Madhyamaka Wiki

Prāsaṅgika–Svātantrika Hierarchy Is a Tibetan Construction

9 min read

The claim

The classification of Indian Madhyamaka into two ranked sub-schools — Prāsaṅgika (higher, definitive) and Svātantrika (lower, provisional) — is a Tibetan codification dating to the late eleventh century (Pa tshab Nyi ma grags and Jayānanda) and elevated into a doctrinal hierarchy by Tsongkhapa in the early fifteenth. The Indian sources themselves — Buddhapālita, Bhāviveka, Candrakīrti, Śāntarakṣita, Kamalaśīla — exhibit diverse methodological positions but no ranked school-designation. Even within the Tibetan tradition, four major voices (Atiśa, Mipham, Ninth Karmapa, Shakya Chokden) reject or dissolve the hierarchy while remaining fully within the Mahāyāna hermeneutical framework.

Evidence for

  • Terminological evidence: the words are not in Indian sources. Ruegg in ruegg-svat-pras-2006 (p. 319) demonstrates that svātantrika and prāsaṅgika are unattested in Indian sources as school-designations. They are codified in Tibet by Pa tshab Nyi ma grags and his disciple Jayānanda at the end of the eleventh century.
  • Reception evidence: Candrakīrti was largely ignored in India. Per Ruegg (p. 319) and confirmed by Apple, Candrakīrti — the figure later Tibetan tradition treats as the founding Prāsaṅgika voice — received “near-silence” from Indian contemporaries and immediate successors. Real recognition came only around the tenth century, possibly through Atiśa’s influence. The “Prāsaṅgika tradition as codified in Tibet is substantially a Tibetan achievement,” not a transparent transmission of an established Indian school.
  • Manuscript evidence: Atiśa taught both without ranking them. Apple’s recovered Kadampa manuscripts (apple-jewels-middle-way-2018) show that Atiśa (982–1054) — at the very moment of Madhyamaka’s full transmission to Tibet — taught an undifferentiated “Great Madhyamaka” (dbu ma chen po) synthesising Bhāviveka publicly/pedagogically with Candrakīrti for advanced private instruction. Neither Atiśa’s writings nor the earliest Kadampa commentaries use the Prāsaṅgika/Svātantrika terminology.
    • The pairing is visible inside a single text. In Satyadvayāvatāra vv. 19–20, Atiśa cites Candrakīrti for the conventional as means and the ultimate as goal, then restates MMK 24:10 and repeats Bhāviveka’s staircase verse (Madhyamakahṛdaya 3.12) almost word for word (apple-jewels-middle-way-2018, n. 504). On the passage from the conventional to the ultimate the two later-ranked authors are cited as saying one thing.
  • Tsongkhapa’s own justification is retrospective. Per Ruegg (p. 320), Tsongkhapa justifies the Tibetan terminology by arguing that the philosophical distinction is consistent with Candrakīrti’s treatment, even though Candrakīrti did not use the formal nomenclature. This is a retrospective alignment, not a transmitted classification.
  • Convergent dissolution within Tibet. Four major Tibetan voices independently minimise or erase the hierarchy:
    • Mipham (Nyingma) via shantarakshita-madhyamakalankara: Svātantrika and Prāsaṅgika differ in pedagogical emphasis (approximate vs. actual ultimate), not in philosophical rank; both converge on the same actual ultimate. Three MA-commentary specifics make this the wiki’s sharpest dissolution: (i) the same “neither one nor many” reasoning is formulable as either a consequence or an autonomous inference (citing “the Madhyamakāloka and other texts”, by which entities of both kinds “may be refuted by Prasangika and Svatantrika arguments equally”) — so the method is presentational, not doctrinal; (ii) the genuine no-assertion ultimate belongs to ārya equipoise, on which “the Svātantrikas, like the Prāsaṅgikas, make no assertion,” and an assertion-making “Prāsaṅgika” is “no different from” a Svātantrika; (iii) Mipham’s own diagnostic — “in Tibet, even the explanation of the Prāsaṅgika view reverts to that of the Svātantrikas” — is an insider witness that the hierarchy does not hold even in practice within the tradition that most insists on it.
    • Ninth Karmapa (Kagyü) in karmapa-feast-fortunate-1578: “the Consequentialists and Autonomists differ in regard to the words they use to communicate, but their intentions are the same” (p. 165); the Karmapa explicitly defends Autonomists as genuine Followers of the Middle Way. What he dismantles is the distinction “imputed by later [Tibetans]”, which “the latter created anew” (heading, p. 166); he keeps a difference in how the two present the relative (p. 162).
    • Shakya Chokden (Sakya, heterodox) per komarovski-visions-unity-2011 (pp. 117, 137–139): the Prāsaṅgika/Svātantrika difference is merely methodological (autonomous reasons vs. consequences) and pedagogical; it cannot serve as a valid division of Madhyamaka. The genuine division is Niḥsvabhāvavāda / Alīkākāravāda.
    • Atiśa (Kadampa, predating all three): undifferentiated Madhyamaka with no nomenclature for the supposed split.
  • Westerhoff 2018 on the “doxographic fiction.” Per westerhoff-golden-age-madhyamaka-2018 p. 137, “the split into two sub-schools is certainly a doxographic fiction.” Westerhoff supplies two further data-points: (i) the terms Prāsaṅgika and Svātantrika are not used by Indian Mādhyamikas of themselves but are retranslations from later Tibetan doxographers; (ii) Bhāviveka’s commentator Avalokitavrata (c. 700 CE), the longest single work in the Tengyur, “does not appear to draw a distinction between the understanding of the two truths the two authors held” — if Bhāviveka had perceived a great rift between his interpretation and Buddhapālita’s, this commentarial silence would be peculiar. Westerhoff also makes the stronger Candrakīrti-side point: from Candrakīrti’s own perspective there are not two valid interpretations of Madhyamaka but only one (Bhāviveka being for him not a Mādhyamika at all but a tārkika). This is Western analytic-philosophical confirmation of the dissolution-tradition’s reading from a scholar who is otherwise willing to read Madhyamaka deflationarily — independent route, same verdict.
  • Westerhoff 2018 (VP) on Nāgārjuna’s actual practice. Per westerhoff-vaidalyaprakarana-2018 (Introduction, “The Aim of the Vaidalyaprakaraṇa”), the Vaidalyaprakaraṇa cannot be reduced to paraprasiddhānumāna (other-acknowledged inference): if Nāgārjuna had bracketed the Naiyāyika categories collectively as “merely the opponent’s premises,” there would be no reason to refute the categories one by one — the entire chapter-length structure of the VP is unmotivated under a pure-prasaṅga-only reading. Nāgārjuna in his most aggressive engagement with non-Buddhist realism takes the categories on their own terms and reformulates them in a svabhāva-free version. This is independent textual evidence, from Nāgārjuna’s own corpus, that the later Tibetan Prāsaṅgika–Svātantrika dichotomy does not fit Nāgārjuna’s actual method. Convergent with Westerhoff’s “doxographic fiction” verdict (above) and with ames-prajnapradipa-ch2-1995 on Bhāviveka’s paramārthatas-qualified syllogism similarly cutting across the typology.
  • Twentieth-century Geluk-internal criticism (Gendün Chöpel) — not a fifth voice of dissolution. Gendün Chöpel (1903–51), trained in the Geluk Prāsaṅgika-hierarchy reading at Bla brang and Sgo mang, attacks its grounds from inside: the Adornment (¶35–¶41, ¶51, ¶100) dismantles the Geluk identification of the object of negation by qualifier and the pramāṇa-warranted account of the conventional (lopez-madmans-middle-way-2006). But he keeps the Prāsaṅgika–Svātantrika distinction and redraws it: “Those who assert the presentation of the conventional as their own system are Svatantrikas, and those who assert it as the system of others are Prasangikas” (Adornment ). So he does not dissolve the distinction; he relocates it in whether the conventional is asserted as one’s own system or as others’. Nor is his independence uncontested: the editor of the 1997 Geluk refutations calls the Adornment an imitation of Gorampa, Shakya Chokden and Taktsang, while Shes rab rgya mtsho says it has no precedent at all; Lopez reads that polarity as “the compulsion to condemn” outrunning engagement with the arguments (“The Critics”). GC is evidence that the Geluk grounds for the hierarchy were contested from inside the Geluk academy, not a fifth independent dissolution.
  • Framework-external 1976 convergence (Kalupahana). kalupahana-buddhist-philosophy-1976 Ch. 11 — pre-reversal Kalupahana, working entirely outside the commentarial tradition — reaches a sympathetic reading of Bhāviveka against Candrakīrti on the ground that prasaṅga alone disables Madhyamaka from offering an empirical counter-thesis: “Bhāvaviveka believed in advancing a counter thesis or one’s own thesis (svatantra) … realising the difficulties inherent in the above position.” Kalupahana adds the historical observation that “this position [Prāsaṅgika] was not accepted wholeheartedly by some of the other Madhyamika teachers, especially Bhāvaviveka” — the Indian record on his reading does not support the Tibetan hierarchy. The point is hostile-witness valuable: Kalupahana arrives at the same verdict as Atiśa / Mipham / the Karmapa from a framework-removed early-Buddhist standpoint by a quite different route (the disabling-prasaṅga charge), strengthening the convergence claim by showing it is not an artefact of any one interpretive framework.
  • Eckel’s evidence on avicāramanohara. Per Ruegg (, p. 337), the “satisfaction of no analysis” — which Tsongkhapa presents as distinctively Svātantrika — had become standard across Madhyamaka by the eighth century, irrespective of supposed sub-school affiliation. Another data point that the categories Tsongkhapa codifies do not cleanly track Indian self-understandings.

Evidence against / objections

  • Ruegg’s mediating position. Ruegg himself does not endorse the strong “Tibetan construction therefore worthless” reading that ruegg-svat-pras-2006 attributes to Huntington. Ruegg’s own conclusion (, p. 345) is that the distinction retains “descriptive, taxonomic, analytical, and heuristic” usefulness when handled with historical awareness, because it tracks at least six interrelated philosophical criteria all converging on the status of saṃvṛti — and these criteria are real Indian philosophical issues. The argument should therefore claim that the hierarchy is a Tibetan construction, not that the underlying philosophical issues are.
  • Tsongkhapa’s defence has substance. Tsongkhapa’s six synonyms for the object of negation and his graduated Svātantrika-to-Prāsaṅgika pedagogy (tsongkhapa-illuminating-intent-1418 Ch 9) constitute a real philosophical position, not merely a back-projection. The argument must distinguish “the hierarchy is not given by the Indian sources” from “the hierarchy is unmotivated.”
  • The Tibetan codification is itself part of the tradition. Saying “the hierarchy is a Tibetan construction” risks the implication that it is less authoritative than the Indian material. The framework-necessity thesis has to maintain that framework-internal Tibetan developments are part of the framework’s productive elaboration, not deviations from it. The argument therefore must avoid sliding into the Western critic’s “back to the Indian sources” move that Burton makes for different reasons.
  • The four-voice convergence may overstate independence. Atiśa, Mipham, the Ninth Karmapa and Shakya Chokden are not perfectly independent — Mipham reads Atiśa’s Kadampa material; Shakya Chokden engages directly with the Sakya context shared with Gorampa. (Gendün Chöpel is sometimes counted as a fifth, but he keeps the distinction, redrawn by whether the conventional is asserted as one’s own system, .) The Ninth Karmapa, too, rejects only the distinction “imputed by later [Tibetans]” (Feast p. 166), not every difference (p. 162). The “convergence” needs to be characterised carefully: similar conclusions reached through different routes, not strictly independent confirmation. Kalupahana 1976 (kalupahana-buddhist-philosophy-1976) is more strongly independent of the Tibetan voices but converges via a different (and hostile) route, which strengthens the non-school-internal-artefact reading of the dissolution.

Linked pages

Graph View

Backlinks