, Asaṅga’s Bodhisattvabhūmi, Yaśomitra. Nihilism need not be incoherent, and framework-engaged interpretation can itself arrive at nihilist-adjacent conclusions | Analytic + textual; opponent-relative | westerhoff-madhyamaka-2009; westerhoff-nihilist-interpretation-2016 |
What the matrix shows
Three constituencies, one structural pattern
The non-Buddhist Indian critics, from Uddyotakara through to Mādhava, all share what Tsongkhapa six centuries later would diagnose as the svabhāva-equation: existence = svabhāva-existence, and therefore niḥsvabhāva = non-existence. None of them had access to the Mahāyāna hermeneutical framework that breaks this equation by making dependent origination the positive content of emptiness (jinpa-tsongkhapa-qualms-1999 is the locus classicus of the diagnosis; the full argument is set out in framework-absence-yields-nihilism). Their convergence on the nihilism verdict is, on this reading, not coincidental but structural.
The Buddhist critics (Asaṅga, Vasubandhu) operate before — or outside — the Prāsaṅgika apparatus that Candrakīrti would later develop. Asaṅga’s no-basis-for-prajñapti worry is the direct ancestor of the modern Williams–Burton argument, and Westerhoff makes this lineage explicit (westerhoff-nihilist-interpretation-2016 p. 354). The internal Buddhist nihilism charge has the same structural shape as the modern academic charge — confirming that the type of move drives the conclusion, not which side of the Buddhist boundary one stands on.
The modern Buddhologists exhibit the same pattern in three distinguishable sub-modes:
- Full methodological removal (de la Vallée Poussin and the early Western consensus; Burton; Hayes; in a different way, Kalupahana) — bracket the framework entirely. Verdicts: nihilism, deflationary pragmatism, or invalidity.
- Framework-as-formalism without pedagogy (Oetke) — retain the Two Truths as a formal sentential-operator scaffolding but empty it of soteriological content. Verdict: nihilism on the analogy with theoretical-level physics.
- Partial removal with overlay (Murti / Kantian, Streng / Gudmunsen / Wittgensteinian) — retain the framework’s name but replace its content with a Western ontology or semantics. Verdicts: absolutism (Kantian overlay) or positivism (Wittgensteinian overlay).
- Argumentative rejection (Wood) — engage the framework on its own terms and reject it: the Two Truths are not “a literal or philosophical distinction”; pragmatic catuṣkoṭi readings are category mistakes. Verdict: nihilism via a consistent prasajya reading of the four extremes. This is the cleanest exhibit because the rejection is philosophical, not methodological.
Williams as the structural hinge
Williams occupies a distinctive position. He accepts the Two Truths as a level-distinction and engages the Indo-Tibetan commentators by name, yet he defects on the universal-emptiness extension by holding (with Abhidharma) that pains have svabhāva. His localised collapse on a single item — pain — illustrates a general point: framework-as-formal-scaffolding without framework-as-ontological-content yields a localised, not a global, collapse. He is the intermediate case between full retention (the commentarial tradition) and full removal (Burton, his doctoral student, who pressed the move to its limit).
Hayes as the fifth verdict — and as an independent builder of this taxonomy
Hayes was added to the matrix on 2026-08-07 and is the only row whose verdict is none of the four this page was built to catalogue. He is explicitly against three of them: the Absolutist reading (Stcherbatsky, Schayer, Murti), the Positivist reading (Warder, Kalupahana), and a Deconstructionist reading (Magliola, Dilworth, Huntington) that this matrix does not otherwise carry. What he concludes instead is that the arguments equivocate and their conclusions are unsupported — the text fails on its own logic. Two consequences.
The typology is less tidy and the scatter claim is stronger. A fourfold catalogue of failure modes was always a claim about what framework-removal produces; a fifth destination is more scatter, not less, and needs scatter rather than tidiness. What it does not touch is the target, which was always the twelve unqualified nihilism₁ rows.
His is this matrix, built by a hostile reader. Hayes independently identifies three of the four distortion types, names their exhibits, and diagnoses all three by the same mechanism this page uses — reading one’s own era into the text, via Tuck’s isogesis. The diagnosis here otherwise rests on della-santina-madhyamaka-western-1986, who is framework-sympathetic and can be discounted as an apologist. Hayes cannot: he has no interest in defending the tradition and every interest in finding the text defective. The corroboration is worth more than the extra row.
There is a reflexive complication, and it should be recorded rather than suppressed. Hayes’s charge against the five interpreters he criticises is that each finds his own century’s preoccupation in the text; his own verdict measures a second-century Indian monastic against twentieth-century formal validity, on the assumption that the MMK is a set of arguments meant to compel a logically astute reader. That assumption is a genre claim, and it is the genre claim disputes. He half-sees this — he distinguishes exegesis from hermeneutics at the close, and quotes Robinson to the effect that Nāgārjuna should be appraised in his own milieu — but does not turn it on his own . Noting this does not answer the equivocation charge, which stands on its merits until answered on them (see Svabhāva for the two incompatible answers now on file, from Siderits and Gokhale).
Westerhoff as the limit case
Westerhoff is the framework-engaged interpreter who nonetheless arrives at a nihilist-compatible conclusion. He partially concedes the critics’ diagnosis while reframing nihilism as defensible: a sophisticated nihilism, distinct from the five forms Madhyamaka explicitly rejects. He shows that the framework alone does not settle the question. What it does is supply the resources — dependence-circles, infinite descent, the Catuṣkoṭi’s pedagogical role, neyārtha / nītārtha — that prevent the naive nihilist inference Burton presses. Whether a sophisticated nihilism that uses those very resources differently is itself the right reading is a separate question. His equilibrium principle pushes that question towards the opponent: against naturalistic realists, the nihilist reading sharpens; against anything-goes relativists, the stability of conventional truth does.
The convergence at a glance
- Verdict = nihilism: 14 of 19 — 6 non-Buddhist Indian + 2 Buddhist + 6 modern (de la Vallée Poussin, whose row also represents the Burnouf / Kern / Walleser / Keith cluster and has no separate row for it; Wood; Oetke; Williams; Burton; Westerhoff). The count of nihilism verdicts is unchanged; the denominator rose from 18 to 19 when Hayes was added on 2026-08-07. An earlier “15 of 18” figure double-counted the cluster.
- Of those 14, unqualified metaphysical nihilism (nihilism₁): 12. Williams’s charge is partial (pain retains svabhāva) and Westerhoff’s is the conceded methodological nihilism (nihilism₂), which does not undertake to refute. Twelve rows are the direct target of the refutation, and that figure is unaffected by the change of denominator.
- Verdict = absolutism: 1 (Murti).
- Verdict = positivism: 2 (Streng, Gudmunsen).
- Verdict = deflationary pragmatism: 1 (Kalupahana).
- Verdict = the arguments are invalid: 1 (Hayes).
Nihilism 14 + absolutism 1 + positivism 2 + deflationary pragmatism 1 + invalidity 1 = 19.
By framework status:
- Framework status = no access: 8 (the 6 non-Buddhist Indians + 2 pre-Prāsaṅgika Buddhists).
- Framework status = full methodological removal: 4 (de la Vallée Poussin et al., Kalupahana, Burton, Hayes; Wood is argumentative rejection rather than methodological removal).
- Framework status = framework-as-formalism without pedagogy: 1 (Oetke).
- Framework status = partial retention with Western overlay: 3 (Murti, Streng, Gudmunsen).
- Framework status = partial removal with localised collapse: 1 (Williams).
- Framework status = argumentative rejection from within: 1 (Wood).
- Framework status = framework-engaged: 1 (Westerhoff).
8 + 4 + 1 + 3 + 1 + 1 + 1 = 19.
The 14-of-19 nihilism count is the very convergence Westerhoff himself invokes (p. 358) against the dismissal that “two millennia of misreaders simply failed to understand.” The companion page framework-absence-yields-nihilism explains that same convergence differently: not by conceding nihilism, but by identifying the methodological move that produces it.
Linked pages
- Arguments: framework-absence-yields-nihilism (the adjudicative companion); framework-necessity (the positive version of the thesis); zhentong-is-refined-cittamatra-not-madhyamaka (the off-axis case, where framework-removal yields a positive substantialism rather than nihilism)
- Concepts: Emptiness, Svabhāva, Two Truths, Provisional and Definitive, Three Turnings, Pratītyasamutpāda, Catuṣkoṭi
- Sources: all interpreters with named source pages are linked in the matrix’s “Primary text” column; the central survey source is westerhoff-nihilist-interpretation-2016