Pramāṇavārttika

Pramāṇavārttika (ཚད་མ་རྣམ་འགྲེལ་)

Commentary on Valid Cognition — dharmakirti’s magnum opus and the single most important text in the Buddhist logico-epistemological tradition. It became the primary text for Tibetan epistemological studies thanks to sakya-pandita’s influence.

Overview

Presented as a commentary on dignaga’s pramanasamuccaya, the Pramāṇavārttika in fact constitutes an independent and far more elaborate treatise. It consists of 1,452 verse-form arguments (kārikā) across four chapters: Chapter 1 contains 340 stanzas, Chapter 2 contains 285 stanzas, and the remaining stanzas fall in chapters 3 and 4. Only the first chapter has an autocommentary (svavṛtti) by Dharmakīrti himself (gorampa-pramanavartika).

The work was translated into Tibetan three times: first by Ma Lotsāwa Gewai Lodrö (1044–89) with Subhūtiśrīśānti; second by Ngok Lotsāwa Loden Sherab (1059–1109) with Bhavyarāja; and third by sakya-pandita with the Kashmiri master Śākyaśrī — it is this third translation that became standard.

Opening Homage and Pledge to Compose

The work opens with Dharmakīrti’s famous homage to the Buddha:

I prostrate to the one in whom the web of conceptions has completely disappeared, who possesses profound and vast forms, and whose light rays of that which is good in every aspect radiate everywhere.

gorampa reads the homage as a praise of the Buddha’s two perfect needs: the needs of self (fulfilled by the three enlightened forms — natural-purity form, enjoyed form profound because confined to Akaniṣṭha, and manifested form vast because suited to all three types of disciples) and the needs of others (the “good in every aspect” of high rebirth and nirvāṇa taught by the light rays of the sacred Dharma) (gorampa-pramanavartika).

Dharmakīrti immediately follows with the candid statement that the work will not directly benefit faulty listeners — “those attracted to the ordinary [who] do not possess the skill of discernment … wholly endowed with resentment, the stain of envy” — and that he therefore composes it primarily to familiarize his own mind with its tenets. Gorampa unpacks this as identifying the three faults of a faulty vessel (no discernment, no aspiration, prejudice) and the three corresponding qualities of a fit vessel, citing Sakya Paṇḍita’s pramanayuktanidhi: “With discernment, freedom from prejudice, and a mind of compassion for others will this be explained.”

The text chapter by chapter

Chapter 1: Inference for Oneself (རང་དོན་རྗེས་དཔག་, svārthānumāna)

340 stanzas. The only chapter that comments directly on the Pramāṇasamuccaya and the only one for which Dharmakīrti supplied an autocommentary (svavṛtti). It establishes inference as a valid means of cognition and lays out the entire logical apparatus of the Dharmakīrtian system — the three perfect reasons, the indispensable relationship, the apoha theory of universals, the proof of momentariness by disintegration, and the long critique of Vedic authority. The first three quarters are the logical machinery; the final quarter is its application to a single polemical target, the Mīmāṃsā doctrine that the Vedas are authoritative because authorless.

Why this chapter comes first. Dharmakīrti reorders Dignāga, who had opened with perception. Gorampa surveys and rejects several Indian explanations of the reordering before giving his own (gorampa-pramanavartika §2). Śākyabuddhi had taken Dharmakīrti’s phrase “separating what is meaningful from what is without meaning” — drawn from the opening of Dignāga’s autocommentary — to refer to the difference between correct (Buddhist) and incorrect (non-Buddhist) definitions of pramāṇa; on this reading, inference is taught first because that separating depends on inference. Devendrabuddhi had read it as “inference is the basis for the explanation of the treatise on the characteristics of valid cognition.” Prajñākaragupta’s Ornament even claimed that the original order was Dignāga’s and that the redacted volume placed Inference for Oneself first only because it happened to have an autocommentary. Gorampa rejects all three. For him “meaningful” denotes Śākyamuni Buddha as a valid being — the ultimate point Chapter 2 establishes — and “without meaning” the views that deny him; the proofs in Chapter 2 depend on entity-based reasoning that must be developed first. The logic of Chapter 1 is propaedeutic to the soteriology of Chapter 2, and the reordering is doctrinally motivated, not redactional.

The three perfect reasons (PV 1.1–6). Inferential reasoning is defined as that in which all three criteria (ཚུལ་གསུམ་, trairūpya) are complete: the reason is established as a property of the subject; the predicate concomitantly pervades the reason in the congruent class; and the counter-pervasion holds in the incongruent class. Reasons satisfying these criteria are exhaustively of three types — by result (འབྲས་བུ་, kārya), by nature (རང་བཞིན་, svabhāva), and by nonapprehension (མ་དམིགས་པ་, anupalabdhi). Result-reasons infer a cause from its indispensable effect (smoke → fire, PV 1.2); nature-reasons infer one feature from another feature of the same entity (the impermanence of sound from its being a product, PV 1.2cd); nonapprehension-reasons infer absence (PV 1.3–6), with four standard types — apprehension of contradictory nature, contradictory result, non-establishment of an appearing cause, non-establishment of an appearing entity — which Gorampa extends to eight and (following Pramāṇaviniścaya) to ten. The exhaustiveness of the threefold scheme is then defended against Mīmāṃsā alternatives, in particular the inference of a result from uncollected causes (PV 1.7–8: this is reasoning by nature, not by result, because the appropriateness of the result inheres in the causal collection) and the inference of form from taste (PV 1.9–10: this is reasoning by result, since both depend on a common prior cause).

The indispensable relationship (འབྲེལ་པ་, avinābhāva, PV 1.11–38). The heart of the chapter, and the foundation of all later logic. Gorampa argues that same-nature (svabhāvapratibandha) and cause-and-effect are the only two grounds of pervasion; there is no third. Without an indispensable relationship between reason and predicate, merely not seeing the reason in the incongruent class does not establish counter-pervasion — Dharmakīrti’s standing example is the termite-hole, where smoke is seen without fire, refuting any inference that takes mere co-occurrence as ground (PV 1.12). The same-nature relationship is then defended at length: product and impermanence are one entity but distinct isolates; commentators who take them to be of separate substances yet still function as reason and predicate “have not understood the intent of the text” (Gorampa §13).

Exclusion and universals — apoha (གཞན་སེལ་, PV 1.40–166). Numerically the largest topic-cluster in the chapter. Verbal expressions and conceptual cognition engage their objects only by exclusion of other (anyāpoha), never by affirmation; only direct perception engages by affirmation, and it does not conceptually distinguish features. The chapter develops two complementary formulas — “the one conceptualised as many” (a single entity supports multiple isolates, ldog pa, vyatireka) and “the many conceptualised as one” (multiple particulars are subsumed under a non-real universal). Universals, wholes, and durations are non-entities; they are conceptual designations imposed by mistaken cognition. Gorampa lays out the canonical statement at PV 1.40 — “All entities dwell by nature in their own individual entities, they are the support of being isolated from entities, both compatible and other” — and proceeds through a sustained refutation of Nyāya-Vaiśeṣika realism: a single universal can abide neither by dependence on its particulars nor by being illustrated by them; no single result arises from a shared universal (PV 1.143–150). Designations engage objects only indirectly, through a universal (a mental image) appearing as if it were a particular (PV 1.78–84).

Divisions of reasoning by nature and disintegration (PV 1.186–197). Disintegration is by the phenomenon’s own nature, not by any cause beyond its direct producer. The argument (PV 1.193–197) is that if disintegration depended on a separate cause such as a hammer, entities would be seen to persist where no such cause arose (“for some it will not be possible,” PV 1.194), and even where a destroyer is present its operation is not certain. Therefore “the destruction of a pot is by way of its own entity, because it was created as an entity of disintegration without dependence upon any cause other than its own.” This is the vināśitvānumāna, the inference from disintegration, which establishes momentariness. dreyfus-recognizing-reality (pp. 65–66) notes that later texts — Pramāṇaviniścaya, Nyāyabindu, Vādanyāya — replace this argument with the sattvānumāna, the inference from existence, reflecting different polemical pressures from the Nyāya school; the PV version is the one Tibetan commentators inherit.

Fallacious reasoning and the refutation of the Vedas (PV 1.213–340). The chapter’s long final quarter, occupying roughly the last hundred stanzas. Gorampa structures the polemic in four stages. (i) Against the Mīmāṃsā thesis that the Vedas are authoritative because unproduced by a person (1.224ff.), Dharmakīrti’s reply is the canonical PV 1.226: “The cause of understanding meaning is linguistic convention, which depends upon people. Thus, even if the words were not produced by a person, the possibility of error remains.” (ii) Against the response that the relationship between Vedic sounds and their meanings is natural rather than conventional (1.227–230), Dharmakīrti notes that a permanent natural relationship would make either all meanings simultaneously available from any sound (1.228) or contradictory meanings derivable from the same sound — closing with the famous “washing an elephant” image (PV 1.230): the Mīmāṃsaka, to remove the dirt of error, asserts the Vedas were not produced by a person, only to be smeared by the wetness of having no determinable meaning. (iii) Examination of permanence vs. impermanence of the relationship (1.231ff.) and the syllables (1.243ff.). (iv) Examination of Vedic mantras and the impossibility of comprehending their definitive meaning.

Gorampa’s introduction (§2) enumerates the specific “mistaken conceptions” Dharmakīrti refutes across the chapter: that there are anywhere from one to six criteria of a perfect reason (rather than three); that a result naturally arises from a complete collection of causes (refuting the Mīmāṃsā move at 1.7); that counter-pervasion can be proven merely by not seeing the reason in the incongruent class (the termite-hole problem); that product and impermanence, though substantially separate, function as reason and predicate; that if reason and predicate are of one nature they cannot stand as reason and reason-possessor; that verbal expressions directly express intrinsic existence; that the universal, substantially one with or distinct from its manifestations, causes verbal expression; that if existence is a reason, existence is also the predicate to be established; that mere non-apprehension in the treatises establishes nonexistence; that a verbal expression can establish the phenomenon expressed; and the Mīmāṃsā argument that the Vedas, not produced by a person and infallible in one area (ritual procedure), are wholly authoritative.

Chapter 2: Establishment of Valid Cognition (ཚད་མར་གྲུབ་པ་, pramāṇasiddhi)

285 stanzas. The “Establishment of Buddha as Valid” — a connected demonstration that the Buddha is a valid person (ཚད་མའི་སྐྱེས་བུ་, pramāṇabhūta), built as an exegesis of the five epithets in Dignāga’s opening homage in the pramanasamuccaya (valid, became, resolved to work for the benefit of others, Teacher, Sugata, Protector). The chapter supplied the Buddhist tradition with its standing reply to Hindu criticism (dreyfus-recognizing-reality, p. 20) and the soteriological scaffolding of pramāṇa as a system. Gorampa structures the explication around two macro-demonstrations: a forward sequence (compassion → training → “Teacher” → “Sugata” → “Protector”; PV 2.29–146) and a reverse sequence that starts from the Buddha’s errorless teaching of the four truths and works back (“Protector” → “Sugata” → “Teacher” → “compassion”; PV 2.147–283).

The definition of validity (PV 2.1–7). The chapter opens with what Gorampa calls “the exclusive Sakya way” of reading Dharmakīrti’s definition (§33). PV 2.1: “Validity is a nondeceptive consciousness existing through having the capacity to perform a function; it is undeceiving.” Gorampa decomposes nondeceptiveness into three features — of the object (what it is undeceiving toward: an object to be engaged or isolated), of the agent (a cognition that “embraces its phenomenon exactly as it exists”), and of the action (nondeceptive toward an intrinsically existing phenomenon as having causal capacity, and toward a non-intrinsically existing phenomenon as lacking it). He then turns to three putative faults of the definition: (i) nonpervasion — cognition arising from sound is undeceiving even though no direct apprehension occurs, because it reveals the speaker’s intention (PV 2.1cd–2); (ii) excessive pervasion — conventional cognition (e.g. a pot-universal induced by direct perception of a pot) is not asserted as undeceiving toward the pot, because it merely re-apprehends what was already apprehended (PV 2.3); (iii) impossibility — against the Vaibhāṣika who places consciousness on the material sense power, Dharmakīrti insists that “for engaging those things to be developed and those to be discarded, consciousness is paramount” (PV 2.3cd–4). The division of cognitions into those self-ascertained (two object-knowers, two self-knowers, and inference [Gorampa: “inferential cognition”, རྗེས་དཔག་]) and those ascertained from others (initial, unengaged, and error-cause cognition) at PV 2.4d–5a is read by Gorampa, following Jamyang sakya-pandita, as operating exclusively at the level of wholly designated valid cognitions: he distinguishes the “consciousness” feature as an innate characteristic (རང་གི་མཚན་ཉིད་), ascertained by self-awareness (Gorampa: “self-knowing”, རང་རིག་), from nondeceptiveness as a cognitive designation determined by subsequent ascertaining cognition (Gorampa: “subsequent ascertaining consciousness”, ངེས་ཤེས་). Dignāga’s other definition — “revealing unknown phenomena” — Gorampa treats as a synonymous formulation of the same definition (PV 2.5d–6), against Devendrabuddhi (two distinct definitions) and the Ornament commentaries (two definienda). The unit closes at PV 2.7: “Possessing that, the Bhagavan is valid.”

Refutation of self-arisen and permanent validity (PV 2.8–28). A focused polemic against the Naiyāyika and Vaiśeṣika doctrine of Īśvara as a permanent self-arisen pramāṇa, and against any account that grounds validity in a permanent substance. The argument is that what is permanent cannot perform a function — and so cannot meet the definition of validity at PV 2.1 — and that what is self-arisen cannot enter relations of cause and effect with anything else.

The Mīmāṃsaka objection and the valid person (PV 2.29–33). PV 2.29 voices the Jaiminīya: a valid person would need to know all hidden phenomena, no such person exists, no cause accomplishes such knowledge, and so there is no point in striving for it. Dharmakīrti’s reply turns the question. Discerning people with doubts will seek out one possessed of the knowledge required — not omniscience as such, but the knowledge of “what is to be developed and discarded, together with methods and causes” (PV 2.32). “Knowing how many insects there are is of no use whatsoever for me” (PV 2.31). The valid person is the knower of the four truths, not the seer of all hidden phenomena. PV 2.33 caps the argument with one of the chapter’s most cited lines: “If seeing far makes for a valid being, come and devote yourselves to vultures!”

Compassion as accomplisher and the rebirth proof (PV 2.34–119). PV 2.34 is the pivot of the forward sequence: “Great compassion is the accomplisher, and that comes from training.” Training in compassion requires many lives, so the entire next block (Gorampa §36, 2.34d–119, the chapter’s longest single section) is given over to the refutation of Cārvāka/Lokāyata materialism and the establishment of past and future lives. The strategy is to deny that the body is the cause of mind. Gorampa walks the reader through a sequence of absurd consequences: if mind arose from a body composed of the elements, “there would be no place where heat and moisture would not produce living beings” (PV 2.37); if the body with sense faculties were the direct cause of mind, damage to any single sense faculty would directly damage mental cognition, which it does not (PV 2.39) — moreover, when the mind is grieved the sense faculties are seen to change, indicating the reverse dependency (PV 2.40–41); if the body is permanent (the Lokāyata thesis) it cannot stand in succession with successive cognitions, so the body must itself be impermanent and successive (PV 2.43–44), which yields a continuum that can be traced past and future. The proof has subsidiary arguments against the body as material cause of mind (PV 2.60ff.), as support (2.63ff.), as direct producer (2.73ff.), and as nature of mind (2.79ff.).

Training admits unlimited increase (PV 2.120–131). Against the objection that repeated practice cannot transcend its own nature (one can only jump so high, water can only be heated to boiling), Dharmakīrti distinguishes physical from mental qualities: the body’s strength bounds physical practice, but compassion, once accomplished, “operates naturally on its own, like fire with wood and quicksilver with gold.” Mental qualities arise from homogeneous causes (mind → mind) and so admit limitless increase.

Compassion → Teacher → Sugata → Protector (PV 2.132–146). The forward sequence completes. From the trained compassion arises the Teacher — one trained in the methods, knowing no-self as the antidote to suffering and analysing the cause, nature, and antidote of suffering (2.132–138). From the Teacher arises the Sugata (2.139c), and Gorampa specifies the threefold force of “Sugata” as an abandonment that is well abandoned (arising from seeing no-self, distinguishing the Buddha from non-Buddhist freedom from desire), never to return (the seeds of the view of self uprooted, distinguishing from the lower paths), and completely abandoned (imprints of mental afflictions eradicated, distinguishing from śrāvaka and pratyekabuddha arhats). From this comes the Protector (2.145c) — one who teaches the path without error.

The four truths — reverse sequence (PV 2.147–279). With “Protector” established by the fact of errorless teaching of the four truths, Dharmakīrti now establishes the four truths themselves in detail. Suffering (2.147–179): the appropriated aggregates are saṃsāric, established as continuous past and future by manifestation-through-habituation (a newborn’s manifest desire requires prior habituation) and by material-cause homogeneity; the four aspects of impermanence, suffering, empty, and no-self are laid out. Origin (2.180–190): craving is the chief cause; the alternatives — causeless arising, a permanent cause, Īśvara — are refuted. Cessation (2.191–205): freedom is established; the conventional vs. ultimate status of the self that traverses bondage and liberation is examined (Gorampa flags here that all Buddhist tenet-holders except the Sāṃmitīya assert the self is not substantial while the aggregates are, and that “these points are essential for the philosophical system of Pramāṇa and Madhyamaka combined”). Path (2.206–279): no-self as antidote — the canonical PV 2.208–209, “the nature of mind is luminous and the defilements are incidental,” read by Gorampa as the strong claim that defilements never had power to harm mind by way of its nature; love is not the antidote to anger because they share the same root, ignorance, which is the view of the transitory collection (§44); the long refutation of ritual paths follows — Īśvara-initiation, mantra-consecration, five-fire reliance, ascetic mortification all fail because body in the sense of “self’s body” is not the cause of suffering, and abandoning attachment to it without applying the antidote of refuting permanent self is “an arduous waste of time”; the chapter also refutes the permanent self of the Sāṃkhya and the “inexpressible self” of the Pudgalavādins.

The reverse sequence completed and the purpose of the praise (PV 2.281–285). With “Protector” established by the four-truths argument, Gorampa supplies a second gloss on “Sugata” — wisdom that knows reality as it is, is stable, and knows the features of all phenomena — reconciling this with the abandonment sense given in the forward sequence (PV 2.281). The chain runs back through “Teacher” (2.282) and “compassion” (2.282). PV 2.283: “Therefore he is a valid being.” The chapter’s closing verses (2.283–285) state the doctrinal point of the entire enterprise: praising the Buddha as valid serves a purpose, “to know that the presentation of validity can be established from the very teachings of the Bhagavan Buddha. It is not that Dignāga invented it.” The Buddha himself taught syllogisms — “whatever has a nature of the slightest production is wholly subject to cessation” (PV 2.285), smoke is known from fire — and so pramāṇa as a system is established by his own teaching, not as an external scholastic imposition.

Chapter 3: Perception (མངོན་སུམ་, pratyakṣa)

Awaiting commentary ingest — current treatment is provisional, drawn from dreyfus-recognizing-reality and westerhoff-dignaga-dharmakirti only.

Covers the definition and types of perception and the arguments establishing it as a valid means of cognition. Contains the canonical Dharmakīrtian definition of existence by causal efficacy at PV 3.3 — “Whatever has causal powers, that really exists in this context. Anything else is declared to be just conventionally existent” (westerhoff-dignaga-dharmakirti, p. 231) — and is the only chapter in which Dharmakīrti consistently adopts his ultimate idealist (Yogācāra) perspective, discussing the identity of perceiver and perceived (Westerhoff p. 255; dreyfus-recognizing-reality ch. 27 on Yogācāra elements in PV III). gorampa-pramanavartika does not cover this chapter.

Chapter 4: Inference for Others (གཞན་དོན་རྗེས་དཔག་, parārthānumāna)

Awaiting commentary ingest — current treatment is a stub.

Deals with the formal presentation of inferences for the purpose of convincing others — debate proper, as distinct from the private reasoning of Chapter 1. Dharmakīrti expands this material in his Vādanyāya (Principles of Debate).

Tibetan Reception

Before sakya-pandita’s promotion of this text, the Pramāṇaviniścaya (Ascertainment) had been the primary text for Tibetan epistemological studies. Sakya Paṇḍita’s decisive influence established the Pramāṇavārttika as predominant. He is said to have taught it daily and is the source of all its lineages in Tibet (dreyfus-recognizing-reality, p. 23). His revised translation (with Śākyaśrī) became the standard, and his pramanayuktanidhi supplied the lens through which the work was henceforth read in the Sakya tradition.

After Sakya Paṇḍita, Uyuk Rikpa Sengé (d. 1235) propagated the text at Sakya Monastery. In the fifteenth century gorampa’s teacher Rongtön Śākya Gyaltsen (1362–1449) and Gorampa himself (1429–89) consolidated the Sakya reading; Gorampa composed his Light of Samantabhadra in 1474. In the early Geluk, Gyaltsab Darma Rinchen (1364–1432), Khedrup Gelek Palsang (1385–1438), and the First Dalai Lama — all of whom had Sakya training — produced major commentaries. Tsongkhapa himself wrote little on pramāṇa but is reported to have wept on reading the Pramāṇavārttika (gorampa-pramanavartika, translator’s introduction).

Dreyfus uses “Commentary” as his abbreviation for this text throughout dreyfus-recognizing-reality.

The Sliding Scales

Westerhoff argues that the Pramāṇavārttika is best understood through the lens of Dharmakīrti’s “sliding scales of analysis” — four ascending levels of philosophical sophistication (ordinary → Abhidharma → particularism → idealism). Most of the text argues from the particularist (Sautrāntika) level, which Dharmakīrti knows to be provisional, to maximise shared assumptions with non-Buddhist interlocutors (westerhoff-dignaga-dharmakirti, pp. 253–256).

Dreyfus arrives at a similar reading independently and uses Sākya Chok-den’s framing: the system uses ascending scales (False Aspectarian Yogācāra-Madhyamaka being the apex), and Dharmakīrti’s apparent oscillation between Sautrāntika and Yogācāra commitments is strategic rather than confused (dreyfus-recognizing-reality, pp. 59, 435–37). Dreyfus’s distinctive twist is that he reads the scale as epistemological rather than soteriological in its primary intent: Dharmakīrti introduces Yogācāra moves to solve epistemological problems (e.g. how perception relates to external objects via aspects), not primarily to lead the reader toward nondual gnosis (pp. 437–38).

Dreyfus also flags an evolution within Dharmakīrti’s corpus that Tibetan commentators tend to ignore: the PV establishes impermanence by the “inference from disintegration” (vināśitvānumāna), whereas later texts (Pramāṇaviniścaya, Nyāyabindu, and exclusively Vādanyāya) use the “inference from existence” (sattvānumāna). Both reasonings reach the same conclusion but reflect different polemical pressures from the Nyāya school (dreyfus-recognizing-reality, pp. 65–66, following Steinkellner and Mimaki).

Mipham connects this strategy to shantarakshita’s Yogācāra-Madhyamaka synthesis: the difficulties of the aspect theory (sākāravāda) forced Dharmakīrti to shift from a Sautrāntika to a Yogācāra framework, mirroring the historical development from Bhāviveka to Śāntarakṣita (shantarakshita-madhyamakalankara-intro, pp. 26–27).

Pramāṇa and Madhyamaka

The logico-epistemological tradition was “primarily and intimately linked” with Svātantrika Madhyamaka from its inception (shantarakshita-madhyamakalankara-intro, p. 10). Candrakīrti rejected the tradition of Dignāga as antithetical to Madhyamaka insight, but his distrust was not shared by most Tibetans. Tsongkhapa’s paradoxical adoption of pramāṇa within a Prāsaṅgika framework — applying the logical method to a view that had traditionally rejected it — was “censured with great severity” by Sa-gya critics (shantarakshita-madhyamakalankara-intro, p. 16).

Key passages

The verses most often invoked elsewhere in the wiki and in the literature, gathered here for cross-reference. Their argumentative role is described in the chapter walk-through above.

• PV 1.40 (apoha, the canonical statement) — “All entities dwell by nature in their own individual entities, they are the support of being isolated from entities, both compatible and other.”
• PV 1.226 (against Vedic authority) — “The cause of understanding meaning is linguistic convention, which depends upon people. Thus, even if the words were not produced by a person, the possibility of error remains.”
• PV 1.230 (“washing an elephant”) — “Not being composed by a person, there is no meaning in any way. If you accept that it was composed, that is actually like washing an elephant.”
• PV 2.1 (definition of validity) — “Validity is a nondeceptive consciousness existing through having the capacity to perform a function; it is undeceiving.”
• PV 2.32 (the valid person) — “One who knows the realities of what is to be developed and discarded, together with methods and causes, is asserted to be valid; one who knows all things is not.”
• PV 2.33d (against far-sight) — “If seeing far makes for a valid being, come and devote yourselves to vultures!”
• PV 2.34 (the pivot) — “Great compassion is the accomplisher, and that comes from training.”
• PV 2.208–209 (luminous mind) — “The nature of mind is luminous, and the defilements are incidental.”
• PV 2.283–284 (purpose of the praise) — “It is to know that the presentation of validity can be established from the very teachings of the Bhagavan Buddha. It is not that Dignāga invented it.”
• PV 3.3 (causal efficacy) — “Whatever has causal powers, that really exists in this context. Anything else is declared to be just conventionally existent.”

Sources

• dreyfus-recognizing-reality — the central subject of the entire book; especially Ch. 2 on momentariness and the evolution of arguments for it, Ch. 16 on the twin definitions of pramāṇa in PV I and PVin, Ch. 27 on Yogācāra elements in PV III, and pp. 437–42 on the soteriological reading of PV II
• westerhoff-dignaga-dharmakirti — PV 3:3 on causal efficacy, sliding-scales model, Ch. 2 on Buddha’s authority, yogic perception, debate with Mīmāṃsā
• gorampa-pramanavartika — Gorampa’s Light of Samantabhadra (Kilty trans. 2022); running scholastic commentary on the 625 stanzas of PV Chs. 1–2, including the long translator’s introduction on the history of pramāṇa in India and Tibet, the chapter-order debate, the threefold definition of validity, and the forward/reverse sequences of Ch. 2
• perdue-buddhist-reasoning-debate — Geluk bsdus grwa pedagogical uptake of PV: the seven-fold gradation of cognitions (Ch. 12, drawing on PV I/III), Sautrāntika ontology, the three modes of a correct sign, “Because objects of comprehension are two, valid cognizers are two” cited from PV III (p. 314), the soteriological frame of PV Ch. 2 as the “valid cognizer” division of person/scripture/mind (p. 317)