Definition
The Two Truths (satyadvaya) doctrine distinguishes conventional truth (saṃvṛtisatya) from ultimate truth (paramārthasatya). In Madhyamaka, conventional truth is the world of appearances, language, and everyday transactions. Ultimate truth is emptiness (śūnyatā), the absence of intrinsic nature (རང་བཞིན་) in all phenomena. The two are not separate domains but two ways of understanding one reality. As Nāgārjuna states in MMK 24:8-10, those who miss the distinction between the two truths miss the profound truth in the Buddha’s teaching.
Comparison matrix
| Thinker | Position | Key text | Wiki link |
|---|---|---|---|
| Tsongkhapa | ”Each phenomenon has two natures — an ultimate and a conventional — found by two distinct cognitive processes; not ‘a single nature that is in fact two truths in virtue of the two perspectives’” (Ocean Ch 24 p. 406). At MA 6.23 the same point is put as “two natures, identical in nature with distinct conceptual identities” (like produced / impermanent). The two are not in tension: the 1407–08 Ocean and the 1418 Illuminating the Intent formulations are doctrinally continuous, with Ocean the sharper anti-subject-side statement. Ultimate truth is obtained by rational cognition of suchness but “not established through its own essence.” It IS an object of knowledge. The division is exhaustive: no third truth is possible. The one explicit object-side reading of the two truths. | Ocean of Reasoning Ch 24; Illuminating the Intent on MA 6.23 | Tsongkhapa |
| Gorampa | Two levels of ultimate truth: quasi-ultimate (rnam grangs pa’i don dam bden pa / don dam pa rjes mthun pa) = the four-extremes-negation reached stepwise by rational analysis, ultimate vis-à-vis a truth-grasping mind but conventional vis-à-vis the equipoise of āryans; real ultimate (don dam mtshan nyid pa) = the simultaneous (cig char du) elimination of all four extremes in non-dual āryan equipoise, where even the appellation “ultimate” is withdrawn (Distinguishing the Views , .1.2). | Distinguishing the Views , .1.2 | Gorampa |
| Dolpopa | Ultimate truth = truly existent thoroughly established nature (tathāgatagarbha, dharmadhātu, self-arisen pristine wisdom); conventional truth = imputational and other-powered natures, empty of own entity. Ultimate truly exists; conventional does not. The two truths are “different… not the same entity… [yet] not different entities” (MD 405) — a categorial separation, not aspects of one reality. The three-natures framework carries the weight. | dolpopa-mountain-doctrine-1333 (Mountain Doctrine 342, 405) | Dolpopa |
| Tāranātha | Follows Dolpopa; distinguishes “Ordinary Middle Way” (rangtong — ultimate is mere negation) from “Great Middle Way” (zhentong — ultimate truly exists). The Ordinary Middle Way’s error is denying the true existence of ultimate truth. | Essence of Other-Emptiness | Tāranātha |
| Candrakīrti | Conventional things accepted “just as they are known in the world” without ultimate-level analysis; ultimate truth is emptiness understood through dependent origination. | Madhyamakāvatāra, Prasannapadā | |
| Mabja | Two truths depend on “two cognitive perspectives” (twelfth-century Tibetan, pre-polarisation): saṃvṛti is “true as the object of a conventional, deluded mind”; paramārtha is “true as the object of an ultimate mind and rational cognition.” The “vessel for water” image on MMK 24:10: “just as one who wishes for water needs a vessel, it is indeed necessary to teach the relative for there to be a realization of the ultimate.” Pre-Tsongkhapa Tibetan voice on the cognitive-perspective construal — sits structurally between the later object-side (Tsongkhapa) and subject-side (Sakya / Kagyü / Nyingma) readings. | Ornament of Reason Ch 24 commentary on MMK 24:8–10 | Mabja |
| Atiśa (Kadampa) | Ultimate reality is one, undifferentiated, beyond all conceptuality — not realised by valid cognition (pramāṇa). Conventional reality is a false projection of ignorance: “mere appearances” (snang ba tsam) without any real basis. Two types of conventional: mistaken (optical illusions, bad doctrines) and correct (dependent arisings that have causal efficacy when unexamined). Buddhas have no mind or wisdom-continuum — fully fused with dharmadhātu. No pramāṇa for ultimate; reasoning dissolves itself like fire consuming the sticks that produced it. | Satyadvayāvatāra (vv. 1–9); Madhyamakopadeśa; General Explanation | Atiśa |
| Kalupahana | Two truths are NON-hierarchical: saṃvṛti = convention (moral, social, linguistic — not merely language); paramārtha = “ultimate fruit/consequence” (pragmatic, not metaphysical). Neither sublates the other. No indication of “two truths with different standing as higher and lower.” | Philosophy of the Middle Way (1986) | Kalupahana |
| Karunadasa (Theravāda) | Theravāda sammuti-sacca / paramattha-sacca are NON-hierarchical — “two ways of presenting what accords with actuality,” not two degrees of truth; neither sublates the other (both amusā, only true). Etymology load-bearing: Pāli sammuti (“convention,” from man) ≠ Sanskrit saṃvṛti (“concealment,” from vṛ), so the Mahāyāna’s saṃvṛti term “clearly indicates” a hierarchy the Theravāda lacks. The distinction is an Abhidhammic innovation drawn out of the suttas (Aṅguttara nītattha/neyyattha; Saṅgīti Sutta sammuti-ñāṇa/paricchede-ñāṇa). Same etymological move as Kalupahana, from the Abhidhamma side. | The Theravāda Abhidhamma (2010) Ch 3 | Karunadasa |
| Śāntarakṣita / Mipham | Conventional truth analysed through Cittamātra (external objects refuted as mind’s projections); ultimate truth through Madhyamaka (mind itself lacks intrinsic nature). Approximate ultimate (rnam grangs pa’i don dam) = emptiness as conceptually cognised; actual ultimate (rnam grangs ma yin pa’i don dam) = beyond all conceptual elaboration. Both Svātantrika and Prāsaṅgika converge on the actual ultimate. | Madhyamakālaṅkāra (vv. 70–71, 92) | Śāntarakṣita, Mipham |
| Ninth Karmapa (Kagyü) | Two truths are neither the same nor different — beyond all elaborations of sameness or difference. Distinction drawn from the perceiving subject’s side: ultimate truth = what is seen by enlightened beings; relative truth = what is seen by ordinary beings. Even the ultimate truth, when presented as a dichotomous opposite to the relative, is itself a relative truth. Emptiness is a pedagogical tool, not a reified ultimate. Three stages: no analysis (conventions accepted), slight analysis (emptiness), thorough analysis (freedom from elaborations). | Feast for the Fortunate (on MA 6.23) | Ninth Karmapa |
| Dzongsar Khyentse (contemporary Khyentse lineage) | Two truths must be drawn subjectively (from the side of the cognising subject’s defect or wisdom) and never objectively (from the side of the object’s intrinsic existence). All non-Madhyamaka schools commit the object-side error. Four admissible bases for the subjective division (mind, pratītyasamutpāda, sūtra contents, knowledge) are interchangeable. | Madhyamakāvatāra commentary on MA 6.23 (1996–2000 Chanteloube teachings) | Dzongsar Khyentse |
| Siderits (analytic) | Two senses of “ultimate truth”: ultimate truth₁ = a fact that must be grasped to attain enlightenment; ultimate truth₂ = a statement that corresponds to the ultimate nature of mind-independent reality. The doctrine of emptiness says “the ultimate truth₁ is that there is no ultimate truth₂.” Only conventional truth survives, understood as semantic anti-realism (truth as warranted assertibility within useful conventions). Disambiguated satyadvaya device reconstructed in analytic-philosophy idiom; structurally close to Tsongkhapa’s bden grub qualifier but derived independently. | Buddhism as Philosophy Ch 9 (pp. 202–203) | Siderits |
| Garfield (analytic, Prāsaṅgika-Geluk) | The two truths are “Nāgārjuna’s greatest philosophical contribution,” introduced as distinct (24:8) but later identified: there is “no difference in entity” between the conventional and the ultimate; “to see them this way is precisely not to see them as conventional. To see that they are merely conventional… is thereby to see them as empty.” The identity holds because emptiness is empty (24:18). The conventional is not “less true,” and understanding the ultimate “is completely dependent upon understanding conventional truth.” Declares the reading “never conflicts directly with… Candrakīrti.” A modern identity-of-the-two-truths-via-emptiness-of-emptiness reading; it lands on the inseparability side, with the Tractatus register sitting in tension. | The Fundamental Wisdom of the Middle Way Ch XXIV | Garfield |
| Shakya Chokden (Sakya) | Conventional existence = nonexistence; if something exists it must be true (bden). No actual (go chod) ultimate truth was taught in Niḥsvabhāvavāda texts — their emptiness is a “mere metaphorical ultimate” (rnam grangs pa’i don dam). The actual ultimate truth is non-dual primordial mind (ye shes), taught in the third dharmacakra texts. Ultimate reality is an impermanent phenomenon. Two valid types of Madhyamaka ultimate: Niḥsvabhāvavāda’s self-emptiness (view determined by reasoning) and Alīkākāravāda’s other-emptiness (view experienced in meditation) — both lead to the same direct experience. | Profound Thunder, Rain of Ambrosia (via Komarovski 2011) | Shakya Chokden |
Textual loci
- MMK 24:8-10 — the locus classicus; understanding the two truths is necessary for understanding the Buddha’s teaching. Candrakīrti’s Prasannapadā commentary (see sprung-lucid-exposition-1979 Translation Ch XVIII, pp. 230–232) glosses saṃvṛti etymologically as “utterly obscured” by ignorance and as social convention (“the world of ordinary language and of transactions between individuals”); it renders MMK 24:10 with the image of ordinary language as “a container for someone who wants water,” the receptacle that carries the water of wisdom
- MMK 24:7 — Candrakīrti’s commentary: “dependent origination” = “emptiness” but “non-existence” ≠ “emptiness” (see sprung-lucid-exposition-1979 p. 234): “śūnyatā has the same meaning as dependent arising; the meaning of non-existence is not the meaning of absence of being”
- MMK 24:18 — pratītyasamutpāda = śūnyatā = prajñaptir upādāya = madhyama pratipad (see sprung-lucid-exposition-1979 p. 238). The translation of prajñaptir upādāya is itself a contested move at this kārikā: Sprung’s Wittgensteinian “non-cognitive guiding notion” rendering against the Yogācāra-Madhyamaka basis-of-designation reading
- Madhyamakāvatāra 6.23-28 — Candrakīrti’s exposition of the two truths
- Bodhicaryāvatāra 9.2 and 9.106 — Śāntideva’s two-truths exposition, read from the Nyingma side (see kunzang-pelden-nectar-manjushri-2007, ch. 9 following Mipham’s Norbu Ketaka). The commentary on BCA 9.2 gives two methods of positing the two truths: by examination of ultimate status (appearing mode / abiding mode) and by examination of relative status (whether appearance matches reality for a cognising subject — a subject-side construal). The commentary on BCA 9.106 supplies a strong framework-as-pedagogy statement: “the system of the two truths is propounded solely for didactic purposes, as an entry to the path. On the ultimate level, the division into two truths has no place.”
- Satyadvayavibhańga (Jñānagarbha) — Indian treatise devoted entirely to the two truths
- Vigrahavyāvartanī v. 28 — Nāgārjuna self-quoting MMK 24:10: “We do not speak without assenting to the conventional truth … Not having had recourse to the conventional, the absolute is not taught. Without having approached the absolute, liberation is not reached.” The self-referential defence of universal emptiness against the Naiyāyika dilemma turns on the Two Truths structure (see westerhoff-vigrahavyavartani-2010). The framework is not optional even at the meta-level of defending Madhyamaka against critics: Nāgārjuna states that all philosophical debate must take place at the conventional level
- Śūnyatāsaptati v. 44 — the most explicit Two Truths hermeneutical statement in Nāgārjuna’s surviving corpus: “Whatever is said by the Buddha has the two truths as its chief underlying thought; it is hard to understand and must be interpreted in this light. When the Buddha says ‘existence’ his chief underlying thought is conventional existence; when he says ‘non-existence’ his chief underlying thought is non-inherent existence.” Direct disambiguation of the Buddha’s “exists” / “does not exist” sayings into conventional existence vs non-inherent existence. It generalises MMK 24:8–10 from a doctrinal claim about the centrality of the Two Truths to a positive interpretive instruction covering the Buddha’s entire teaching (see komito-seventy-stanzas-1987). Śūnyatāsaptati v. 1 sets the Two Truths register at the outset (the Buddha’s own categorial vocabulary is saṃvṛti-level usage) and v. 70 supplies Nāgārjuna’s own diagnosis of why the nihilist misreading recurs (“Those who do not understand … are frightened by this teaching”)
- Vigrahavyāvartanī v. 70 — close parallel of MMK 24:14: “For whom there is emptiness, there are all things. For whom there is no emptiness there is nothing whatsoever.” Glossed by Nāgārjuna’s own autocommentary as: emptiness = dependent origination = the four noble truths = the path = the three jewels. The Two Truths structure is the precondition of the Buddhist path, not a threat to it
- Āryadeva’s Catuḥśataka — the Two Truths as architecture, not just doctrine. The Four Hundred Verses is organised in two halves: chapters I–VIII on conventional truth and the path (the means), IX–XVI on ultimate truth and the analysis of emptiness, with chapter VIII the explicit pedagogical hinge (the conventional must be correctly understood before the ultimate can be taught). The framework is thus built into the organisation of a treatise by Nāgārjuna’s immediate disciple, evidence that it is not a later commentarial imposition. Candrakīrti’s Catuḥśatakaṭīkā nonetheless criticises Dharmapāla for treating the two halves as “two distinct parts,” insisting the two truths are interconnected — so the structure witnesses the inseparability of the truths, framework-as-pedagogy rather than a two-level filing system (see aryadeva-four-hundred-sonam-2008)
Interpretations
Gorampa’s two-level ultimate truth (see gorampa-distinguishing-views-1469 , .1.2): Gorampa distinguishes between (a) the quasi-ultimate (rnam grangs pa’i don dam bden pa / don dam pa rjes mthun pa), which is the freedom-from-the-four-extremes that is the direct object of stepwise rational analysis but which is, “compared to the stainless equipoise of āryans, only a conventional truth”; and (b) the real ultimate (don dam mtshan nyid pa), reached only when the four extremes are eliminated simultaneously (cig char du), the realised reality and the realising mind do not appear as two, and the appellation “ultimate truth” is itself withdrawn — “at that time, there is no apprehension whatsoever of the fact, this is the ultimate truth.” This means that even the emptiness cognised through philosophical reasoning is not yet the real ultimate truth — it is its rjes mthun, “analogue.” This directly challenges Tsongkhapa’s position that emptiness as a non-affirming negation is the proper object of inferential cognition and the genuine ultimate. Translator’s note 386 records that the rnam grangs pa / mtshan nyid pa terminology has its origin in Indian Svātantrika texts (Jñānagarbha) — Gorampa repurposes Svātantrika terminology for a Prāsaṅgika reading. (From gorampa-distinguishing-views-1469)
Tsongkhapa’s position (in his own words, MMK-anchored): tsongkhapa-ocean-of-reasoning-1408 Ch 24 (p. 406) supplies Tsongkhapa’s most explicit object-side primary statement: “Each of the internal and external phenomena has two natures: an ultimate and a conventional nature. The sprout, for instance, has a nature that is found by a rational cognitive process … and a nature that is found by a conventional cognitive process. The former nature is the ultimate truth of the sprout; the latter nature is the conventional truth of the sprout. … But this does not show that a single nature is in fact two truths in virtue of the two perspectives of the former and latter cognitive processes.” Same Ch 24 develops the etymologies (saṃvṛti read three ways: concealment by ignorance, mutual dependence, signifier-nominal-convention; paramārtha as “fact-and-supremeness” rather than as the object of the wisdom of equipoise, against Bhāviveka). Cf. the parallel articulation at MA 6.23 in Illuminating the Intent: “two natures — conventional and ultimate — identical in nature but with distinct conceptual identities” (like “being produced” and “being impermanent”). The 1407–08 Ocean statement is the sharper anti-subject-side formulation; the 1418 MA-anchored statement is the more reconciliatory one. The decisive point: ultimate truth IS an object of knowledge, yet the ultimate is “not established through its own essence” (the pre-emption of the “grasping at emptiness” objection). Within conventional, veridical (unimpaired senses) vs distorted (impaired senses) is drawn only from the worldly perspective, not the ārya perspective. (From tsongkhapa-ocean-of-reasoning-1408 Ch 24 and tsongkhapa-illuminating-intent-1418 Ch 11)
Tsongkhapa’s bivalence defence — “no third truth”: Tsongkhapa explicitly defends the exhaustiveness of the two-truths division at MA 6.23 against critics (later including Mipham and the Ninth Karmapa) who treat the division as a pedagogical convention beyond which a third option lies. He argues by direct opposition: “deceptive” and “nondeceptive” are mutually exclusive predicates that admit no third term, citing Kamalaśīla’s Madhyamakāloka on direct opposites — once one is eliminated, the other is established by exclusion. He extends this to defend bivalent reasoning generally within Prāsaṅgika (“whether something does or does not exist,” “whether it is one or many”) against those who claim Prāsaṅgika rejects all direct opposites. The Meditation on the Definite Revelation of Suchness is cited verbatim: “There is the conventional and likewise the ultimate; there is never such a thing as a third truth.” (From tsongkhapa-illuminating-intent-1418, Ch 11)
Tsongkhapa’s three criteria for conventional existence: Conventional cognition (tha snyad pa’i shes pa) — what Candrakīrti calls “unexamined cognition” (ma dpyad pa’i shes pa) — is the kind of cognition that engages its objects within the context of how they appear, without analysing whether the appearance corresponds to actual mode of existence. It occurs in everyone whether or not their minds have been exposed to philosophical systems. To be conventionally existent, on Tsongkhapa’s account, a fact must satisfy three criteria: (1) it is acknowledged within conventional cognition; (2) it is not invalidated by another conventional valid cognition; (3) it is not invalidated by analysis probing the ultimate nature of reality. This is Tsongkhapa’s positive content for “deferring to the world”: not abdication to cowherds but appeal to the unanalysed perspective shared by reflective philosophers and ordinary persons alike. The three criteria distinguish his full-blooded conventional existence from the rope-snake (which fails criterion 2) and from intrinsic existence (which fails criterion 3). (From tsongkhapa-illuminating-intent-1418, Ch 11, on MA 6.25–28)
Tsongkhapa on the buddha’s single gnosis with two modes (Ch 25): The two-truths division as defined at MA 6.23 holds in general but admits a special case at buddhahood. Tsongkhapa argues — against the position that a buddha’s gnosis perceives only emptiness and that conventional perception is reserved for trainees — that a buddha possesses a single gnosis with two distinct modes of knowing defined by its two distinct objects (the way things really are, ji lta ba; and things in their diversity, ji snyed pa). At buddhahood the alternation between equipoise and post-equipoise ceases; the buddha continuously abides in equipoise on emptiness and yet perceives all of conventional reality without dualistic delusion. The cost of denying this, Tsongkhapa argues, is that the buddha would no longer be omniscient — the ten powers, defined in terms of knowledge of specific facts, would become incoherent. He concedes this is “a formidable but crucial challenge for Madhyamaka” rather than a free-standing claim, and explicitly rebukes “those who say that in this system there is no nonconceptual gnosis realising the ultimate truth” as denigrating the supreme realisation of the āryas. (From tsongkhapa-illuminating-intent-1418, Ch 25, on MA 11.10–17)
Gorampa’s critique of Tsongkhapa: Gorampa charges that this grasps at emptiness as a conceptual object, failing to transcend proliferation. The real ultimate is ineffable and beyond inferential cognition. (From gorampa-distinguishing-views-1469)
Gorampa’s verse-level reductio of Tsongkhapa on MA 6.23 (see gorampa-removal-wrong-views pp. 173–175 [Tib. 113–115]): Beyond the thematic charge of “grasping at emptiness,” Gorampa attacks Tsongkhapa’s “single nature with distinct conceptual identities” reading directly at MA 6.23 in five steps: (1) the single-nature thesis entails that “the nature found by a delusive perception is the nature found by an authentic perception”; (2) it entails that the cognitive procedures of the two truths are indistinguishable; (3) it entails that the meditative balance of the holy perceives the very basis found by delusive perception; (4) it makes the bden grub qualifier necessary — Gorampa derives Tsongkhapa’s central methodological move as a cost of the underlying error rather than as its solution; (5) Nāgārjuna’s Bodhicittavivaraṇa “produced/impermanent” analogy that Tsongkhapa cites in fact epitomises the non-existence of separate natures, not the existence of a single nature — the Centrist Consequentialist accepts neither. At MA 6.29 [Tib. 125, p. 188]: “This text [MA] also negates that the two truths are a single nature” — Candrakīrti’s own hairs/bell-metal image at MA VI.29 cited as a witness against Tsongkhapa. The disagreement is sophisticated and stays within the shared hermeneutical framework.
Dolpopa’s inversion (via Gorampa’s critique): Dolpopa identifies ultimate truth with truly existent Buddha-nature, reversing the standard Madhyamaka understanding. Gorampa classifies this as eternalism. (From gorampa-distinguishing-views-1469)
Zhentong Two Truths (via sympathetic Jonang account): In Tāranātha’s presentation, the conventional includes all imputational and other-powered natures — the entirety of apprehended-object and apprehending-subject. The ultimate is the thoroughly established nature: self-arisen pristine wisdom, tathāgatagarbha, dharmadhātu. The critical move is that the ultimate truly exists — it withstands analysis by the reasoning of dependent-arising and the lack of being one or many. Dolpopa’s Mountain Doctrine argues that if something is not ultimately true, it is not suitable as an ultimate truth — and therefore ultimate truth is ultimately true and conventionally untrue, while conventional truth is conventionally true and ultimately untrue. This is a strict asymmetry, unlike the rangtong positions where the two truths are aspects of the same reality. The “Ordinary Middle Way” errs by extending the negation of true existence even to the ultimate, leaving nothing truly established — which Dolpopa regards as falling into the extreme of non-existence. (See dolpopa-mountain-doctrine-1333 342, 405; “existing in the dispositional mode of subsistence” is what ultimately existing means, MD 225; concise synthesis at taranatha-essence-other-emptiness-2007, pp. 57–58.)
Kalupahana’s flat Two Truths: Kalupahana argues that the Buddha used sammuti to refer to all conventions (moral, social, political, linguistic, religious) — not merely language as Candrakīrti maintains. The specific terms for linguistic convention were nirutti-patha and adhivacana-patha. When Nāgārjuna speaks of two truths at XXIV.8-10, he distinguishes loka-saṃvṛti (worldly convention — the fruit of everyday existence) from paramārthataḥ (in terms of ultimate fruit). Artha and paramārtha are both truths (satya) — neither is presented as un-truth in relation to the other. Crucially, paramārtha is a pragmatic concept meaning “highest consequence” or “ultimate fruit,” not a metaphysical “ultimate reality.” Freedom (nirvāṇa) is not something that transcends conventional existence but an extension of ordinary goodness — “without the former the latter is not expressed” (XXIV.10) is a statement of pragmatic dependence, not ontological hierarchy. This directly opposes every other position in the comparison matrix. (From kalupahana-mmk-1986)
Karunadasa’s Theravāda non-hierarchical Two Truths (Abhidhamma-side companion to Kalupahana): In karunadasa-theravada-abhidhamma-2010 (Ch 3), Karunadasa argues the Theravāda sammuti-sacca / paramattha-sacca distinction is an Abhidhammic innovation whose antecedents lie in the suttas — chiefly the Aṅguttara nītattha / neyyattha pair (whose commentary explicitly correlates it to the two truths) and the Saṅgīti Sutta’s sammuti-ñāṇa / paricchede-ñāṇa. The Theravāda truths are emphatically non-hierarchical: “there is no hierarchical difference between the two”; they are “two ways of presenting what accords with actuality,” and the Buddha “speaks what is only true” whether teaching by sammuti or by paramattha (the “different dialects” simile). Strictly, paramattha-sacca “really means the truth expressed by using the technical terms expressive of the ultimate factors of existence” — the label attaches to the vocabulary, not to a higher grade of truth. The load-bearing argument is etymological and matches Kalupahana’s: Pāli sammuti (from man, “convention/consent”) is not Sanskrit saṃvṛti (from vṛ, “to cover/conceal”), and “the very use of the term saṃvṛti… shows that particular truth is less truthful and therefore inferior to… paramārtha-satya.” Two cautions. First, this is a tradition-located Sri-Lankan-Theravāda-academic reading, the Abhidhamma-side twin of Kalupahana’s MMK-side reading and convergent with Della Santina — not a neutral lexicographical finding. Second, Karunadasa’s claim that “in almost all other Buddhist schools one truth is higher” is too coarse: the non-hierarchical “two sides of one coin” reading he treats as uniquely Theravāda is precisely the Mahāyāna subject-side reading (Ninth Karmapa, Mabja, Mipham, MMK 24:10), while the hierarchy he ascribes to “almost all” schools really characterises only the object-side / zhentong minority (Dolpopa; arguably object-side Tsongkhapa). His charge thus collapses the very object-side / subject-side distinction it depends on. (From karunadasa-theravada-abhidhamma-2010)
Dzongsar Khyentse — Two Truths drawn subjectively, not objectively: In dzongsar-khyentse-madhyamakavatara-2003 (on MA 6.23), Khyentse Rinpoche makes the strongest version of the subjective-side claim available in any contemporary source: “Whatever these people decide about the two truths, they always make the distinction between relative and ultimate truth based on the object and its intrinsic nature or existence. The Madhyamikas refute this totally, and say that the distinction between the two truths must be made subjectively.” All four non-Madhyamaka Buddhist schools (Vaibhāṣika, Sautrāntika, Cittamātra) draw the distinction from the side of the object’s intrinsic existence; on this analysis, that is the error that defines them as non-Madhyamaka. Four admissible bases for drawing the distinction subjectively — mind, pratītyasamutpāda (interdependent origination, རྟེན་འབྲེལ་), the contents of sūtra, knowledge — are interchangeable “since in reality, there are no ‘two truths’.” Structurally close to the Ninth Karmapa position below and to Gorampa’s two-level ultimate, but framed more sharply: object-side two-truths is not Madhyamaka at all. The teaching is arranged according to Gorampa’s sa bcad, so the position is best read as a contemporary oral transmission of the Sakya/Gorampa reading rather than an independent thesis.
Siderits’s two-senses-of-ultimate-truth disambiguation (2007 textbook): In siderits-buddhism-philosophy-2007 (pp. 202–203), Siderits resolves the apparent self-contradiction of “the ultimate truth is that there is no ultimate truth” by distinguishing ultimate truth₁ (a fact that must be grasped to attain enlightenment — soteriological) from ultimate truth₂ (a statement corresponding to the ultimate, mind-independent nature of reality — correspondence-theoretic). The Madhyamaka thesis then reads coherently as “the ultimate truth₁ is that there is no ultimate truth₂.” Conventional truth survives without correspondence-grounding via this wiki-currency-off-the-gold-standard analogy (pp. 202–204): value (truth) derives from role in human practices, not from backing by intrinsically natured items. Structural significance: this is an independent analytic-philosophical reconstruction of the function Tsongkhapa achieves with the bden par grub pa qualifier — the negandum is not existence/ultimate-truth as such but a particular kind of existence/ultimate-truth (intrinsic, correspondence-theoretic). Siderits arrives at the device without naming the lineage. The Geluk move is therefore not idiosyncratic; it is what a careful analytic reader of MMK reconstructs when pressed. The disambiguation is structurally distinct from Gorampa’s quasi-ultimate / real-ultimate (Gorampa retains a yogically realisable real ultimate as a determinate state); from Westerhoff’s deflationary antirealism (Westerhoff retains a deflationary ultimate as conventional corrective); and from Sprung’s “the higher truth, in so far as it is a theory, falls within the lower truth” (Sprung’s move is Wittgensteinian, Siderits’s is correspondence-theoretic-deflation).
Siderits’s analytic-philosophical reach for the graded-teaching device (2000 academic): In siderits-reality-altruism-2000 (pp. 421–422), Mark Siderits — an analytic Buddhist Reductionist not given to commentarial-tradition pieties — defends Śāntideva’s argument from impartial benevolence (BCA 8.101–103) against Paul Williams’s “destroyed the bodhisattva path” charge by reaching explicitly for the neyārtha / nītārtha (provisional / definitive) structure: BCA 8 (“Perfection of Meditation”) precedes BCA 9 (“Perfection of Understanding”), “and it is only in the latter chapter that the doctrine of emptiness is established. Śāntideva is, in other words, giving us a ‘graded’ or progressive teaching (cf. MMK XVIII.8).” The cross-reference to MMK 18:8 — the Buddha’s graded teaching on self / no-self — is direct. Siderits does not name neyārtha / nītārtha in Sanskrit but the device is operative. It shows the graded-teaching framework doing real work even in a contemporary analytic register that has no antecedent commitment to it. Williams (Siderits’s interlocutor) refuses the device and arrives at incoherence; Siderits invokes the device and reconstructs a coherent reading. The text’s own internal warrant — BCA 9.1’s “all this was taught by the Buddha for the sake of gaining prajñā” — supplies the strongest single verse anchor.
Della Santina’s anti-Kantian Two Truths (1986 academic): In della-santina-madhyamaka-western-1986, Della Santina mounts a sustained critique of T. R. V. Murti’s The Central Philosophy of Buddhism (1955) and the wider mid-twentieth-century Western reading on which paramārtha is glossed “absolute” and saṃvṛti / vyavahāra “relative.” Three principal objections: (a) the etymology refuses these renderings — paramārtha denotes “highest end, purpose, or meaning”; saṃvṛti / vyavahāra denote “convention, usage, language” or “obscured, covered, veiled”; (b) “the division between the ultimate and conventional truths, for the Madhyamaka, is nothing more than a pedagogical device. Emptiness is said to be the ultimate truth, but emptiness is not an ontological category, but a soteriological therapy” (p. 49); (c) Kant’s noumenal/phenomenal split is itself ontological and dualistic and so cannot be mapped onto a non-ontological Madhyamaka. Della Santina also rejects the Kantian closure of the noumenal: citing Madhyamakālaṅkārakārikā v. 75, he holds that the ultimate is knowable mediately (“by the reason that sunders conceptualism”) as well as immediately by non-dual perception, contra Kant’s epistemic closure. Structural significance: an early (1986) framework-respecting Western-published rejection of the Kantian-absolutism reading from the Indo-Tibetan side, contemporaneous with kalupahana-mmk-1986 but pulling in the opposite direction. The “soteriological therapy, not ontological category” formulation is doctrinally underspecified — it does not engage the Tsongkhapa / Gorampa debate over whether ultimate truth is itself an object of knowledge — but the diagnostic against Murti is sharp.
Mabja’s twelfth-century cognitive-perspective gloss (see mabja-ornament-of-reason): Mabja Jangchub Tsöndrü’s Ornament of Reason — the earliest extant Tibetan MMK commentary — glosses MMK 24:8 as making the two truths the objects of “two cognitive perspectives”: “the relative truth of the world is characterized by being something that obscures the perception of reality, and so is true as the object of a conventional, deluded mind. All cognizable, expressible, and measurable phenomena are, thus, the bearers of this characteristic of the relative truth. The ultimate truth is characterized by being true as the object of an ultimate mind and rational cognition.” On MMK 24:9: “From the perspective of the two minds referred to above, there is, respectively, appearance and the complete lack of any established nature. The division into two truths, thus, depends on these two mental perspectives.” On MMK 24:10: “Just as one who wishes for water needs a vessel, it is indeed necessary to teach the relative for there to be a realization of the ultimate. Without relying on convention, moreover, the ultimate cannot be taught by any means, such as language and signs, because the essence of the ultimate itself lies beyond the sphere of language and mind.” Structural significance: Mabja is the pre-sectarian Tibetan voice on the cognitive-perspective construal, pre-dating both the Tsongkhapa object-side two-natures reading and the Sakya–Kagyü–Nyingma subject-side coalition. The “vessel for water” image converges with Candrakīrti’s “container for someone who wants water” at the parallel verse (sprung-lucid-exposition-1979 p. 232). Mabja deploys the Two Truths as the basic interpretive gesture at the outset of his commentary, before glossing any verse: the framework is the twelfth-century Tibetan default, not a Tsongkhapa-era over-reading.
Sprung’s Wittgensteinian Two Truths (1979 academic): Sprung’s Translator’s Introduction to sprung-lucid-exposition-1979 reads the Two Truths as a thesis about the cognitive vs non-cognitive uses of language. Saṃvṛti is the domain of cognitive (delusive) reference; paramārtha is reached only through the non-cognitive function of language — prajñapti as “a guiding, not a cognitive, notion presupposing the everyday” (p. 17). The decisive claim: “the higher truth, in so far as it is a theory, falls within the lower truth” (p. 16): any theory of śūnyatā must use entitative vocabulary and is therefore, qua taught, saṃvṛti. This anticipates the Ninth Karmapa’s “even the ultimate, presented as opposite to the relative, is itself relative” but reaches the position from a Wittgensteinian rather than a Madhyamaka starting-point. Sprung leaves standing an unresolved tension over whether MMK 24:10 implies two truths or three (he notes saṃvṛti / paramārtha / nirvāṇa could all be distinguished). Translation choices encode the reading: paramārtha = “higher truth which surpasses it” (anti-Kantian, refusing Stcherbatsky’s “transcendental reality” gloss); prajñaptir upādāya = “non-cognitive guiding notion presupposing the everyday.” Predates kalupahana-mmk-1986 and is the Western academic-Wittgensteinian predecessor to the deflationary line that runs through burton-emptiness-appraised-1999.
Westerhoff’s analytic anti-realist Two Truths (modern academic): In westerhoff-candrakirti-2024, Westerhoff glosses Candrakīrti’s MA 6:023–6:032 in three intertwined claims. (a) The “fire screen” image (MA 6:028, p. 79): conventional truth is a decorative screen constructed to conceal the empty cavity behind it; the figures on the screen are the fabricated entities and true statements about them are the conventional truths — but the whole arrangement is structured to hide, not to reveal. (b) “Semantic insulation” between the truths (MA 6:031, p. 81, citing Siderits): conventional statements cannot refute ultimate statements, but they can refute mistaken conventional renderings of ultimate statements (the “cars don’t really exist therefore I can steal yours” rebuttal goes through at the conventional level via legal codes, not via Madhyamaka). (c) Negation operates at both truths (MA 6:032 commentary, p. 82): Candrakīrti rejects intrinsic-nature production at both levels, with the qualification svalakṣaṇa-production rather than ordinary “mere conditionality.” Westerhoff explicitly warns against reading the Two Truths as an appearance/reality divide with a “rock-bottom layer of reality” behind appearances (MA 6:035, p. 84). The framework is tipped toward the Geluk/Tsongkhapa transactional-pragmatic reading of conventional truth (especially MA 6:082, p. 112: “those truths which are pragmatically successful within this framework constitute conventional truth”) rather than the Sakya/Gorampa internal-veridical reading.
Ninth Karmapa’s Kagyü Prāsaṅgika Two Truths: For the Ninth Karmapa (following the Eighth Karmapa Mikyö Dorje), the two truths are neither the same entity nor different entities — they are beyond all conceptual elaborations of sameness or difference. If they were the same, all sentient beings would see ultimate reality; if they were different, ultimate truth would not be the true nature of relative phenomena, making liberation impossible. The distinction is drawn from the perceiving subject’s side: ultimate truth is what is seen by realised beings free from ignorance; relative truth is what is seen by ordinary confused beings. The Karmapa then adds a sharper point: “even the ultimate truth, when presented as a dichotomous opposite to the relative truth, is a relative truth, because it exists merely as a conventional counterpoint.” This echoes the emptiness of emptiness — even the two truths framework is a pedagogical tool, not an ontological claim. The three stages of analysis (no analysis, slight analysis, thorough analysis) structure the relation between the truths: at thorough analysis, both truths dissolve into freedom from elaborations. (From karmapa-feast-fortunate-1578)
Śāntarakṣita / Mipham’s Yogācāra-Madhyamaka Two Truths: Śāntarakṣita’s Madhyamakālaṅkāra establishes a distinctive two-step approach: conventional truth is analysed through Cittamātra — external objects are refuted as mind’s projections — while ultimate truth is established through Madhyamaka reasoning that shows mind itself lacks intrinsic nature (v. 92). The text further distinguishes between an approximate ultimate (rnam grangs pa’i don dam) — emptiness as conceptually cognised, “attuned” to the ultimate — and the actual ultimate (rnam grangs ma yin pa’i don dam), which “is free from constructs and elaborations” (vv. 70–71). Mipham’s commentary uses this distinction to reconcile the Svātantrika-Prāsaṅgika divide: Svātantrika emphasises the approximate ultimate, Prāsaṅgika emphasises the actual ultimate, but both converge on the same actual ultimate truth. This is structurally parallel to Gorampa’s quasi-ultimate / real ultimate distinction but deployed reconcilingly rather than polemically. (From shantarakshita-madhyamakalankara)
The MA commentary adds three points on how the two truths function as a path rather than a static division: (a) a two-stage pedagogy — the approximate ultimate “as a first step… destroys clinging to reality” (the extreme of existence); then the actual ultimate “halts clinging to non-reality” (the extreme of non-existence), so the second truth is reached by consuming the negation, not by adding a thesis (raw p. 105); (b) the inseparability of appearance and emptiness as the content of the actual ultimate — “emptiness entails appearance; appearance entails emptiness… the two can never be separate” (raw p. 29), which is why Mipham calls a two-truths division that stops at the approximate ultimate a “mere juxtaposition… like a black thread and a white thread twisted together” rather than a genuine union; (c) the anti-Hashang guard — the freedom-from-elaboration of the actual ultimate is emphatically not the mere absence of mental activity (“the mere arresting of mental movement… is not even a cause of the dispelling of the ontological extreme of existence”, raw p. 106), so the second truth is self-cognising wisdom, not blank vacuity. This is the path reading of the two truths that underwrites madhyamaka-is-method-not-standalone-system.
Mipham’s “epistemic, not ontological” reading of MA 6.23 (see mipham-introduction-middle-way-2002): Mipham’s Word of Chandra commentary on stanza 23 glosses Candrakīrti’s “twin identity” as a feature of cognition, not of being: “this double identity (i.e., the two truths) is the conclusion of correct reasoning; it has no reality on the level of being.” Padmakara’s translators’ fn. 95: “The difference is epistemic, not ontological.” There is one reality (the inseparability of appearance and emptiness — the ultimate-in-itself); the appearance of two truths is generated by the structural difference between deluded consciousness and primordial wisdom. This is a substantively distinct fourth Tibetan reading of MA 6.23 alongside Tsongkhapa’s “single nature with distinct conceptual identities,” Gorampa’s “neither single nor separate” mtha’ bral, and the Ninth Karmapa’s apophatic “beyond same and different.” The Padmakara fn. 95 is the single sharpest formulation of the position in the primary sources. Polemical consequence (Supplementary Discussion ): a Madhyamaka that stops at the approximate ultimate (Mipham’s reading of how Tsongkhapa’s med dgag ultimate functions) cannot transcend the four extremes — “ultimate reality, qualified as a non-affirming negative, is able to refute the extreme of existence. But given that the refutation of the extreme of non-existence involves an appeal to relative truth, ultimate reality, from its own side, constitutes an ontological extreme (non-existence).” This is the structural argument that the Geluk position is not yet mtha’ bral, derived independently of Gorampa from the same MA 6.29 verse-base. Sakya in the 1470s and Nyingma around 1900 converge on the same architecture by separate routes.
Khenpo Kunzang Pelden’s two methods of positing the two truths (Nyingma, via the Norbu Ketaka; see kunzang-pelden-nectar-manjushri-2007): the commentary on Bodhicaryāvatāra 9.2 — following Mipham’s reading “closely, often verbatim” — distinguishes positing the two truths by examination of ultimate status (“the way phenomena appear is their relative truth; the way they actually are is their ultimate truth” — the appearing-mode / abiding-mode division) from positing them by examination of relative status (“when subject and object appear in such a way that there is a discrepancy between the way they appear and the way they really are, this is the relative truth… when subject and object appear in accordance to the way they actually are, this is the ultimate truth”). The second is a subject-side construal — the truths sorted by whether appearance matches reality for a cognising subject — and it lands the Nyingma firmly in the subject-side coalition with Mabja, the Ninth Karmapa, and Dzongsar Khyentse against object-side Tsongkhapa. The same commentary, at BCA 9.106, makes the framework-as-pedagogy point explicit: the two-truths division “is propounded solely for didactic purposes, as an entry to the path. On the ultimate level, the division has no place” — yet the device is indispensable (the lower five perfections are “blind” without the wisdom it carries). Pedagogy and necessity stand together. This is the Two-Truths reading on a Mahāyāna path-treatise (the BCA) rather than on the MMK/MA, broadening the subject-side base to a fourth textual locus.