Chapter 1: Perception
Chapter 1 argues that there are exactly two means of valid cognition because there are exactly two kinds of object: the own-character, which perception takes, and the universal-character, which inference takes. Genuine perception must be free of the conjunction of name and class, and divides into sense, mental, self-aware and yogic; pseudo-perception has seven kinds, six of them conceptual. The chapter ends with three successive accounts of the result of valid cognition, from which Dignāga derives the division into object-aspect, subject-aspect and self-awareness-aspect, and then refutes the accounts of perception given by the Vādavidhi, the Nyāya, the Vaiśeṣika, the Sāṃkhya and the Mīmāṃsā. The prefatory material of the Chinese edition — the publisher’s account of the text’s transmission, a life of Dignāga, and Fazun’s own preface — is kept here at the head of the chapter.
Condensed from Fazun’s Chinese translation-and-exposition of the Pramāṇasamuccaya and its auto-commentary, and rendered into English. Root verses are given as Fazun gives them; the prose beneath summarises his exposition. On the provenance and the limits of this rendering, see the series introduction.
Chapters: 1. Perception · 2. Inference for Oneself · 3. Inference for Others · 4. Example and Pseudo-example · 5. Apoha · 6. Sham Rebuttals
Publisher’s Preface
The Pramāṇasamuccaya is Dignāga’s major work (c. 5th–6th century) and one of the eleven foundational treatises of the Yogācāra school. The Sanskrit is lost; Yijing’s Tang translation (711) was lost soon after; only the two Tibetan translations survive in the Tengyur — Vasudhararakṣita with Zhā-ma seng-ge (Xin-hui), and Dānaśīla with Ma-gzhis dge-ba’i blo-gros (Yamashen-jia). Modern partial work includes Lü Cheng’s Abridged Extracts (1928), Randle’s Fragments (1926), Iyengar’s Sanskrit restoration of chapter 1, Jambuvijaya’s restoration of about a third, and studies by Hattori, Kitagawa, Takemura, Miyasaka and Frauwallner. No complete Japanese translation exists.
Each of the six chapters has two halves: establishment of one’s own position and refutation of rival views. The first half is the primary source for Dignāga’s logic and fills gaps in the surviving Chinese material; the second preserves the doctrines of the Vādavidhi, Nyāya, Vaiśeṣika, Sāṃkhya and Mīmāṃsā.
Chapter 1 covers perception. Its first half argues that there are only two pramāṇas because there are only two kinds of object: the own-character (svalakṣaṇa), the object of perception, and the universal-character (sāmānyalakṣaṇa), the object of inference. Scripture (śabda) and analogy (upamāna) are nominal pramāṇas only. True perception must be free from the conjunction of name and class, i.e. wholly free of concept, and is fourfold: sense, mental, self-aware, yogic. Pseudo-perception is sevenfold — six conceptual, one non-conceptual but lacking a real entity. The chapter closes with three successive accounts of the pramāṇa-phala, from which Dignāga derives the threefold division into object-aspect, subject-aspect and self-awareness-aspect: one of his major advances over earlier Buddhist epistemology. The second half refutes the five rival accounts of perception.
Fazun worked from the Derge Dānaśīla–Yamashen translation, collated against the Peking Vasudhararakṣita–Xin-hui version, and annotated throughout with reference to the Nyāyamukha. He completed the work in his eightieth year, with failing eyesight, and was not himself satisfied with it.
Preface to the Abridged Exposition
Dignāga was one of Vasubandhu’s four great disciples, and his work in hetuvidyā surpassed his teacher’s; post-Dignāga logic is accordingly called the “new hetuvidyā.”
Born to a South Indian brahmin family, he first studied under the Vātsīputrīya teacher Nāgadatta, who set him to search for the “inexpressible self distinct from the aggregates.” He practised, found nothing, and — suspecting some hindrance — took to exposing his body and staring in all directions by sunlight and lamplight. Questioned, he reported his failure; the report was in fact a quiet exposure of the doctrine’s vagueness. Expelled for attacking his own school, he declined to refute his teacher out of propriety and went to Vasubandhu, studying the Mahāyāna and Maitreya teachings and excelling in logic. Tāranātha reports he could recite five hundred works from memory.
Summoned from a cave retreat at Odiviṣa to Nālandā, he thrice routed the tīrthika disputants. Tradition holds that when he chalked the opening homage-verse on his cave wall, light blazed and the earth shook; a jealous tīrthika, Kṛṣṇa-mūni-rāja, rubbed it out three times, and after a debate in which Dignāga defeated him three times over the man still refused to concede. Discouraged, Dignāga received a revelation from Mañjuśrī urging him to compose the treatise, and vowed to do so.
He wrote the root verses first, then the auto-commentary (vṛtti). Indian successors include Dharmakīrti’s Seven Treatises, Śubhagupta’s Bahirarthasiddhi and Sarvajñasiddhi, and Dharmottara’s Kṣaṇabhaṅgasiddhi and Apohasiddhi. In Tibet, where the work is called “the Pramāṇa-Sūtra” (Tshad-ma’i mdo), commentaries were written by Sakya Paṇḍita and Tsongkhapa, and specifically on this text by Gyaltsab Darma Rinchen (Tshad-ma’i mdo’i rnam-bshad).
Author’s Preface [Fazun]
Written at the urging of friends studying hetuvidyā, without oral instruction on the treatise and with very few commentaries available. Fazun presents it as a draft, not a definitive version.
Abridged Exposition of the Pramāṇasamuccaya
- Composed by Bodhisattva Dignāga
- Translated and edited by the Śramaṇa Fazun
Opening homage
Homage to the Teacher, the Sugata, the Protector — the one who, intent on benefiting beings, has become the authoritative means of knowledge; in order to establish pramāṇa, from my own treatises I have gathered the scattered teachings and compiled them into one.
Dignāga pays homage to the Buddha as pramāṇa-bhūta, “become the authoritative means of knowledge,” through the gateway of his perfections in cause and in fruit. Causal perfection covers motivation (compassion, bodhicitta) and application (teaching the true Dharma). Resultant perfection covers benefit for oneself — attaining “Sugata,” in three senses: supremely good, like a man of refined bearing; permanently non-regressing, like a plague removed for good; and complete without remainder, like a jewel-vase filled to the brim. These three respectively surpass the tīrthikas’ freedom from desire, the learner’s and the non-learner’s newly attained self-benefit. Benefit for others is teaching according to capacity and leading beings to nirvāṇa.
The second half of the verse states the vow to compose: to establish genuine pramāṇa, Dignāga gathers the scattered statements of his own earlier works — the Nyāyamukha and others — into one place.
Chapter 1 — Perception (Pratyakṣa)
Two parts: establishment of one’s own position, then refutation of rival positions. The first divides into the general exposition of the two pramāṇas (positive statement, then resolution of objections) and the special exposition of perception.
(1) The positive statement
Perception and inference are the pramāṇas; the two aspects [of the object] are the prameya.
Knowing the object of knowledge depends on the means of knowing, so the means must be settled first. There are exactly two: perception and inference. Scripture and analogy are nominal pramāṇas, not genuine ones.
The count is fixed by the object, not by the subject: the prameya has only two aspects, own-character (svalakṣaṇa) and universal-character (sāmānyalakṣaṇa). The cognition whose appearing object is the own-character is perception; the cognition whose appearing object is the universal-character is inference. There is no third aspect of the knowable, hence no third pramāṇa. This is the structural argument on which the whole chapter rests.
(2) Resolution of objections
Because they consist in combining [cognitions], there is no other pramāṇa. Nor is it [a separate pramāṇa] to know repeatedly, for [if so] there would be infinite regress, as with memory and the like.
[Objection] What of cognitions that combine — grasping form and then judging “this is impermanent” — or of cognising the same thing again? Are these further pramāṇas?
Reply: No, because they cognise the same prameya already covered. Grasping the own-character is perception; reasoning out “form is impermanent” is inference. Merely tying the two together in mental consciousness without reaching a settled cognition is neither, and so is not a pramāṇa at all.
Repeated cognition likewise: if each re-cognition of one object counted, pramāṇas would multiply endlessly. The criterion is novelty — where there is no new cognition there is no pramāṇa. This is why later recollection of an object one previously desired or hated is not a pramāṇa.
B. Special exposition of perception
Three parts: true perception, pseudo-perception, and the result of pramāṇa.
(1) True perception
(a) The general mark
Perception is free from conceptual construction associated with name, class, etc.
Freedom from concept is the general mark, but the implication runs one way only: all perception is non-conceptual, while not every non-conceptual cognition is perception. The eye’s cognition of a double moon under cataract is non-conceptual yet is pseudo-perception.
The concepts excluded are those that fuse the bare object with a word. Fazun lists five kinds by the type of word: arbitrary names (“Devadatta,” “Yajñadatta”), class-words (“ox,” “sheep”), quality-words (“blue,” “white”), action-words (“offering,” “giving”), and substance-words naming a property-bearer (“stick-bearer,” “horned one”). Every cognition produced through words of these kinds is conceptual and therefore not perception. Dignāga defines here by exclusion only, and does not yet say from what kind of object perception arises.
(b) The kinds
Because it is the distinctive cause, the cognition is named after the faculty. Because it arises from many objects, its object-domain is the aggregate of its own sense-objects. A bearer of many natures is not what the faculty cognises. What is cognised by it is not expressible; its own-character alone is the object of the faculty.
[Objection] Sense consciousness arises from two conditions, faculty and object. Why then name it “eye-consciousness” after the faculty rather than “colour-consciousness” after the object?
Reply: Because the faculty is the asādhāraṇa (distinctive, uncommon) cause. The colour is a shared cause — the same patch can be seen by another person’s eye-consciousness and cognised by one’s own mental consciousness — whereas my eye faculty produces only my eye-consciousness. Worldly naming follows the same rule: “drum-sound,” “barley-sprout.” Fazun cites the Abhidharma in support: “Through accomplishment of the eye-consciousness one knows ‘blue’; one does not think ‘this is blue.’ In the object one has the notion of the object, not in the object the notion of ‘dharma’.”
[Objection] If perception is wholly non-conceptual, how can the five consciousnesses cognise aggregates, and how can they be said to take the own-character of their sphere?
Reply: They arise from many objects — the union of many atoms forming the object-support — yet take their own sphere as object under a general form. Eye-consciousness takes the form-sphere, not the individual atom. Two things follow. The property-bearer compounded of colour-, smell-, taste- and touch-atoms is not what a sense faculty cognises: the faculty does not take what is different as non-different. And what it does cognise cannot be put into words, since it is the bare own-character of that sense-base stripped of verbalising construction. Fazun refers the fuller treatment to the Pramāṇavārttika.
Mental [perception] too takes the object. [Mental factors such as] desire etc., and self-awareness, are non-conceptual. What the yogin is taught, unmixed [with verbal construction, and seen in] its mere object [as such], — even conceptual cognition is [thus] self-awareness, though not with respect to its [external] object, because that is different.
Mental perception. The sixth consciousness also takes objects such as colour, operating with a receptive mode and without concept, and so is perception. Dignāga’s account is very brief; Fazun fills it out from Dharmakīrti, who restricted mental perception to the single moment of mental consciousness arising immediately after the last moment of sense-consciousness and cognising the same colour. Later moments cannot take colour directly and are therefore conceptual and not perception. Dharmakīrti also denied that two cognitions of one object and one aspect can arise simultaneously, which removes any “concomitant mental consciousness” running alongside the five sense consciousnesses; two cognitions may coexist only if object and aspect both differ.
Self-awareness. Desire, hatred, delusion, pain and pleasure are not obtained through any faculty, so their cognition is not sense-perception; it is perception by self-awareness.
Yogic perception. The practitioner who does not mix in the teacher’s verbal instruction sees the principle alone, and this too is perception.
[Objection] If the self-cognising aspect of desire is perception, must conceptual cognition also be perception?
Reply: Yes as to its own nature, no as to its object. Conceptual cognition is self-aware perception with respect to itself, and conceptual with respect to what it intends. The two are compatible because they concern different things: being conceptual towards the object does not block self-aware perception of its own nature, and self-awareness of its own nature does not make it non-conceptual towards its object-sphere. This is the move that lets Dignāga admit self-awareness universally without making every thought a perception.
(2) Pseudo-perception
Deluded cognition, conventional cognition, inference and what arises from inference, memory and expectation, and [non-conceptual cognition] with cataracts — [these are] pseudo-perception.
Seven kinds; the first six are conceptual, the seventh non-conceptual.
- Deluded cognition — taking a mirage for water.
- Conventional cognition — seeing a jar or cloth and supposing a real jar-substance. Jar and cloth are posited by conceptual cognition as the socially agreed referents of words; in the own-character of the substance there is no “jar.” Conventionally such cognitions are correct knowledge, not errors; measured against substance they are illusory, conceptual, and so pseudo-perception. 3–4. Inference and post-inferential cognition — both discriminate on the basis of a previously received object. 5–6. Memory and expectation — memory takes the past, expectation imagines the future; neither has a real present object.
- Cataract-cognition — sky-flowers, hair-circles, a double moon. Non-conceptual, since sense-consciousness does not cling to words, but with no real entity, hence still pseudo-perception.
The label “pseudo-perception” applies to all seven. Reasons for the list are given at length in the Pramāṇavārttika.
(3) The result of the means of knowledge (pramāṇa-phala)
Since there is a confirming function, the result itself is what measures. It is likewise posited as the measuring instrument, and yet is not without [instrumental] function. Moreover, self-awareness is the result, because by it the own-nature of the object is determined. The object-image is the measuring; by it the measuring is done. One says: when the [cognition’s] appearance arises, the object-measured, the means, and the result — because of the grasping and the knowing — are not three separate things. Knowing of the object and knowing of that [knowing] are distinct; therefore cognition has a twofold aspect. By subsequent recollection too, the twofold aspect and self-awareness are established, for if not experienced, it would not [be remembered thus]. If experienced by another cognition, infinite regress — and that would be remembered too. Likewise, [the other cognition] does not turn toward a different object, [only] seeing this one.
Against the tīrthikas, the result is not something separate from the act of cognising: the measuring cognition is itself the result. As when one fells a tree with an axe, the tree’s being cut is the result, and there is no additional “cutting-result” standing apart from the tree.
Three successive accounts are given, each refining the last.
-
First account. The object is the prameya; the cognition that measures it is the pramāṇa; the cognition’s function of confirming the object is the result. The cognition arises bearing the object-image and carries the confirming function, so it counts as the instrument and yet is not without instrumental function. By simile: an effect is said to bear the form of its cause, as a son resembles his father. (Some editions read “not without difference,” on which the point becomes that instrument and result are not wholly undifferentiated.) Khenchen Kunga Wangchuk’s annotation commentary (ཚད་མ་ཀུན་བཏུས་ཀྱི་མཆན་འགྲེལ་ཚིག་དོན་གསལ་བྱེད་) supports the “not without difference” variant — བློ་ངོ་བོ་གཅིག་ལ་དོན་རྟོགས་པ་དང་དོན་དང་འདྲ་བའི་དབྱེ་བ་མེད་པའང་མ་ཡིན་ནོ།, “nor, within one cognition, is there no distinction between cognising the object and resembling the object.” The annotation divides the section exactly as Fazun does: conventional positing of object-cognition as instrument and result, then self-awareness as the result on analysis, then the establishing of that.
-
Second account. Self-awareness is the result; the cognition’s image is the prameya and its object-mode is the pramāṇa. Since the object is not apart from cognition, the cognition by its own essence determines the object-sense. When it bears the object-image, the self-awareness following on it is what knows whether the experience is pleasant or unpleasant. Where an external object is taken as prameya, the cognition bearing its image, though in itself self-awareness, counts as the instrument. Black, white and the rest appear within the cognition, and it measures by bearing that image. On this account one and the same appearance can be called both prameya and pramāṇa: self-awareness arises from two images, the object-image being the prameya, the own-image the instrument, and self-awareness the result.
-
Third account. The object-aspect (nirbhāsa/ākāra) is the prameya, the grasping-aspect (grāhaka-ākāra) is the instrument, and the knowing is the result. The three are one entity, posited as three only from different points of view. This is the basis of the later threefold division into object-aspect, subject-aspect and self-awareness-aspect.
Proof that cognition has two aspects. [Objection] How is this known? Reply: by the distinction between knowing the object and knowing that knowing — the consciousness conforming to the object-image and the consciousness manifesting the own-image. Without a self-cognising aspect, saying that form “cognises itself” would be indistinguishable from cognising an object. Saying it is known by a later cognition fails, because a long-past object should then not appear, not being present.
Proof from memory. One can later remember not only the object but the cognition of it, which shows the cognition was itself experienced at the time; what was never experienced could not be remembered. Memory of form presupposes prior experience of form, and the same holds for the cognition.
Against experience by a further cognition. [Objection] Form is experienced by another consciousness; so let consciousness be experienced by another consciousness.
Reply: Infinite regress. The second consciousness would need a third, and so on without end, whether the further consciousness is taken as simultaneous or subsequent. Three further faults follow: one would have to find later memories of those further consciousnesses too, which we do not have; the further consciousness, occupied solely with the prior one, could take no other object; and no term is ever reached. Self-awareness must therefore be admitted, and with it the status of pramāṇa as result. Perception is free from conceptual construction.
Refutation of rival positions
Five divisions: the Vādavidhi, the Nyāya, the Vaiśeṣika, the Sāṃkhya, and the Aupavarṣa/Mīmāṃsā.
1. Refutation of the Vādavidhi
The Vādavidhi is not his [Vasubandhu’s] composition; he held it to contain no essential meaning. [As for] the rest, they should be expounded with divisions — therefore I shall examine it. It says: “The consciousness arising from that object is perception.” The phrase “that object” if taken universally — but [perception] is not only from it. If “that object” means “the object-support” — [then] memory and the like, the cognitions [of them], do not depend on another [object].
Further:
As what appears is not existent, therefore in the ultimate sense — since the five [sense-consciousnesses] take aggregates as their condition — no verbal designation is made [to them]. This is an intermediate verse. Because they are remote from the verbalisable aspect of the object, its object is not expressible; since [expression proceeds] by way of the universal, no verbal designation is made.
Dignāga opens by disowning the text’s authority: the Vādavidhi is not Vasubandhu’s own composition, and the master, holding it to contain no essential doctrine, never expounded it. Since he set it aside, Dignāga will examine its definition of perception himself. (Fazun leaves the line “the rest should be expounded with divisions” unexplicated, marking it as needing further study.)
The definition under attack: “The consciousness arising from that object is perception.” Two readings are taken in turn.
First reading — “that object” as a universal-expression (sarvopasaṃhāra). Then any cognition arising with respect to any object would be perception. Too wide: on the Buddhist account mind and mental factors arise from many causes (a variant reads “from four,” i.e. the four conditions), so arising from an object-condition cannot by itself be the mark.
Second reading — “that object” as the object-support (ālambana). Then memory, inference and expectation are excluded, since they arise without a present object-support. Worse, if a cognition of X required X to be present as object-support, one could never cognise fire from smoke, and inference collapses.
Third move — the dilemma of appearance. If they say the object just is what is cognised in the form of colour and the rest, ask: does the consciousness arise as whatever it appears to be, or does it arise from its cause while appearing in a different form?
- If it arises as it appears: what appears is the aggregate, so the five sense consciousnesses would cognise only conventional existents, seeing aggregated form and never real atoms. And if one admits genuine entities with real aspects, then since substance-aspects appear, one would be forced to admit substance, number and the rest as perceived — the Vaiśeṣika categories smuggled in.
- If it arises in accordance with its cause without appearing in that form: the fault of substance-appearance is avoided, but nothing then appears corresponding to the atoms, so they cannot be called perceived at all, none being cognisable separately. Each atom is a cause of the aggregate, and the aggregate is not admitted to be substantially real.
Hence the summary verse. The Vādavidhi’s object is either taken as mind takes it — an aggregate-appearance, not ultimately existent — or as real substance, which the mind does not bear the form of; on neither reading can the word “perception” be applied. A further consequence: were mere causal role enough, the non-existent second moon would count as the cause of a cognition. Verbal designation in this sphere proceeds only through the universal-character, never through the own-character, so the objects of the five sense consciousnesses are named “form” and so on by their common nature and are in themselves inexpressible.
(Fazun records variants: the commentary renders the third line “what the mind cognises is of five kinds,” and Xin-hui’s Tibetan translation gives “If [one cognises] not in the aspect as it appears, then since the five take aggregates as their condition, from what [can they be said to arise] in the ultimate sense? They make no verbal designation.”)
2. Refutation of the Nyāya position on perception
[Nyāya] “The cognition that arises from the contact of the faculty with the object, which does not employ verbal designation, is non-erroneous, and whose nature is determination (vyavasāyātmaka), is perception.” Four marks: born of faculty–object contact; free of verbal designation; non-erroneous; of the nature of determination.
That too is untenable. In cognition born of the faculty, verbal designation etc. are absent [and so need not be specified]. [On their account,] it could not cognise what is at a distance; consciousness could not take what is larger [than the faculty]; and the faculty, not leaving its support, is not able to cognise the object. Pleasure etc. would not be objects-of-knowledge, or else there would have to be another faculty. If one says “there is a mental faculty, obtained because nothing excludes it,” then the word “other [sense-]faculty” becomes meaningless. If definition [means the nature of perception is determination], then there is no result; or it must be said [the result is] distinct from the cognition. [If] another object is the result — no, for the distinct are different things. Nor is it both [means and result]; nor is the qualified [similar]. “Not-knowing etc.” are not universal; since there is no [such] reversal, [that] is not the result.
“That too is untenable” is the general verdict; each mark then fails on its own.
“Free of verbal designation” is idle. Only inferential consciousness has an object set up by verbal designation. No sense cognition can make verbal designations, so nothing is excluded by the qualifier.
“Non-erroneous” is idle. Error belongs to mental consciousness, its proper locus; and to say a cognition is non-erroneous says no more than that every consciousness takes its own object-sphere.
“Determination” is fatal. Determination is ascertainment, and ascertainment is the pramāṇa-phala — the result reaches further than the instrument, which merely cognises the object while the result ascertains it. If perception is itself ascertainment, no result is left for it to produce. A subsidiary point: if these qualifiers were meant to display the cognition’s own essence rather than to define perception, one would have to go on adding “not produced by quality or substance,” “not the sphere of the non-empty,” and so endlessly.
“Contact of faculty and object” fails on the facts. If perception required full contact, mountains could not be seen, yet distant seeing is directly observed; nor could the eye take an object exceeding its own magnitude. If the faculty is said to travel out to the object, that too fails: the faculty does not leave its support — the eye stays in the body and still sees a distant mountain — and were it to travel, it would have to grasp whatever lay against its seat. Eye and ear reside at their own internal support and take neither distant nor oversized objects by going out.
Pleasure and the rest are left unknowable. The cognition grasping one’s own pleasure, pain, desire, hatred and effort relies on no reason, so it is not inference; it arises from no faculty–object contact, so on the Nyāya definition it is not perception either. If the grasping cognition is no pramāṇa, what it grasps is no prameya — a great fault. The alternative, admitting a further pramāṇa or a further faculty, is equally unwelcome. [Nyāya reply] There is a mental faculty (manas), established because nothing excludes it. [Dignāga] Then “ear faculty” and the rest become empty words, since nothing excludes any faculty either; mere non-exclusion cannot establish anything.
Instrument and result cannot be split by objects. [Nyāya] Let the cognition grasping the universal property be the instrument and the cognition grasping the particular be the result. [Dignāga] No — distinct objects are distinct things. One does not chop the dān-tree and fell the rāṣa-tree; such a thing is never seen. Nor does causing the cognition of the particular make the first the agent of the second, or every agent would collapse into one agent; only where a thing operates upon a thing is the second reasonably called its result.
Nor can one cognition be both. [Nyāya] The cognition of the qualified is both instrument and prameya, as when I know my own “I,” at once object known and knower. [Dignāga] Then the cognition of the qualifier is likewise both; the two must be coupled, and the position gains nothing.
Removal of not-knowing is not the result. [Nyāya] The result is the turning back from not-knowing, doubt and misapprehension. [Dignāga] No cognition dispels these universally; a cognition arises with a moment’s reflection on one class of objects only. And the mere absence of not-knowing is a non-entity, never observed as a prameya, so it cannot be the result. The Nyāya account of perception therefore fails throughout.
3. Refutation of the Vaiśeṣika position on perception
[Vaiśeṣika] “Formed from [the combination of] self, faculty, mind, and object, [perception] is something distinct.” Their texts add: “Born only of [such] conjunction, it is perception with respect to substance,” and “Formed from the conjunction of self, faculty, and object, it is something distinct.” Interpreters divide: some take the “something distinct” from the object measured as the asādhāraṇa cause and make the faculty–object conjunction the pramāṇa; others, on grounds of primacy, make the self–mind conjunction the pramāṇa.
Because it is seen to [cognise only] its object, it does not conjoin with the distinctions. Not one; form etc. are not one. If seen [grasped by another], it is not by the faculty. Since “other faculties” would be pointless, another object would also be grasped. All faculties should grasp everything — but not substance etc. If so: being absent, it is not another’s object. If one admits the objects are the same, non-pleasure too should follow. Indeterminate; said to be other. Not all that is asserted is the thesis. Even where faculties are not different, cognitions differ, so [the objects] differ too; on what basis [should elsewhere there be] non-difference?
Preliminary faults. Their own claim that what is produced by ascertainment is perception and inferential cognition conflicts with the four-factor account, since ascertainment presupposes investigation whereas perception is the bare seeing of the object, in which no inquiry occurs. If faculty–object conjunction is the pramāṇa, then everything contacted would be grasped, since everything contacts in its essence. If self–mind conjunction is the pramāṇa, instrument and result fall into different objects — already refuted against the Nyāya.
The main target is the claim that perception cognises the universal and the particular, and substance, quality and action. It cannot: sense cognition is seen to take only its own object, so the distinguishing properties are not conjoined with it. To judge “this is the universal of that” or “this is conjoined with substance” one must already have cognised two objects and brought them together — an act of recollection, performed by mental consciousness alone. “This sweet scent” cannot be a perception, since scent belongs to the nose and sweetness to the tongue.
One substance grasped by many faculties? Then it is not one — colour and the rest would become many dharmas, and colour has never been seen grasped by many faculties as one. If it is grasped by many yet counted non-different, colour too would have to be one, like substance.
An undifferentiated cognition instead? [Vaiśeṣika] Though faculties differ, they see the substance-object by an undifferentiated cognition, as with existence and quality, so colour is neither one nor many. [Dignāga] Then it is not seen by a faculty at all; one cannot know another faculty’s objects through one’s own gate. Otherwise, if faculties took each other’s objects, positing many faculties for colour and the rest would be pointless — and the eye would have to take tangibles, since difference is admitted as no bar to grasping (blue and its varieties differ, yet one faculty takes them all). Either difference bars grasping, in which case one cannot take differing objects at all, or it does not, in which case every faculty grasps everything.
The appeal to determinateness. [Vaiśeṣika] Each faculty’s object is fixed by its own class-property: what lacks colour-ness is not the eye’s object. [Dignāga] Then substance, number and action, having neither colour-ness nor tangible-ness, are grasped by neither eye nor touch — which contradicts their thesis. Read the other way, the position contradicts their own canon, which says error is absent “due to the absence of that” — the absence of colour in sound — not due to its presence.
How, then, is substance grasped as non-different? By a recollection-cognition arising simultaneously, i.e. mental consciousness, not by sense cognition, which takes each distinct object separately. The cognition that arises towards an object as “non-different” by excluding the other is mental throughout. So cognition of existence and quality is not perception, since they do not appear; to claim perceptual grasp of them out of superiority-conceit is wrong imagination.
If substance were equally the object of all faculties, then since substance, quality and action are substantive like existence, substance would be qualified by all faculties — and being in all, it would be existence, which the Vaiśeṣika denies. Their canon states: “since there is [existence] in action and quality, existence is neither action nor quality.” If existence enters substance, substance is one; if it enters others too, substance is not one — so substance must become different. And when the eye perceptually cognises “not hot,” the tangible becomes the eye’s object.
Plurality from difference of faculty is indeterminate. A single faculty also grasps distinct substance, quality and action, and distinct shades of blue, so being grasped by different faculties does not establish plurality; the inference is not universal. And if difference of cognition alone made objects different, then where the faculty is one and the cognitions differ there would be no point at which anything could be called one — nothing would ever be non-different. The cognition of quality is refuted on the same ground. Their “conjunction of four factors” fares no better, since not every conjunction produces cognition, as shown against the Nyāya. The Vaiśeṣika account is therefore extremely hard to establish. (Fazun gives Xin-hui’s variant rendering of this passage.)
4. Refutation of the Sāṃkhya position on perception
[Sāṃkhya, the followers of Kapila] “The ear, skin, eye, tongue, nose etc., empowered by manas, operate on objects, i.e. in the present cognise sound, tangible, colour, taste, and scent respectively — this is perception.” Their supporting account: sound and the rest differ in kind because the three guṇas — sattva, rajas, tamas — stand in different proportions.
Either infinite regress, or one faculty. Grasped by two, not three objects. Obtaining one object, without distinction, it does not take the own-nature. It does not take the distinction of the object. If it takes it, it would be like the mind — it would also be conceptual. It should be according to that stage, not sattva etc. Not of another nature, therefore. Not-another should be not-result — or if not of another nature, it should not be grasped. The various transformations — their object should be alike. Many are grasped, therefore. The Sāṃkhya has a sub-school that does not accept [this]. All these each-own-natures have a distinct ultimate atom (最勝塵). Through differences of combination the result’s essence is manifested. By what are the natures of the three atoms proved to be one result? If the combination is of disparate kinds, transformation is not admitted. If two natures are not admitted, the result has only one nature. In the many natures of the object, faculty and sphere are distinct. Therefore the Sāṃkhya school is to be abandoned. Each-own-nature is supreme, operating on non-one objects; it is not called the means. Not memory, since not experienced. If the two arise simultaneously, that is not pramāṇa — it looks elsewhere. Dissolved, or memory. Because one speaks predominantly of “memory,” the rest being excluded, [positing] other faculties is meaningless.
The opening dilemma. If the object a faculty operates on is individuated by guṇa-proportion, then a single sound, admitting endless variation in those proportions, would require endlessly many faculties. If instead the guṇas are undifferentiated, one faculty would grasp sound and tangible alike, and all faculties collapse into one — for apart from the three guṇas there is no sound-class at all.
Objects already cross faculties. The form of a mountain is grasped both by the eye and by touch, which breaks the rule that each faculty keeps to its own object. And if form is made a distinct class, ear, nose and tongue cannot reach it, so what we plainly perceive as sound, scent and taste would all be non-perception.
Sameness of form proves too much. Gold bowls and gold ornaments would be undifferentiated; gold, sound and substance would come out the same, and the faculties would no longer each operate on their own object.
The second dilemma: what does a faculty take? Either its own kind alone, or the object as qualified by pleasure and the rest.
- Own kind alone: then it cannot determine the pleasure-nature of sound, nor grasp the object’s body as against a mere appearance of form — and it could not distinguish lute-sound from drum-sound, since these have no difference of form.
- The qualified object: then, like mind operating on its own sphere, the faculty becomes conceptual, which destroys it as perception.
Grasped separately or together? Not separately: the faculties operate on sound and the rest, not on sattva and the rest, and sattva is not a nature other than sound. If sattva is not other than sound, then sound is not its result, sattva is not its cause, and the causal doctrine collapses; also sattva would be of one undifferentiated nature with sound, which is incompatible with positing cause and effect at all. And if pleasure and the rest are not of another nature than sound, they cannot be distinctly grasped — nor can the atoms, nor the five tanmātras — and the faculty would degenerate into a cognition of the universal.
Not together either: if the faculties grasp jointly, all of them operate on every object, the objects become alike, and again one faculty results.
The sub-school. [Ancient Kapila] The three-aspect nature of tangibles is not the same as that of sound, so pleasure and the rest differ, each faculty keeps its own object, and the faculties do not go infinite. [Dignāga] Then each own-nature — pleasure, pain, delusion, sound, tangible, action — has its own distinct supreme atom, and the effect is manifested by differences of combination alone. But conjunction of disparate kinds yields no transformation, so three natures cannot produce one result. And if two natures are not admitted in the effect, the result has only one nature, so among an object’s many natures only the one on which cognition operates is the faculty’s object — a single kind, and the same for tangibles. Either way the position fails.
The experience of the faculty’s operation. If the faculty operates only on the present sound and that is perception, then not all prameyas are covered and there is no pramāṇa of the whole; and the mental faculty that experiences the faculty’s operation is subsumed under no pramāṇa at all. [Sāṃkhya] That experience is memory, a distinction of the operation of perception. [Dignāga] Memory requires prior experience, and the uninterrupted mental consciousness never experienced the operation, nor previously experienced the external object. [Sāṃkhya] Then the two arise simultaneously. [Dignāga] Then neither is a pramāṇa, and both the sense cognition and the mental consciousness experiencing it fall. Only admitting self-awareness avoids the fault. Finally, if mind operates on the external object directly, the other faculties become pointless, since a person’s aim would be accomplished by mind alone. The Sāṃkhya account therefore cannot determinately grasp the object’s own-nature and is not perception.
5. Refutation of the Mīmāṃsā (Aupavarṣa) position on perception
[Mīmāṃsā] “The sense-cognition of a person, arising from a correct conjunction with the existent, is perception.”
[Positing] the negation of the non-existent as “existent,” the conjunction-word cognises that. Correct-conjunction displays what is definitely existent. If it displays what is opposite, then one should state faculty-distinctions. Mere-existence, not unknown — what further is established? If the faculty is admitted to have union-with-blessing toward something, union is had also elsewhere; collyrium and so on also [bless]. By convention this is posited; but [for] “faculty-object” the word [blessing] does not exist. If the object is thoroughly conjoined, [then] sound and form, which have distance — and what is larger is also seen — [their grasp] refutes non-separation. Setting aside the talk of the cognition-producing cause-aggregate — which is the means? Is the means or is it the object? — why add “the arising of cognition”? Apart from object, faculty, mind, self (puruṣa), their conjunction and activity — what else are you saying is the cognition-producing cause-aggregate? Through conjunction with “cow” etc., the object of the means is “cow” etc.; but [the cognition], even when properly related to its object, — sense-cognition has no such capacity. What is cognised does not express itself; the field of the sense-faculty is form-as-such; [yet on their view] every non-object cognition would rest on perceptual awareness. If one admits the arising of cognition, and says another object is the result — [no], it cannot obtain another object; the cognition-of-the-appearing-object is itself the result, because there is no other result. If the arising of the cognition is other, [then it is] from the conjunction of its own causes, from the conjunction of [the several] pramāṇa[-factors]. If not-another, meaningless. Through transformation of the self, the cognition arises — [then] it should be impermanent. If the self does not change, it cannot be a means of knowledge.
“Existent” is redundant. It is added to exclude the non-existent, but “conjunction” already carries that: to speak of a faculty’s conjunction is already to speak of something existent. If the word is meant to mark off what each faculty faces, they must state the faculty-distinctions severally, pairing each faculty with its own object; if it means conjunction with existence in general, that is established already and needs no further proof.
It does not even exclude illusions. A mirage arises from existent conditions — the terrain, the sun’s heat, the differences among the great elements — conjoining with the eye and developing by stages into an erroneous consciousness. So “existent” fails to rule out the very cases it was introduced for.
“Union and blessing” fails too. [Mīmāṃsā] “Existent” means what unites with the faculty, or what makes the faculty finely clear. [Dignāga] Atoms also unite with faculties, and collyrium and foot-ointment also clarify the eye, so conjunction with these would count as perception. [Mīmāṃsā] Convention restricts the term, as “born-of-the-act” names only the cow though other classes are also born of action (compare the Chinese usage of “draught-beast” for the ox). [Dignāga] That convention is established for “born-of-the-act”; no such convention exists whereby “existent” signifies union-and-blessing with respect to faculty-objects. The appeal is imagination, not usage.
“Thorough conjunction” is refuted by observation. If the faculty had to be conjoined with every part of its object, sound and colour could not be grasped at a distance, nor could objects larger than the faculty be seen; it would take only what is contiguous, as scent does. Both are directly observed, so non-separation falls.
The “cause-aggregate from which cognition arises.” [Mīmāṃsā] Since no result is seen apart from the arising of cognition, they count the aggregate producing cognition as perception. [Dignāga] Which member is the instrument? If the object alone is the cause-aggregate, then “with proper conjunction, the sense faculties of a person constitute perception” would suffice, and the added clause “from which cognition arises” is idle. And apart from object, faculty, mind, self, their conjunction and their activity, what else is being named? That aggregate does not produce perception; faculty–object conjunction does not, since “present” would then attach to both terms; and proper conjunction is not separately established.
Sense cognition cannot determine “this is a cow.” To determine so requires conjoining the name “cow” with its meaning, and even on their own account sense cognition has no such examining capacity — it merely observes. Qualifier and qualified, expressed and expressor, and every conceptual construction are set up by mind, not by sense cognition.
The faculty’s field is the uncommon own-character. Though the object has many facts, what the faculty obtains as appearance is its uncommon essence, which produces the cognition arising with its own-character; such an essence can display nothing expressible, since the expressible is a general object. If the universal-aspect were also perceptual, every objectless cognition would count as perception, and cognitions ranging over all the qualities of colour would trespass on other consciousnesses’ fields, making the several faculties pointless. Fazun notes that the word “present” (現) operates in three places — pramāṇa, consciousness, object — of which only the pramāṇa is truly perception; the object is called present as the object of present measuring, and consciousness because it depends on the faculty and resembles the pramāṇa. Both are nominal.
On the result. [Mīmāṃsā] The result lies in a further object. [Dignāga] It cannot: the cognition bearing the appearance of the object is itself the result, since what makes an increased ascertainment is the result and the pramāṇa is not other than the cognition. If the arising of cognition is other than the cognition, it comes from the conjunction of its own causes, as the Vaiśeṣikas hold of existence — but conjunction so understood is permanent and can produce nothing. If it is not other, the word “arising” is useless, the cognition itself being the perception. And if cognition arises by a transformation of the self out of a prior stage, the self becomes impermanent, which they do not want; while if the self does not transform, the pramāṇa state is indistinguishable from the non-pramāṇa state and no settled pramāṇa can arise from it.
Thus the perception-doctrines imagined by the other schools are all difficult to establish.