Dharmakīrti

Dharmakīrti (ཆོས་ཀྱི་གྲགས་པ་)

Every later Buddhist logico-epistemological text in India and Tibet argues with Dharmakīrti or from him. Two of the three root texts of this vault are his pramanavartika and its Tibetan reworking in sakya-pandita’s pramanayuktanidhi.

Westerhoff spells the name Diṅnāga; this page uses Dignāga throughout, as the vault does elsewhere.

Life

Dharmakīrti studied dignaga’s pramanasamuccaya with the latter’s disciple Īśvarasena, forming the lineage Vasubandhu → Dignāga → Īśvarasena → Dharmakīrti. Īśvarasena recognised that Dharmakīrti’s understanding surpassed his own and equalled Dignāga’s, encouraging him to compose the pramanavartika (westerhoff-dignaga-dharmakirti, p. 218).

Dating

Frauwallner dates Dharmakīrti to 600–660 CE, on the negative evidence that the Chinese pilgrim Xuanzang (mid-7th c.) did not mention him while Yijing (late 7th c.) did. Westerhoff calls this “hardly decisive”. Balcerowicz, reading the Jaina discussions of Dharmakīrti, puts him a half-century earlier, 550–610 CE (westerhoff-dignaga-dharmakirti, p. 217). Nothing in this vault settles the question, and the two datings put him on opposite sides of Xuanzang’s visit. [unverified]

Legendary Debates

Traditional accounts describe Dharmakīrti infiltrating Kumārila Bhaṭṭa’s household incognito as a servant in order to learn all his doctrines before later defeating him in debate. He is also said to have defeated Śaṅkara, who drowned himself in the Ganges, only to be reborn and defeated again, twice (westerhoff-dignaga-dharmakirti, pp. 218–219).

Bu-ston’s version of the Kumārila episode adds a striking detail: in his telling the teacher is not named Kumārila but identified as Dharmakīrti’s maternal uncle, and Dharmakīrti is said to have learned his uncle’s most guarded doctrines by having the uncle’s wife relay difficult questions to her husband during intercourse, with Dharmakīrti listening in and “pulling on a cord tied around her leg” when subtle subjects were broached (westerhoff-dignaga-dharmakirti, p. 218 n. 9). Stcherbatsky reads the whole cycle of debate stories against itself: it is “an indirect confession that these great brahmin teachers had met with no Dharmakīrti to oppose them”, evidence that Buddhism in Dharmakīrti’s time was no longer flourishing as it had under Asaṅga and Vasubandhu (westerhoff-dignaga-dharmakirti, p. 219 n. 11).

What these biographies dwell on is not meditation or teaching but success in public debate against trained non-Buddhist opponents. That setting explains a feature of the system that otherwise looks arbitrary: before the doctrines can be argued at all, the two sides need a shared account of which sources of knowledge either may appeal to, and neither can simply cite its own scripture (westerhoff-dignaga-dharmakirti, p. 219).

Dharmakīrti’s Own Assessment

The opening and closing verses of the Pramāṇavārttika read, on Westerhoff’s account, as a reflection on how little his contemporaries took up: ordinary beings, attached to common things, cannot grasp the depth of the work, and so “like the water of the sea, his work will remain in itself, without being absorbed by anybody” (westerhoff-dignaga-dharmakirti, p. 247).

Philosophical Position

Tibetan doxography files Dharmakīrti as a Sautrāntika following reasoning (རིགས་པའི་རྗེས་འབྲང་གི་མདོ་སྡེ་པ་). Dreyfus calls the term oxymoronic, and says the Tibetans coined it precisely because the view defended through most of Dignāga’s and Dharmakīrti’s texts is not the view usually ascribed to Sautrāntika; the label exists to force them into the four-school scheme rather than to describe them (dreyfus-recognizing-reality, pp. 20–21).

Ontology

Dharmakīrti holds a sharply antirealist ontology:

• Reality consists exclusively of momentary (སྐད་ཅིག་མ་, kṣaṇika), causally effective individual thing-events
• The real is defined as the specifically-characterised (རང་མཚན་, svalakṣaṇa)—that which has definite spatio-temporal location and individual essence
• All universals, commonsense enduring objects, and abstract entities are conceptual constructs—generally-characterised (སྤྱི་མཚན་, sāmānyalakṣaṇa)
• The system needs universals to ground inference and denies them any place in reality. That is the tension the apoha theory and the sliding scales below are both built to absorb

Epistemology

• valid-cognition (ཚད་མ་, pramāṇa) is defined as nondeceptive cognition (མི་བསླུ་བའི་ཤེས་པ་, avisaṃvādi-jñāna)
• Two types: perception (མངོན་སུམ་, pratyakṣa) and inference (རྗེས་དཔག་, anumāna)
• Nondeceptiveness involves both a practical dimension (readiness to perform a function) and a normative/intentional dimension (revealing a hitherto unknown object)
• The novelty requirement excludes memory from being valid cognition
• Unlike Dignāga, Dharmakīrti makes immunity from error (abhrānta) part of the very definition of perception (Pramāṇaviniścaya 1.4: pratyakṣaṃ kalpanāpoḍham abhrāntam) (westerhoff-dignaga-dharmakirti, p. 223 n. 23). Every error is thereby relegated to the mind, which has a consequence Eltschinger draws out: “as far as perception is concerned, there is no difference between an ordinary mind and one that is liberated”. The Buddha’s perceptual contact with reality (tattvadarśana) is direct and error-free, and he resorts to concepts only when teaching others (westerhoff-dignaga-dharmakirti, p. 223 n. 23).
• Dharmakīrti’s perception theory connects the perceiver to the world through both causation (the external object causes the ākāra via the sense faculty) and resemblance (the ākāra has the form of the external object), the structure conventionally associated with the Sautrāntikas (westerhoff-dignaga-dharmakirti, p. 223 n. 24).

Philosophy of Language

• Developed dignaga’s apoha theory: words and concepts function by exclusion (སེལ་བ་), not by picking out positive properties
• Concepts are fundamentally mistaken—they superimpose a false unity on reality’s discrete particulars—yet they can be pragmatically valid
• Innovation over Dignāga: Dharmakīrti anchors apoha in causal power relativised to human desires.

Dignāga’s apoha leaves a gap that Dharmakīrti’s does not. If the concept “fire” is built by excluding non-fire, something must already have collected the fire-particulars together, and Dignāga does not say what. Dharmakīrti’s answer is that nothing in the objects does it. Fire-particulars differ from each other in every respect that matters ontologically: wood fire and dung fire, smoky and smokeless, red flame and blue. What they share is that they all satisfy one insufficiently specific human want, warmth. The grouping is therefore picked out relative to human interests and not by any characteristic of the objects held apart from those interests.

The objection this invites is that the objects must at least share the causal power. Dharmakīrti’s counter-example is antipyretic drugs: they all answer to the want for something that lowers fever, and they do it by causally different routes. Satisfying one need is all the grouping consists in; no common underlying property, and so no object-type, is needed behind it (westerhoff-dignaga-dharmakirti, p. 237).

Against Mīmāṃsā, Dharmakīrti’s target is the inference from the Vedas’ having no author (apauruṣeya) to their authority. Mīmāṃsā reaches it by way of a second thesis, that the relation between words and their referents is “primordial” rather than conventional (p. 263), so that a well-formed Vedic sentence must be meaningful, and, since falsity comes from defects in a speaker and the Vedas have no speaker, must be true. Dharmakīrti turns the last step around: if falsity derives from defects in the speaker, truth derives from the speaker’s reliability, which rests in turn on other good qualities, compassion in the Buddha’s case. A statement with no author at all is then either false, being backed by nobody reliable, or without meaning (westerhoff-dignaga-dharmakirti, p. 265).

Seven Treatises on Valid Cognition

Dharmakīrti is credited with seven treatises (ཚད་མ་སྡེ་བདུན་), of which the pramanavartika (Commentary on Valid Cognition) became the primary text in Tibet thanks to sakya-pandita’s influence.

Sliding Scales of Analysis

Westerhoff attributes to Dharmakīrti four levels of philosophical analysis in ascending order of sophistication (westerhoff-dignaga-dharmakirti, pp. 253–254):

1. Ordinary beings—pragmatically successful but philosophically flawed (satkāyadṛṣṭi)
2. Abhidharma reductionism—persons and partite objects are mere verbal designations on conglomerations of dharmas
3. Particularism (Sautrāntika)—spatial, temporal, and conceptual extension are all cognitive errors; reality consists only of utterly distinct momentary particulars
4. Idealism (Yogācāra)—the duality of perceiving subject and perceived object is illusory; all phenomena are mental

Dharmakīrti argues mainly from level 3, which he knows to be false. It is the lowest common denominator, the best available balance of conceptual sophistication against broad acceptance for inter-doctrinal debate (westerhoff-dignaga-dharmakirti, p. 255). Only one chapter of the Pramāṇavārttika consistently takes up the idealist point of view.

Mipham quotes Dharmakīrti: “When I investigate outer phenomena, I take the Sautrantika as my starting point” (shantarakshita-madhyamakalankara-intro, p. 27). The Padmakara translators reconstruct why he could not stay there. On the Sautrāntika account the aspect (རྣམ་པ་) is supposed to bridge mind and matter, but the aspect is itself a feature of consciousness, so a new and unexplained gulf opens between the material object and the mental aspect. Only two exits are available, and both abandon Sautrāntika premises: posit a third principle that is neither mind nor matter, or deny that the mind-matter separation is real. Dharmakīrti takes the second, which is to say he evades the difficulty rather than answering it (shantarakshita-madhyamakalankara-intro, pp. 26–27).

The sharpest form of the same problem is that the causal relation is itself unreal by Dharmakīrti’s own criteria. Causal power is what makes a particular real; but relations, like object-types, are neither identical with their relata nor separate from them, and the neither-one-nor-many argument therefore cuts them away (westerhoff-dignaga-dharmakirti, p. 256). The mark of the real is not real. On Westerhoff’s reading the sliding scales absorb this rather than answer it: level 3 was never Dharmakīrti’s final view, so an inconsistency internal to it is not a refutation of him. What that leaves unsettled is how far up the scales the same argument keeps working, since it is not obvious why it should stop at consciousness.

Double Readings and Apologetic Drift

Westerhoff picks out two features of Dharmakīrti’s writing that a reader approaching him as a system-builder will miss. The first is the double nature of the project: many passages are phrased so that they carry both a Sautrāntika and a Yogācāra reading, and which one dominates is left to the interpreter. That fits his stated practice of putting his ideas in vocabulary acceptable to non-Buddhist interlocutors, borrowing terms from rival traditions and assigning them altered senses, so that the appearance of continuity survives across the boundary even where the sense does not (westerhoff-dignaga-dharmakirti, p. 246).

The second is the apologetic character of the writing. The system is recruited to establish not general epistemological theses but specifically Buddhist ones: the four noble truths, karma, rebirth, momentariness, the Buddha’s omniscience, an innate Buddha-nature.

The thesis that this marks a drift in the whole school, from Dignāga’s relatively neutral epistemology towards the defence of orthodoxy, is Richard Hayes’s, and Westerhoff reports it without endorsing it: whether the notion of “genuinely disinterested philosophical investigation” underlying that reading is a useful concept for ancient Indian thought he calls “at least questionable”. He is equally careful about the further suggestion that the apologetic turn is why Dharmakīrti’s standing among Buddhists is the direct opposite of his repudiation by Nyāya, Vaiśeṣika and Mīmāṃsā thinkers, and calls that “debatable”; what he does say is that the reaction to Dharmakīrti from those quarters was clearly negative, and may have contributed to hardening the boundaries between Indian philosophical schools (westerhoff-dignaga-dharmakirti, p. 246). Eltschinger sets the whole development wider still, reading both the logico-epistemological school and Buddhist tantra as responses to sixth-century brahmanical pressure, dialectical and ritual counter-moves against an increasingly powerful Brahmin establishment (westerhoff-dignaga-dharmakirti, p. 234 n. 53).

Scriptural Authority

Dharmakīrti subsumes scripture under inference, so that testimony (śabda) need not be posited as a third epistemic instrument. The Buddha’s authority he establishes by unpacking the epithets of Dignāga’s homage verse at the head of the Pramāṇasamuccaya: “seeking the benefit of the world” names great compassion, which set him to practise the means of liberation, which made him a teacher, which made him a sugata. Telescoped, “compassion is the proof” of his embodying an epistemic instrument (westerhoff-dignaga-dharmakirti, p. 239).

The argument’s own premise is what Dharmakīrti then refuses to claim knowledge of. Mental states are supersensible, so nobody is in a position to verify that the Buddha has the qualities the argument runs on. Scriptural authority is accordingly a practical necessity (agatyā) for someone who wants liberation and has no other route to it: a defensible and fallible choice, not an epistemic instrument (westerhoff-dignaga-dharmakirti, pp. 242–244). See scriptural-authority.

Yogic Perception

Yogic perception (yogipratyakṣa) is not an epistemic super-power but the last of three stages, after intensity (prakarṣa) and termination (paryanta). Unlike the Vaiśeṣika version it is not validated by giving access to the fundamental structure of the world; Dharmakīrti puts it in the same class as vividly appearing dreams and fantasies, and separates it from them only by result, the one deepening saṃsāra and the other providing the way out. Its authority is therefore its effect, and that has a consequence he draws: the practitioner needs to be sure of directing it at the right object (the four noble truths), and that guidance can only come from the instruction of earlier authorities. Yogic perception cannot generate liberating insight on its own; it has to be combined with scriptural authority (westerhoff-dignaga-dharmakirti, pp. 247–249).

Indian Commentarial Traditions

Stcherbatsky’s Buddhist Logic divided Dharmakīrti’s Indian commentators into three traditions, and Dreyfus adds a fourth:

1. Literal exegesis: Devendrabuddhi (630–690 CE), his student Śākyabuddhi
2. Religious school: Prajñākaragupta (740–800 CE)
3. Philosophical school: dharmottara (750–810 CE), Śaṃkarānanda
4. shantarakshita (725–783 CE) and his disciple Kamalaśīla (740–795 CE)

The line Dreyfus draws is between the first tradition and the rest. Devendrabuddhi’s school attempts a literal rendering; the later ones, under repeated attack from Hindu orthodoxy, let the need to answer take precedence over gloss, and accuse Devendrabuddhi of having misunderstood Dharmakīrti. The fourth group is not itself a “Mādhyamika interpretation” of Dharmakīrti: Śāntarakṣita and Kamalaśīla transformed the apoha theory and influenced the later Indian commentators who read Dharmakīrti as a Mādhyamika, a reading early Tibetan commentators then took up (dreyfus-recognizing-reality, p. 21).

Tibetan Reception

Dharmakīrti’s vocabulary (valid cognition, elimination, specifically and generally characterised) is used across Tibetan philosophical writing, Madhyamaka included, and the monastic debate format rests largely on his logic. Dreyfus notes the reversal of purpose that came with the transmission: in India the tradition earned its place by defending Buddhism against Hindu critics, and in Tibet, where no such opponent existed, that justification was gone.

Two rival interpretive traditions emerged:

• Sakya antirealism: sakya-pandita, gorampa, sakya-chok-den hold that Dharmakīrti is genuinely antirealist about universals
• Gelug moderate realism: gyel-tsap, kay-drup, Ge-dün-drup reinterpret him to accommodate real universals and commonsense objects

Which side is reading him correctly is the question realism-vs-antirealism takes up.

What this page does not rest on

Nothing here is drawn from Dharmakīrti’s own text. The Pramāṇavārttika is held in this vault (raw/root-texts-tb/pramanavartika/) and all four chapters are drafted in translations/pramanavartika/, but every claim on this page comes from a secondary account. The two that carry the most weight, the sliding scales and the double reading, are both Westerhoff’s interpretive frame rather than a doctrine Dharmakīrti states, and the primary evidence for the four levels is not reproduced in the chapter. [unverified]

Sources

• dreyfus-recognizing-reality—the central subject of the entire book
• westerhoff-dignaga-dharmakirti—life, epistemology, metaphysics, inference, scriptural authority, yogic perception, the sliding-scales model. All page references verified against the extraction; the p. N markers there label the text that follows them
• shantarakshita-madhyamakalankara-intro—the aspect theory’s collapse and the shift from Sautrāntika to Yogācāra, read through Mipham
• katsura-dignaga-lectures—Dharmakīrti’s Vādanyāya as full treatment of Points of Defeat; endorsement of Dignāga’s twofold-appearance theory in the Pramāṇaviniścaya