Reverberations of Dharmakirti’s Philosophy

by Birgit Kellner | 2020 | 264,305 words

This page relates ‘Mandana Mishra’s Excursus on the Buddha’s Omniscience’ of the study on the philosophy of Dharmakirti (6th century) and his predecessor Dignaga (5th century). This collection of articles reflects philosophical currents in India, China and Tibet during their time and investigates the Buddhist theories of Pramana (“instruments of trustworthy awareness”).

Maṇḍana Miśra’s Excursus on the Buddha’s Omniscience

(By Hugo David)

This article is the partial outcome of the project “Language and Action in Early Brahmanism” carried out at the Austrian Academy of Sciences (Vienna) in 2013–2016 and funded by the Austrian Science Fund (FWF-project no. P25287–G15). Research for this paper also received support from the Royal Society of Great Britain (Newton International Fellowship). I thank Vincent Eltschinger, Sara McClintock and Vincenzo Vergiani for their very useful remarks on an earlier version of this paper, and Cristina Pecchia for allowing me to use a draft of her unpublished article on omniscience according to Dharmakīrti.

1. Introduction

1.1. The proof of the Buddha’s omniscience occupies, as is well-known, a prominent place in Indian Buddhist philosophy, and also constitutes an essential part of Mahāyāna Buddhism’s dogmatic construction, both as an essential component of its soteriological ideal and as a foundation for the transmission of Buddhist teachings through Scripture.[1] Yet, surprisingly, we do not find a systematic consideration of this topic in the ‘pramāṇa’-school until a comparatively late date. Although both Dignāga and Dharmakīrti deal in some detail with the perception of yogins,[2] their consideration of the ‘fourth’ type of perceptual cognition is done independently of the question whether a yogin could become omniscient by this means.[3] One must wait until the 8th century and the works of Śāntarakṣita (725–788), Kamalaśīla (740–795) and Prajñākaragupta (750–810?)[4] to find an articulate defence of omniscience in the school claiming Dharmakīrti’s heritage, further developed by thinkers like Śaṅkaranandana (800–980?), Jñānaśrīmitra (980–1040?) and Ratnakīrti (990–1050?).[5]

1.2. It is now generally admitted that the development of this new field of investigation within the ‘epistemological’ school of Buddhism owes a lot to the critique propounded in the 6th–8th centuries by Brahmanical thinkers, especially those belonging to the ritualis-tic school of Mīmāṃsā. From an early date, these thinkers regarded the possibility for a person to grasp the totality of being perceptually as a serious threat to their conception of the Veda as the unique means to know dharma, the ritual and ethical system defining what we call ‘Brahmanism.’ As far as we know, the first Brahmanical thinker to present a systematic attack against the belief in an omniscient being is Kumārila Bhaṭṭa (600–650?),[6] both in the codanā-section (ad MīSū 1.1.2) of the Ślokavārttika (ŚlV) (vv. 110cd–155) and in the corresponding fragments of the Bṛhaṭṭīkā (BṬ) quoted in Śāntarakṣita’s Tattva-saṃgraha (TS) (vv. 3127–3245).[7] There, it is principally the idea of the Buddha as the omniscient founder of a religion which is attacked, a fact that might explain the importance accorded to Kumārila’s ideas in later Buddhist thought.[8] Kumārila, however, was not the only Mīmāṃsaka to have offered a critique of the Buddha’s omniscience by the beginning of the 8th century. An important examination of this concept was also carried out by Maṇḍana Miśra (660–720?) in the first, aporetical part–generally (though somewhat improperly) known as ‘the pūrvapakṣa’–of his treatise on action and injunction, the Vidhiviveka (ViV), commented upon in the 10th century by Vācaspati Miśra in the Nyāyakaṇikā (NyK).[9] In this work, the existence of an omniscient being is made the subject of a lengthy refutation occupying more than a third of the whole pūrvapakṣa (ViV 15–25), which has not received much attention so far though it constitutes one of our main sources for the history of this debate in Mīmāṃsā before the time of Śāntarakṣita.[10] My purpose in this essay is to give the reader a first glimpse into this important text, concentrating on its treatment of the Buddhist idea of omniscience.

1.3. Considering the ViV was written perhaps no more than a century after Kumārila’s death, one is struck by how little Maṇḍana apparently owes to the old Master, whose works he simply never quotes in that section. Kumārila’s almost exclusive preoccupation with Buddhism is also not discernible in Maṇḍana’s work. For sure, the Brahmasiddhi (BS) (presumably Maṇḍana’s last work)[11] still mentions “the Buddha, wrongly believed to be omniscient” (sarvajñābhimatabuddha) as the prototype of the false teacher of dharma.[12] Vācaspati is also probably right in considering that the brief description of the Omniscient in the prose introduction to ViVK 15 refers in priority to the Buddha:[13] “an instructor of creatures (niyoktā bhūtānām), deserving to be obeyed, directly perceiving the means for realizing the Supreme Good as prescribed [in the Scriptures] (sākṣātkṛtānuśravikaśre-yaḥsādhana), who loves [to do] what is useful [to others] (hitakāma) and is omniscient (sarvajña).”[14] But in fact, only ViV 15 is directly concerned with Buddhism,[15] while the rest of the section discusses arguments from other schools[16] and even contains the earliest critique of divine omniscience in a Mīmāṃsā text, if not in Indian philosophy overall (ViV 20–24).[17] The tone of harsh religious polemics transpiring from Kumārila’s writings (especially the later ones)[18] is also hardly discernible in Maṇḍana’s text. Thus we do not find in the ViV anything comparable to Kumārila’s critical examination of non-Vedic Scriptures (āgama) in the ‘appendix’ to Tantravārttika 1.3.4,[19] a critique that also occupies a substantial part of his discourse on omniscience in the ŚlV (codanā –118–136) and BṬ (= TS 3186–3213). In other words, it seems omniscience has become, in the hands of Maṇḍana, less a matter for religious preachers and apologetes than for philosophers, the latter more likely to find in his work tools to convince fellow dialecticians than the former powerful incentives to attract faithful crowds and benevolent patrons.

1.4. The purpose of this study is to show that, in spite of all this, Maṇḍana plays a key role in the early debate on the Buddha’s omniscience, and entertains an intense dialogue with his two main predecessors in the field: Kumārila and Dharmakīrti. I will show, first, how the beginning of ViV 15 can be read as a systematic rejection of Kumārila’s argumentative strategy against omniscience, based on Dharmakīrti’s newly popularised logic (§ 2); second, I will examine how Maṇḍana uses Kumārila’s main argument in terms of a relation between perception and time to introduce a new type of epistemological consideration on the nature and cause of the Omniscient’s cognition (§ 3). Finally, I will consider the possibility that this evolution of the mīmāṃsaka discourse on omniscience influenced later stages of the debate in Buddhism as well, by tracking possible echoes of Maṇḍana’s ideas in a slightly later Buddhist work, Prajñākaragupta’s Vārttikālaṃkāra on Pramāṇavārttika 2.29 (§ 4).

2. Maṇḍana Miśra on non-apprehension and the Buddha’s speech

[A ‘Dharmakīrtian’ response to Kumārila?]

Every discussion of omniscience in Brahmanism must start from its prototype in Kumārila’s ŚlV and BṬ. In these works, the great Mīmāṃsaka makes it clear that his intention is not so much to prove that a human being cannot become omniscient (a possibility he actually leaves open) but rather to maintain an absolute distinction between entities that are accessible to the senses (aindriyaka) and others essentially beyond their reach (nendriyagocaraḥ), like the relation between the elements of a Vedic sacrifice (dharma)–actions, qualities and substances–and their expected result.[20] What is most disturbing, then, to Kumārila is that someone who would “see all things in a [single act of] perception” (sarvapratyakṣadarśin), as he defines the Omniscient in the BṬ (= TS 3138c), would also know dharma(s) perceptually: being a “knower of all things” (sarvajña), he would also be a “knower of dharma” (dharmajña), and this would contradict the exclusivity of the Veda on ethico-ritual matters.[21] Maṇḍana shares the same preoccupation: for him, the point is not the existence of an omniscient being in general, but of an “instructor of creatures”[22] promulgating a teaching on matters inaccessible to ordinary perception in the form of injunctions (vidhi/codanā).

The search for such a being is occasioned by Maṇḍana’s reflection on Prabhākara’s concept of “commandment” (niyoga)[23] which, in his view, cannot be operative without supposing such an instructor:[24]

A commandment is [in itself] a mere instigation (pravartanāmātra), and [only] that is the object of an injunctive suffix (liṅādi). What is grasped, then, through [that] speech[-unit] is [only] that “I am prompted [to do this]” (pravartito’ham); but the awareness that “I have to do [this]” (kartavyatāvagama)[25] comes from the fact that someone who deserves to be obeyed (anuvidheya) is the author of the commandment. [Only] when a commandment is given by such a [person] do I understand that “I have to do this;” otherwise, I feel a mere instigation, as it has been said [by Prabhākara]: “the commandment has the obligation (kartavyatā) as its content, it does not affirm (āha) the obligation.”[26]

2.2. Given this essential agreement on the main point at stake, one would expect to find at least an echo of Kumārila’s arguments in Maṇḍana’s text.

Instead of that, the prose development on ViVK 15 starts with the following statement, which seems at first entirely foreign to Kumārila’s main argumentative strategy:[27]

Even though (1’) the absence (nivṛtti)[28] of one’s own perception does not prove the non-existence (abhāva) of [objects] that are at a distance (viprakṛṣṭa),[29] and [although] (1”) the absence of a perception for all [beings] is impossible to establish unless one sees everything;[30] (2) [even though] the absence of all means of knowledge [regarding a certain object] does not prove that [this] object does not exist–for (2A) the absence of a non-pervasive [property] (avyāpaka) does not necessarily imply (ni-yam) the absence of the [corre-sponding] non-pervaded [property] (avyāpya),[31] and (2AI) [the existence of a means of knowledge–pramāṇa] is [indeed] non-pervasive [with respect to the existence of an object to be known–prameya], for it is not the cause (hetu) [of the existence of the object to be known],[32] and because there is [also] no essential connection (tanmātrapratibandha) [between these two properties],[33] for otherwise[34] everybody would see everything as there would be no difference [between us and an omniscient being][35] –; and (3) even though speech, etc. do not rule out [the existence of an omniscient] as there is no contradiction [in his making use of speech, etc.]; even so, (4) since we understand the absence of an effect (kāryanivṛtti) from the absence of [its] cause (kāraṇanivṛtti), from the impossibility (anupapatti) of a means of knowledge [embracing the whole realm of being] we infer the absence of its effect, namely an awareness of all things (sarvārtheṣu saṃvid).

This dense passage is for the most part a discussion of inferential matters, and is indeed a remarkable example of how logical considerations can successfully be applied to the solution of a philosophical issue. Maṇḍana discusses here the use of the negative inferential reason (hetu), “non-apprehension” (anupalabdhi), and its capacity to establish the non-existence (abhāva) of a given object–in our case, an “awareness of all things”–, which is the property to be proved (sādhya). His theoretical background is clearly Dharmakīrti’s logical system, as the mention besides anupalabdhi of two possible grounds for ‘positive’ inference (causality and identity) suffices to prove. In substance, Maṇḍana proposes to replace a series of illegitimate uses of anupalabdhi (conclusion to the non-existence of an entity by mere lack of perception of it, etc.) (1–3) by another, legitimate one, known to Buddhist logicians as kāraṇānupalabdhi (“non-apprehension of the cause”) (4).[36]

His argument can be reconstituted as follows (the sign “←” indicates a logical relation: “justified by”):

1. Absence of perception of an entity cannot establish its non-existence.

  1’. Case of one’s own perception.

  1”. Case of everyone’s perception.

2. Absence of valid knowledge of an entity cannot establish its non-existence.

  ←2A. A property p’s non-existence cannot justify positing another property q’s non-existence unless p pervades (√vyāp) q.

  2AI. Valid knowledge (pramāṇa) does not pervade its object (prameya).

    ←2AIa. No relation of causality.

    ←2AIb. No relation of identity.

3. Speech, etc. do not rule out omniscience in their possessor.

4. Omniscience can be negated, as an effect can be negated through the negation of its cause.

Now this reasoning is clearly not Maṇḍana’s invention; in fact, the very structure of the argument recalls Dharmakīrti’s ‘second’ consideration of anupalabdhi in the Pramāṇavārt-tika (PV) 1.198–204, especially PV 1.198–202 and the corresponding Svavṛtti (PVSV).[37] What is comparatively new, however, is the application of these reflections to the particular case of omniscience, which is not mentioned in this section of PVSV although Dharmakīrti suggests other possible consequences of his theory for religious philosophy.[38] The only text I could find where Dharmakīrti applies a similar reasoning to omniscience (sarvajñatva) is the Nyāyabindu (NB).[39] I suspect this original and quite massive reinvestment of Dharma-kīrti’s ideas must be interpreted in a polemical way. For the first victim of this exercise in ‘applied logic’ is, no doubt, Kumārila, whose arguments against omniscience are–with one exception to which I shall return later on (§ 3)–easily associated with one or the other ‘defective’ use of anupalabdhi. Thus, far from rejecting Dharmakīrti’s elaborations on adṛśyānupalabdhi as ‘Buddhist,’ Maṇḍana appropriates them and adapts them in order to free the classical Mīmāṃsaka argumentation against omniscience of some of its most flagrant weaknesses.[40] Let us now consider in more detail a few aspects of this strategy.

2.3. Kumārila’s refutation of omniscience, in the ŚlV and BṬ similarly, takes place in three successive stages: after having established that omniscience is intrinsically impossible (ŚlV[41] 112–115/TS 3157–3183), he shows that no evidence supports the assumption (kalpanā) of an omniscient being in the past (ŚlV 117–136/TS 3184–3236) and finally argues that, even if there had been such a being, he would have been unable to teach (ŚlV 137–140/TS 3237–3239).[42]

The second part of his demonstration, which is by far the longest, starts with a very simple argument showing how our main means of knowledge (pramāṇa) have no grip on an omniscient being:[43]

First, people like us do not see an omniscient being now; nor is it possible to postulate that there was [such a being], as [one can] deny [it]. Nor [can one postulate] an omniscient being on the basis of Scripture, for his [Scripture would have the undesirable consequence of having] mutual reliance [with his being an omniscient being]. If [Scripture] is composed by others, how is it understood to be a means of valid cognition?

Despite its simplicity and extreme popularity in later philosophical literature,[44] the ar-gument is not even mentioned in the ViV.[45] A plausible explanation for this could lie in Dharmakīrti’s newly introduced distinction of two kinds of non-apprehension (anupala-bdhi): non-apprehension in general (anupalabdhimātra [PVSV 101.23]; anupalambhamā-tra [PVSV 103.3]) and non-apprehension of what, under normal circumstances, would be fit for (perceptual) apprehension (upalabdhiyogya [PVSV 101.18]).[46] As Dharmakīrti explains it, the first type of non-apprehension can only produce doubt (saṃśaya) as to the existence of the object, but cannot prove its non-existence (asattā);[47] from our point of view, an object we do not perceive (say, ghosts, or a particular deity) may as well exist or not.[48] The second type of non-apprehension, on the other hand, positively establishes its non-existence,[49] like when we do not see a pot in front of our eyes even in the clear light of day. Following this important distinction, echoed in the passage of the ViV translated above (§ 2.2), the fact that an omniscient being “is not seen” (na dṛśyate) has no value whatsoever to prove that there is no such being, this regardless of whether we speak of the perception of a single person (svapratyakṣa) or of everyone’s perception (sarvapratyakṣa).[50] But the point is not only about perception. Dharmakīrti further claims that even complete lack of evidence about an object cannot persuade us of its non-existence (unless, of course, it fits all conditions for present perceptual apprehension): “one cannot ascertain that [objects that are at a distance] do not exist, even in the absence of [all] three means of knowledge [i.e.: perception, inference, and teaching through Scripture] (pramāṇatrayanivṛttāv api).”[51] Maṇḍana’s adoption of this argument without any change in the ViV can therefore be read as a rejection of Kumārila’s whole strategy for denying the existence of an omniscient being by mere lack of a pramāṇa capable of establishing it. To put it differently, Kumārila’s proof against the probability of the existence of an omniscient being–quite sufficient for the apologetic purpose of ‘weighing’ kalpanās–is systematically put aside by Maṇḍana, whose only concern is now with the intrinsic impossibility of omniscience.[52]

2.4. The topic of non-apprehension is not the only one where Maṇḍana chooses to distance himself from Kumārila by siding with Dharmakīrti. In v. 137 of the codanā-section of the ŚlV, Kumārila famously points to a contradiction in the assumption of the Buddha’s omniscience:[53] had the Buddha really been omniscient, he would have been unable to teach; teaching is a form of operation (vyāpāra) that naturally implies some kind of intention to speak, thus a form of desire/passion (rāga), incompatible with the supposedly dispassion-ate (rāgarahita) state characterizing the Omniscient.[54] As is well-known, Dharmakīrti is familiar with Kumārila’s argument (or a similar one)[55] and considers it a fundamentally flawed use of the inference from the effect (kāryānumāna). Surely, some teachings are made out of desire for a benefit, as we observe among worldly teachers, but this need not be the case for all teachings and does not rule out other reasons for teaching, less incompatible with the state of omniscience, such as compassion (karuṇā): “Just as a passionate [person] (rakta) speaks, so does the impassionate (virakta) one, too. Therefore it is not apprehended from speech as such [whether one is passionate or dispassionate].”[56] That this is precisely the argument alluded to by Maṇḍana when he says in the ViV that “speech, etc. do not rule out [the existence of an omniscient being], as there is no contradiction [in his making use of speech, etc.].” (argument no. 3 in the above-quoted text) is proved, besides the explicit mention of “speech, etc.” (vacanādi),[57] by a further allusion to the SV on PV 1.12 in the first book of the BS.

In that passage, Maṇḍana distinguishes between two concepts of desire–icchā (desire in general) and rāga (passion, which is an obstacle to liberation from saṃsāra)–, and it is again Dharmakīrti’s definition of rāga in PVSV 9.5–6 that he calls for support:[58]

Passion (rāga) is not mere desire (icchāmātra); they call “passion” that attach-ment to unreal qualities [of the object] brought about by nescience (avidyākṣi-ptam abhūtaguṇābhiniveśa).[59] But the mind’s inclination (prasāda) towards reality–i.e. its delight (abhiruci) [in reality] or desire (abhīcchā) [for it]–, following [its] purification through the vision of reality, does not fall into the category of “passion” (rāga), just as aversion produced by one’s vision of that reality which is the worthlessness of transmigration does not fall into the category of “hatred” (dveṣa).

2.5. As we can already see from our analysis of its initial portion, the discussion of omniscience in ViV 15 reveals Maṇḍana’s deep familiarity with the works of Dharmakīrti–especially PV 1 and the Svavṛtti, but also the Nyāyabindu and, as we shall see later on (§ 3.5), the Pramāṇaviniścaya –, far deeper in any case than that of any Mīmāṃsaka before (and perhaps even after) him. But it also testifies to the complexity of his engagement with the famous Buddhist logician, which contrasts with Kumārila’s strategy of frontal confrontation with Buddhism in general. It is noteworthy in this respect that the debate is never on matters of pure logic, an impression that can only be confirmed by the remaining part of ViV 15. The two kinds of non-apprehension, the negation of the effect through the negation of its cause (see below, § 3), even Dharmakīrti’s conclusions about fallacious uses of the inference from the effect: everything is accepted in block by Maṇḍana and never questioned again in his work. His way to consider anew the topic of the Buddha’s omniscience in the ViV is therefore to play, so to say, Dharmakīrti’s part, and to reconstruct what could have been a ‘Dharmakīrtian’ answer to Kumārila from elements scattered throughout Dharmakīrti’s works. But this is just a prelude to Maṇḍana’s real encounter with Dharmakīrti, carried out once again with his own weapons. The field of this encounter, however, would not be pure logic any more, but the epistemology of perception.

3. Time, perception and Scripture: on a mīmāṃsaka use of kāraṇānupalabdhi

3.1. If the bulk of Kumārila’s arguments against the Buddha’s omniscience, as we have seen, finds no favour in the eyes of his most immediate successor in Mīmāṃsā, a small sec-tion of the ŚlV (codanā° –112–115) and BṬ (= TS 3157–3183) remains to be investigated, in which Kumārila argues for the incompatibility between omniscience and the very nature of perception. That preoccupation, at least, is well in line with Maṇḍana’s philosophical agenda and the problem is in fact at the heart of his enquiry in ViV 15. Once again, the occasion for this reflection is given to him by a logical remark by Dharmakīrti.

As we have seen before (§ 2.3), Dharmakīrti insists in the PV that mere absence of perception (or even of knowledge in general) of an object that is not fit for perceptual apprehension produces no certitude (niścaya) of its non-existence, but only doubt (saṃśaya) regarding its existence. However, there are other ways to produce such a proof of the non-existence of an object, one of them being to find evidence for the absence of a cause (kāraṇa) of the object, as explained in the following passage of the PV and SV:[60]

[PV:] But the fact that we do not know the efficient [cause] (kārakājñāna) of [a given] effect proves that [such an effect] does not exist. [SV:] In case the [property] to be proved is the non-existence (abhāva) of a [given] nature (svabhāva), [we just] said[61] that the [mere] absence of apprehension of that [nature] does not constitute a [valid] means of knowledge (apramāṇam). But the non-apprehension of [its] efficient [cause] (kārakānupalambha)[62] is a [valid] means of knowledge (pramāṇam eva), for it is impossible (nāsti saṃbhavaḥ) that an effect should take place without a cause.

Dharmakīrti’s reasoning is clearly alluded to in the passage quoted above (§ 2.2) by Maṇḍana, who does not speak, however, of “non-apprehension of a cause” (kāraṇānu-palambhaanupalabdhi), but simply of the “absence of a cause” (kāraṇanivṛtti). And indeed, a major issue of Dharmakīrti’s reasoning–which is not entirely clear from this passage of the SV–is that “non-apprehension of the efficient [cause]” (kārakānupalambha) cannot be mere non-apprehension (anupalabdhimātra), but has to be non-apprehension of the second kind, where the object is fit for (perceptual) apprehension, lest the inference becomes equally inconclusive.[63] This, of course, is not the case of most objects placed at a distance (viprakṛṣṭa) or supersensible (atīndriya) objects like particular deities, etc., whose cause is very likely to be also beyond the reach of the senses. The case of omniscience is somewhat peculiar, though, due to the presupposition–apparently shared by all partic-ipants in this debate–that (valid) knowledge of all things must be, to begin with, valid knowledge over all. Just as “persons of exception” (puruṣātiśaya) remain persons all the same, “exceptional visions” (darśanātiśaya) differ in degree (of precision, intensity, clarity, etc.) from ordinary perceptions, but they obey the same principles as any other perceptual cognition.[64] Maṇḍana’s task is, then, to prove that the cause of perception is such that it can never produce a knowledge of all things, and thereby to undermine the very possibility (saṃbhava) of omniscience.[65]

3.2. Kumārila, who already had some thoughts on that topic, mostly insists on the mutual delimitation of our senses, which disqualifies a cognition of all things at once: acute as it may be, the eye–be it a Divine Eye–[66] will never perceive sounds or smells; yet sounds or smells are, no doubt, part of the totality of being.[67] In order to grasp everything at once, the senses of the Omniscient would therefore have to function simultaneously in an eminent way, a possibility which contradicts the widely admitted principle that two cognitions (say, the smell of a strawberry and the vision of its red colour) never take place exactly at the same time.[68] As before, but for reasons that are far less clear, this popular reasoning did not find its way into the ViV.[69] Maṇḍana prefers to concentrate on another aspect of the mīmāṃsaka theory of perception, going back to Jaimini’s treatment of pratyakṣa in MīSū 1.1.4, namely its relation to the present time (vidyamāna). Perception, so the sūtra goes, cannot be a cause (nimitta) for our knowledge of dharma “because it grasps [something] present” (vidyamānopalambhanatvāt).

Applying this conclusion to the debate on omniscience, Maṇḍana makes it into a general statement as to the nature of sensory perception:[70]

Perception, when brought about by the eye, etc. [does] not [apply] to all objects, for the [eye, etc.] have a restricted domain: their domain is exclusively some particular [object], which is present (vartamāna) and related (saṃbaddha) [to the senses], and not all objects are like that. Now it is true that, since it is possible to be aware of all sorts of knowable objects, a restriction (niyama) [of the domain of perception] in terms of form (rūpa) is hardly defendable, and so is also [its restriction] in terms of acuity, feebleness, etc. (paṭumandatādi). [Moreover,] since we cognize (pari-chid) objects at all sorts of distance and in [all sorts of] measures, a restriction of relation in terms of distance (deśa) [or] measure (parimāṇa) [is also not possible]. But [a restriction] in terms of time (kāla) is defendable (nirūpyate), for in [the view that the eye, etc.] operate by reaching [their object] (prāpyakāritve),[71] they cannot reach it if [the object] is not present; the same [is true] in [the view that the eye, etc.] operate without reaching [their object] (aprāpyakāritve), since [in that case] one needs a [special] capacity (sāmarthya) of the object [to be known], [and objects] that have not come into being or have ceased to exist have no [such] capacity, for they are inexpressible (anupākhyeya) [in terms of being and non-being].[72] And therefore the eye, etc., should they have a special excellence (ati-śī ), may only make their own domain known in an eminent way (adhikam) in terms of distance, measure [and] number, but not what is beyond their domain, [namely] what has not come to existence, [and] what has ceased to exist. Therefore it is said [in Mīmāṃsāsūtra 1.1.4]: “because it grasps something present” (vidyamānopalambhanatvāt).

Both characteristics of the object of perception highlighted in this passage (saṃbandha/ vartamānatva) have their source in MīSū 1.1.4, where “contact with the senses of a person” (saṃprayog[aḥ] puruṣasyendriyāṇām) is mentioned besides “being grasped at the present time” (vidyamānopalambhanatva). Yet it is obviously the second characteristic that, above all, captivates Maṇḍana’s attention. His four-fold suspension of restriction (niyama) in terms of form, acuity, measure and distance (to which number can be added) also makes the very idea of a relation with the senses practically useless. For what kind of ‘relation’ would there be between the senses and objects placed at an extreme distance, for instance, if not their mere coexistence in one and the same moment? And what would be the point of restricting the object of perception to what is ‘related’ to the senses if virtually everything can be related to them? Thus it is possible to read in this text a form of thought experiment, allowing us to discriminate between factual limitations of perception (in terms of form, size, etc.) which, in principle, can be suspended, and a natural limitation of perception, i.e. its relation to the present time, which no artificial extension of our faculties can allow us to surpass. This insistance on the temporal dimension of perception might be a natural consequence of Maṇḍana’s main thesis in the field of ontology, voiced in ViV 12, identifying existence (sattā) with being present (vartamānatā): if only present things are perceptible, it is perhaps because they only ‘exist’ in the true sense of the term.[73] But this might also be his one true link to Kumārila, whose core argument against the possibility of foresight (obviously an essential component of omniscience) in ŚlV (codanā°) 115 is precisely the natural limitation of sensory perception to the present moment.[74]

3.3. Maṇḍana’s strategy against omniscience thus appears, at this point, essentially as a reduction of Kumārila’s arguments to a single one: perception, relying on the operation of the senses, can only grasp things in the present. His use of this argument in ViV 15, however, marks a radically different approach to Buddhist theories of perception. As we have seen, all arguments of the ŚlV (codanā°) 112–114 and the corresponding verses of the BṬ are based on the capacity of the senses to grasp all things in a single moment of perception, and this may safely be extended to his remark in v. 115 as well. This presupposes that perception can occur only through the senses, an assumption justified, in Kumārila’s perspective, by his rejection of all kinds of supersensory perception in the chapter of the ŚlV dealing with pratyakṣa (ad MīSū 1.1.4). Quite the opposite, Maṇḍana chooses to confront Buddhist epistemologists on their own ground in order to show that even supersensory perception as they conceive it is incompatible with omniscience. This ‘dialectical’ attitude, so characteristic of Maṇḍana’s philosophical style, allows him to open an entirely new field of philosophical enquiry into the nature of the Omniscient’s cognition, which was to acquire some prominence in later stages of this debate.[75]

3.4. Three types of perceptual awareness are considered in ViV 15, corresponding to Dharmakīrti’s four types[76] with the exception of self-awareness (svasaṃvedana): perception “born from the eye, etc.” (cakṣurādijanman), “mental” (mānasa) perception and perception “born from meditation” (bhāvanāmaya), which is also how Dharmakīrti defines the cogni-tion of yogins (yogināṃ jñānam).[77] Among them, the greatest attention is not devoted to the last kind of perception, as we would probably expect,[78] but to mental cognition. For sure, external senses are riveted to the present time, but the mind need not be; in fact we observe that dreams, fantasies and other creations of the mind have no evident link to the world of sensation, and also deal with past and future events.[79] Could omniscience be a cognition of that kind? We cannot be sure whether Maṇḍana had a particular Buddhist thinker or school in mind while refuting that possibility, but I find it unlikely that his opponent should be Dharmakīrti, whose conception of mental cognition (manovijñāna [NB 1.9]) explicitly excludes independence of the mind from the senses. In fact it seems Maṇḍana chooses once again not to refute Dharmakīrti’s ideas–at least, not at first–, but skilfully to take them out of their original context to fit his own purpose. As is well-known, mental perception for Dharmakīrti does not only cover internal mental phenomena such as awareness of pleasure and pain, but also the (non-conceptual) moment of attention immediately following a sensation, that of a patch of blue for instance, in which we become aware that there is ‘something’ in front of us without yet knowing that it is ‘blue.’ His main preoccupation in PV 3.239–248 is precisely to show that the content (viṣaya) of that moment of perception is different (anya) from what has been previously experienced (pūrvānubhūta)–so that it can be considered valid knowledge (pramāṇa)–,[80] but is also not entirely “unseen” (adṛṣṭa), so that awareness of sound, for instance, cannot follow from a sensation of blue, or awareness of blue from no sensation at all, as in the case a blind man.[81]

Taking the best party of these remarks, Maṇḍana insists, in his turn, on the dependence (pāratantrya) of the mind (manas) on the senses when it comes to external objects (bahir):[82]

Even mental perception (mānasaṃ pratyakṣam) [cannot produce omni-science],[83] for the mind (manas) has no autonomy (asvatantratvāt) [with regard to what is] outside (bahir).[84] If it had [such an] autonomy (svātantrya), the undesired consequence would be that nobody would be blind, deaf, etc. To explain: regarding perceptive awareness (pratyakṣa[ṃ] vedana[m]) of forms/ colours, etc. (rūpādi), the [mind] is dependent (paratantra) on [senses] like the eye, [and] it is limited by their very limitation (niyama);[85] otherwise, as [we have just] said, the undesired consequence would be that nobody would be blind, etc. If [you object] that [this undesired consequence, namely] that nobody would be blind, etc. does not occur, for [mental perception] depends on the [objective] correlate of [its] homogeneous and immediate cause [i.e. a cognition] born from the [senses] (tajjasamanantarapratyayasahakāryapekṣa-ṇād),[86] [our answer is that,] in all cases, the dependence [on the operation of the senses] does not disappear, for the content [of mental perception] is born of the content of a [particular sense].[87] Were it not, since [mental perception] would take place because of a [moment] that would cooperate with [the cogni-tion] born from another sense (indriyāntarajasahakāriṇo’pi),[88] there would indeed be no blind, etc.!

Despite the presentation of the arguments in a polemical form, there is probably little here that Dharmakīrti would actually disagree with. I find it in general unlikely that a philosopher arguing for the mind’s autonomy (svātantrya) in grasping external objects would really draw any benefit from Dharmakīrti’s theory of mental awareness, and from his distinction of two objective ‘moments’ corresponding to sensory and mental perception. Maṇḍana’s detailed discussion of that theory in the ViV (S 474.2–542.1), leading him to reaffirm the orthodox mīmāṃsaka view that “the mind never operates directly (sākṣāt) and independently (svatantram) on an [object] that is not internal (āntara),”[89] thus conscientiously fulfills the epistemologist’s task, but has little bearing on the topic of omniscience proper.

3.5. A more profound divergence between the two philosophers comes out of Maṇḍana’s brief discussion of perception “born from meditation” (bhāvanāmaya) at the end of ViV 15 (S 542.1–555.2). While Kumārila spoke in very general terms of “the perception of past or future objects, or of one that is very small or obstructed, believed by some to belong to yogins and liberated souls (muktātman),”[90] Maṇḍana specifically addresses the Buddhist epistemologists’ conception of yogic perception, especially their belief that it results from “repeated practice” (abhyāsa) or habituation to cognitive contents reached by some other means.[91] As is well-known, Dharmakīrti thinks that the main cause of a yogin’s perception is mental cultivation or “meditation” (bhāvanā), half way between ‘rumination’ of an idea in view of its perfect assimilation and ‘imagination’ understood as the vivid representation of something formerly conceived. In more Buddhist terms, meditation consists of the repeated presentation to the mind of the practitioner of the four Nobles’ Truths, culminating in their direct apprehension (sākṣātkāra) or “vision” (darśana).[92] To that activity Dharmakīrti ascribes the power–well-attested in persons subject to hallucinations born from desire, fear, madness, etc.[93] –to produce an awareness with all external traits of perception: clarity (sphuṭābhatā [PV 3.8b]/spaṣṭa[tva] [PV 3.281d]/spaṣṭābha[] [PV 3.284c], etc.), non-conceptuality (kalpanāpo[h]a [PVinK 1.4a/PV 3.123]/akalpaka[tva] [PVinK 1.28d]/ akalpa[tva] [PV 3.285d], etc.). The difference between mere hallucination and the cognition of a Buddhist Saint only lies, then, in the latter’s being “veridical” (saṃvādin [PV 3.286a]) or “non-erroneous” (abhrānta [PVinK 1.4b]), applying as it does to an object whose reality (bhūta[tva] [PV 3.285]) has been ascertained through Scripture (āgama) and reasoning (yukti). This last characteristic, which makes the cognition of yogins into valid knowledge (pramāṇa) or perception (pratyakṣa) in the true sense of the term,[94] is also the main topic of Maṇḍana’s critique. For to claim, as Dharmakīrti does, that yogic perception is non-erroneous as it follows on hearing Buddhist Scriptures and pondering over their content amounts to saying that meditation is essentially non-productive.

As Maṇḍana puts in the Brahmasiddhi:

“[An injunction] concerning a cognition of the second type [= mental cultivation][95] does not concern the comprehension of reality (tattvāvabodha), but only the repeated practice (abhyāsa) [of that comprehension].”[96]

This ‘borrowed’ character of the content of meditation, a warrant for its validity in Dharmakīrti’s view, is precisely what leads Maṇḍana to disqualify it as mere second-hand knowledge:[97]

Even [perception] born from meditation (bhāvanāmaya) is about an object [previously] heard about [in the Scripture] and/or known by inference (śrutā-numitaviṣaya), for it is impossible to meditate at random; since it conforms to a former cognition’s having a real object (bhūtārthatva) or the contrary (viparyaya), it is dependent (paratantra) on Scripture and inference, and re-lies upon [another means of knowledge to ensure its validity] (sāpekṣatvāt). Therefore it is not a means of valid knowledge (apramāṇam).

In these lines, which I propose to read as a direct response to Dharmakīrti’s opposite statement in PVin 1.28,[98] Maṇḍana inaugurates what would be the invariable position of the Advaita tradition on the nature of meditation or “contemplation” (nididhyāsana) for centuries to come. Meditation being only the repeated and progressively intensified thought of an object, it cannot produce by itself any knowledge of that object. So, in order to be mentally cultivated, the object must be reached by some other means, scriptural or inferential. If meditation, then, has its use as a means of assimilation or ‘realization’ of what has been grasped, it does not bring anything new and because of that it is “not a means of valid knowledge” (apramāṇa).[99] While Dharmakīrti interprets the transition from conceptual knowledge originated from Scripture and reasoning to the immediacy and non-conceptuality of the yogin’s insight as a passage from illusion to truth, Maṇḍana rather insists on the identity of content of all three cognitions: no matter how we take it, it is always the same reality that is “heard, thought and meditated upon,” as the Upaniṣad has it,[100] and neither perception nor reasoning can grasp it without the help of Scripture. Beyond the limited controversy about the yogins’s cognition and its capacity to account for omniscience, we sense a deeper disagreement concerning the very function of Scripture and its place in the path to liberation: from a mere preliminary (and to a certain point superfluous)[101] stage leading the adept to a more authentic and direct apprehension of reality in Dharmakīrti’s view, the audition of Scripture has become for Maṇḍana the very centre of his Vedāntic soteriology, the means of knowledge par excellence that other pramāṇas may of course supplement, but never entirely replace.

3.6. With this last point it seems we have exhausted most of what Maṇḍana had to say on the topic of the Buddha’s omniscience. It is now time to enquire whether his arguments aroused any response from the Buddhist side in the following centuries, as was the case for Kumārila. Our main field of investigation will be the work of an immediately later Buddhist philosopher, Prajñākaragupta, on whom Maṇḍana’s influence–so is at least my contention–is most easily discernible.

4. An early Buddhist paraphrase of ViV 15

[Prajñākaragupta’s Pramāṇavārttikālaṃkāra (PVA) ad PV 2.29 (vv. 2.358–370)]

4.1. Although Maṇḍana’s influence on later Buddhist thought is yet to be properly valued, it is nevertheless certain that the ViV was read and extensively used by some at least among later Buddhist logicians. Of the several texts one could invoke in support of this claim,[102] none is perhaps as revealing as Prajñākaragupta’s commentary on PV 2.29 (PVA 2.358–370).[103] Prajñākaragupta is probably the first commentator on Dharmakīrti’s Pra-māṇavārttika to regard the Buddha’s omniscience (sarvajñatva) as a central topic of the Pramāṇasiddhi-chapter (= PV 2).[104] His commentary on PV 2.29–33 is therefore, along with chapter 26 of Śāntarakṣita’s Tattvasaṃgraha, among the oldest available testimonies of that debate stemming from the Buddhist pramāṇa-tradition. Prajñākaragupta’s long discussion of PV 2.29, where Dharmakīrti objects to the possibility of a knowledge of objects beyond the senses (parokṣārthajñāna) in the absence of a means (sādhana) to accomplish it, forms the pūrvapakṣa of that section, and is for the most part a web of mīmāṃsaka arguments set against the possibility of omniscience. This section of the Vārtti-kālaṃkāra is thus of high significance for the early history of this debate in Buddhism and Mīmāṃsā alike.[105] In his recent study of that section, Sh. Moriyama (2014) rightly points out Prajñākaragupta’s indebtedness to Kumārila’s works–both the ŚlV and the BṬ–[106]but also notes (pp. 59–60) that this cannot account for the pūrvapakṣin’s argumentation as a whole, which has no clear equivalent in the works of the great Mīmāṃsaka. Adding to Moriyama’s remarks, I shall argue that Prajñākaragupta’s model in building his pūrvapakṣa is not only Kumārila, but also Maṇḍana, and that the first half of the text (PVA 2.358–363) in particular can be read as a paraphrase of ViV 15.[107] Incidentally, this identification of one of Prajñākaragupta’s main opponents will help us, it is hoped, to solve certain difficulties in the interpretation of that delicate passage, and also to highlight certain minor divergences between the Buddhist scholar and his Brahmanical source.

4.2. Prajñākaragupta’s fundamental distinction, to begin with, between two possible inter-pretations of the word sādhana (“means”) in PV 2.29bc (tatsādhanasya ca / abhāvāt)[108] –namely, as the (efficient) cause (hetu/kāraṇa) of the Buddha’s omniscience and as the (informing) cause of our certitude (niścaye hetuḥ) of an omniscient being’s existence–[109] has generally been interpreted in reference to the ŚlV or BṬ.[110] Yet nowhere does Kumārila formulate such an opposition, which on the other hand closely corresponds to Maṇḍana’s distinction, already found on the threshold of ViV 15, between the (efficient) cause (hetu/kāraṇa) that should account (upa-padcaus) for omniscience112 and the cause of our knowledge (jñāna) of an omniscient being.[111] It is thus simpler to assume that Maṇḍana’s distinction is the source of Prajñākaragupta’s twofold interpretation of the word sādhana in Dharmakīrti’s verse.

4.3. The assumption of a debt to Maṇḍana further allows us better to understand the struc-ture of Prajñākaragupta’s pūrvapakṣa, which already caused some difficulty to its Indian commentators. Thus Yamāri (11th c.) tentatively identifies four parts in PVA 2.358–363, corresponding to four possible causes (rgyu) of the Buddha’s omniscience: sensory cog-nition (dbang po’i shes pa; Skt. *indriyajñāna?), the senses and the object (?) (dbang po dang don, Skt. indriyārtha?), mental cognition assisted by repeated practice (goms pa dang bcas pa’i yid kyi shes pa, Skt. *abhyāsavanmanojñāna?) and inference (rjes su dpag pa; Skt. *anumāna?).[112] To this rather unlikely organisation of the pūrvapakṣin’s proof, Moriyama (2014: 57–62) opposes his own twofold division, which sounds much more promising: (1) refutation of omniscience as a form of sensory perception (v. 358–359), (2) refutation of omniscience as a form of mental perception (v. 360–362). The problem is that Kumārila, whom he considers to be Prajñākaragupta’s main model, never seems to consider that omniscience could be something other than sensory perception, such as for instance mental perception. So, either one considers that Prajñākaragupta himself introduces that possibility[113] or one has to admit that he draws from some other source, which is then very likely to be Maṇḍana’s set of three (not two) hypotheses: omniscience as a form of sensory, mental or yogic cognition (see § 3.4). The following table presents the various hypotheses in presence regarding the nature of the Omniscient’s cognition:

ŚlV/BṬ ViV 15 PVA 2.358–363 PVA 2.358–362
(Moriyama 2014)
[indriya-pratyakṣa] cakṣurādijanma pratya-kṣam (S 461.2–468.1) indriyajñāna (1.358–359 + 50.24) indriyajñāna (1.358–359 + 50.24)
  mānasaṃ pratyakṣam (S 468.1–542.1) manovijñāna (1.360) manovijñāna (1.360–362)
  bhāvanāmayaṃ pratyakṣam (S 542.1–555.2) abhyāsāt [vijñānam]/bhāvanā (1.361–363)  

4.4. Arguments set against omniscience as a form of sensory or mental perception are clearly similar in the ViV and PVA: the limitation of the domain (viṣaya) of sensory perception to what is related/proximate (saṃbaddha [ViV]/saṃnihita [PVA]) to the senses,[114] the mind’s lack of autonomy (svatantra[tva]) from the senses in knowing external objects.[115] A more delicate issue is whether there is any allusion to the perception of yogins in this passage of the Vārttikālaṃkāra, as is my contention.

The three kārikās at stake (PVA 2.361–363) read as follows:[116]

(361) The clarity (spaṣṭatā) of that [cognition acquired] through repeated prac-tice (abhyāsa) cannot encompass all [objects]. If it (tasya) relies on Scripture, [its] erroneousness (bhrāntatā) also (api) follows; (362) on the other hand (tu), one does not reach all things (sarvaṃ vastu) as they are established by an inference (anumānaprasiddha), so there is no meditation (bhāvanā) that can encompass everything; how [then would one become] omniscient? (363) Through repeated practice of the [sacred] treatises, etc. (śāstrādi) one can only understand what is taught by them (śāstraprabhṛti); how [then] will anyone (tasya) attain knowledge of the totality [of being] (sākalya)?[117]

Although these verses still pose considerable problems in the detail of their interpretation, one can clearly recognize in them an elaboration on Maṇḍana’s main thesis regarding meditation (see above, § 3.5): omniscience cannot result from meditation (bhāvanā –362cd), for it only consists in the repeated practice (abhyāsa –361a/363a) of what has already been obtained by some other means (i.e. Scripture or inference). It is also possible that v. 363ab should be read in reference to Maṇḍana’s idea of meditation as an essentially non-productive activity, although this is far from certain.[118] It seems in any case that Prajñākaragupta substantially changes the nature of Maṇḍana’s argumentation by insisting, above all, on the incapacity of the two ‘root-pramāṇas’ to apprehend all things,[119] while Maṇḍana rather insisted on the “heteronomy” (pāratantrya) of meditation, preventing it from becoming a pramāṇa in the full sense of the term. So, if the general structure of this pūrvapakṣa seems to follow that of ViV 15, the detail of the argument is a blend of Kumārila’s, Maṇḍana’s and–one may think–Prajñākaragupta’s own reflections.

4.5. Assuming, as I did, that most arguments in the first part of this pūrvapakṣa are drawn from the ViV will also, it is hoped, allow us to solve problems in the detail of the interpretation of that passage of the Vārttikālaṃkāra. Two verses are particularly problematic, namely vv. 2.358 and 2.359cd. The first verse is interpreted by Moriyama (2014: 57–58 and 245, n. 12) as an allusion to ŚlV (codanā°) 112–114 which is, as we remember, a crucial group of stanzas dealing with the mutual limitation of the senses, barring them the access to the totality of being.

One has some difficulty, however, to read this argument in Moriyama’s translation of PV 2.358:[120]

If an omniscient being arises despite the non-distinction regarding sense fac-ulties and objects [between omniscient beings and ordinary people], everyone would become omniscient because of the non-distinction regarding sense faculties and objects. (Moriyama 2014: 245)

Although Moriyama does not provide much explanation for this, the logic behind his translation seems to be the following: since the senses of the (putative) Omniscient–the historical Buddha for instance–are not different from ours, they share the same limitations (358ab); if we suspend this limitation (admitting, for instance, that the eye could grasp sounds or smells), then there is no reason why everybody should not become omniscient (358cd). The interpretation of the compound indriyārthāviśeṣa as the “non-distinction of the senses and the object [in us and in an omniscient being]” looks quite forced though, which makes me suspect that this translation somehow misses the point. The argument becomes clearer, I think, if we relate Prajñākaragupta’s remark to Maṇḍana’s reasoning on the relation of identity (tanmātrapratibandhaanubandha) possibly underlying a relation of invariable concomitance (vyāpti) between the means of valid knowledge (pramāṇa) and the object to be known (prameya) (see above, § 2.2 [argument 2AIb] and nn. 31–32):[121] should an object be identical with the means to know it (e.g., the senses), there would be no difference between existing and being known, and everyone would become omniscient.[122]

On that basis, we can hopefully render Prajñākaragupta’s argument in a more faithful way:

Even if an omniscient being arose out of the absence of difference between a sense and [its] object (indriyārthāviśeṣa), everybody would become omniscient because of this absence of difference between a sense and [its] object!

This is not at all a central point in Maṇḍana’s argumentation, and I am struck by the amount of knowledge of Mīmāṃsā Prajñākaragupta expects from his reader (unless, of course, Maṇḍana himself is borrowing the argument from a Buddhist source). The same impression results from another possible hint at ViV 15 in PVA 2.359cd. As we have seen above (n. 107), the first half of this verse (aśucyādirasāsvādasaṃgamaś cānivāritaḥ /) is almost certainly a paraphrase of a verse of the BṬ (= TS 3144) arguing that an omniscient being, who would experience all things, would also experience the taste of impure things, etc. (aśucyādirasa).[123]

The second half of the verse (prāpyakārīndriyatve ca sarvavit katham ucyate //) is read by Moriyama (2014: 245) as a continuation of this argument, and translated as follows:

And [thus,] if [an omniscient being’s] sense faculties function after having had a direct connection [with an object], how can he be [honorably] called an omniscient being?

This translation is in itself impeccable, and it also makes perfect sense to say that the perception of impure things is especially problematic if the senses operate while reaching (prāpyakārin) their object. The presence of ca in pāda c is disturbing though,[124] and suggests another argument may be alluded to. As we saw (§ 3.2), the difference between prāpyakārin and aprāpyakārin is also mobilised by Maṇḍana while dealing with sensory perception to establish that neither explanation of perception (i.e. with and without a contact between the senses and the object) can account for a knowledge of past and future objects (S 465.1–466.2). I find it plausible that Prajñākaragupta reminds us of this argument, a possibility that would also match our main hypothesis that he is following the chronological order of ViV 15. If this proved correct, the allusion would be even more elliptic than in the preceding case, and would presuppose a reader fully conversant with the detail of Maṇḍana’s argumentation.

124 ViV 15: () tanmātrapratibandhābhāvāc ca, anyathā sarvasya sarvadarśitvaprasaṅgaḥ, aviśeṣāt (S 459.4–460.1–translated above, § 2.2). Supposing Prajñākaragupta is indeed alluding to that argu-ment, it is possible that he interprets aviśeṣa in ViV 15 as well as the absence of difference between the senses and the object (indriyārthāviśeṣa). This intepretation would differ from Vācaspati’s under-standing of that term as referring to the absence of difference between us and the Omniscient. See NyK 460.7–8: aviśeṣād asmadādīnāṃ bhavadabhimatena saha sarvavidā. My translation of the passage (above, § 2.2) follows this last interpretation.

All this suggests that the ViV was not only known to Buddhist scholars, but that its contents were also fairly widespread in learned Buddhist circles by the end of the 8th century. The following table, which also integrates evidence found in previous scholarship, summarizes my hypotheses concerning the sources of this pūrvapakṣa:

ViV 15 (S 445.1–458.1) 2.358–370
(prose introd.)
ViVK 15a/ViV 15 (S 461.2–462.1) 2.358–370
(prose introd.)
ViV 15 (S 459.4–460.1) 2.358    
BṬ (= TS 3144) 2.359ab    
Idem/ViV 15 (S 465.1–466.2) 2.359cd    
ViVK 15d/ViV 15 (S 468.1–470.1) 2.360    
ViV 15 (S 542.1–555.2) 2.361–363    
ŚlV (cod°) 115cd/BṬ (= TS 3173cd) 2.362ab    
ŚlV (cod°) 134/BṬ (= TS 3191) 2.364    
ŚlV (cod°) 137/BṬ (= TS 3238–3240) 2.365    
≈ TS 3249/ViV 18 (S 675.1–676.1) 2.366    
PV(SV) 1.335 2.358–370
(prose introd.)
ViV 18 (S 676.1–3) 2.370/PVAS 114.26    


Having reached the term of this enquiry, Maṇḍana Miśra appears to us, without contest, as the other great voice of Mīmāṃsā in the early debate on the Buddha’s omniscience. Less massive, less uncompromising, less influential also than Kumārila’s, his critique is nevertheless more complex, and philosophically more ambitious. It may also have served a slightly different purpose. For sure, Maṇḍana’s final view essentially coincides with that of his predecessor: no Omniscient can legitimately claim to instruct people about their religious duties, their origin or destiny, or about the path leading them to beatitude.

Yet this reaffirmation of the basic Mīmāṃsā position on religious authority does not imply, in the case of Maṇḍana, a complete hostility to the ideal of omniscience, as shown by the following passage of the Brahmasiddhi:[125]

[Objection:] but, for that non-dual [entity] having the nature of Brahman, for whom there is no proliferating [universe], there is nothing to be known (jñeya) and nothing to be ruled over (īśitavya), so knowledge (vijñāna) and sovereignty (aiśvarya) are unaccountable [in its case]. If it is so, scriptural passages [mentioning Brahman as] “omniscient” (sarvajña) and “lord of all” (sarveśvara) must be trusted blindly!–This is not true, because sovereignty is not brought about by those who are ruled over, nor is it the known [object] that makes one into a knower (jñātṛ). Quite the contrary! [Only] when the form of knowledge is established, or the ability to reign, may one attain a knowable [object] or assign tasks and govern those to be ruled over. [It all happens] as in the case of light (prakāśa) and burning (dāha): [only] when the sun’s luminosity is established may it shed light on [objects] to be illu-mined; its luminosity is not due to [there being] something to illumine, no more than a fire’s capacity to burn is due to [there being] something to burn. Thus, since the whole proliferating [universe] manifests itself only thanks to His consciousness, as stated in scriptural passages like “All this shines only through His splendour,”[126] “There is no other seer than Him [= the ātman],”[127] [Brahman] is [said to be] omniscient (sarvajña).

In this text, quite unique in Maṇḍana’s work, we see the lineaments of an alternative concept of omniscience, virtually escaping the objections raised in the ViV. Omniscience for Maṇḍana is not a matter of apprehending past and future, or perceiving the extremely large or extremely small; indeed it is not at all about knowing objects. Omniscience is understood here, in a negative way, as the absence of obstruction of the natural property of awareness (jñāna) pertaining to Brahman, inversely proportional to the presence of a multitude of knowable objects (jñeya).[128] As such it would be vain to ask for its cause, and the means to ascertain it is, of course, none but the eternal Veda. To put it differently, we can read in ViV 15–25 an attempt to release omniscience from its ties with the problem of dharma, which are as tight in the case of the Buddha as they are in the case of Īśvara for those who uphold Him. In that sense, his critique certainly contributes to Kumārila’s apologetic enterprise of (re)affirming the Veda as the one source of all religious authority. But at the same time it also paves the way for a reevaluation of omniscience as part of a Vedāntic soteriology, and thereby for its integration into the conceptual architecture of Uttara-Mīmāṃsā.

References and abbreviations

Bhatt 1989 G.P. Bhatt, The Basic Ways of Knowing. An In-depth Study of Kumārilas Contribution to Indian Epistemology. Delhi 1989.

Bhāvaśuddhi Bhāvaśuddhi of Ānandapūrṇamuni and Abhiprāyaprakāśikā of Citsukha-muni. Two Commentaries on Brahmasiddhi, ed. N.S. Anantakrishna Sastri. Madras 1963.

Brahmasiddhivyākhyā See BS.

Bṛhadāraṇyakopaniṣad Bṛhad-Āraṇyaka-Upaniṣad (traduite et annotée par Émile Se-nart). Paris 1967.

Bṛhatī Bṛhatī of Prabhākara Miśra (on the Mīmāṃsāsūtrabhāṣya of Śabarasvāmin) with Ṛjuvimalāpañcikā of Śālikanātha. Vol. 1.1, ed. S.K. Rāmanātha Śāstrī. Madras 1934.

BS Brahmasiddhi by Ācārya Maṇḍanamiśra with [the] Commentary by Śaṅkhapāṇi, ed. S. Kuppuswami Sastri. Madras 1937.

BṬ Bṛhaṭṭīkā (Kumārila). See TS.

Bühnemann 1980 G. Bühnemann, Der Allwissende Buddha. Ein Beweis und seine Pro-bleme. Ratnakīrtis Sarvajñasiddhi (übersetzt und kommentiert). Vienna 1980.

Chatterjee 1978 S. Chatterjee, The Nyāya Theory of Knowledge. A Critical Study of some Problems of Logic and Metaphysics. Calcutta1978.

David 2013 H. David, Action Theory and Scriptural Exegesis in Early Advaita-Vedānta (1). Maṇḍana Miśra on upadeśa and iṣṭasādhanatā. In: Scriptural Authority, Reason, and Action. Proceedings of a Panel at the 14th World Sanskrit Conference, Kyōto, September 1st–5th, 2009, ed. V. Eltschinger and H. Krasser. Vienna 2013.

David 2017 H. David, Une ontologie du commandement? Réflexions sur l’idée d’exis-tence, l’impératif et l’objet du Veda. ThéoRèmes 11 (2017). theoremes.1216

David (forthcoming) H. David, Action Theory and Scriptural Exegesis in Early Advaita-Vedānta (3). Maṇḍana Miśra on Ontology, Time and Commandment. In: A road less travelled, ed. V. Eltschinger, B. Kellner, and E. Mills.

Eltschinger 2005 V. Eltschinger, Recherches sur la philosophie religieuse de Dharmakīrti. I. Le Bouddha comme Śāstṛ et comme Sugata. Asiatische Studien/Études Asiatiques 59 (2005) 395–442.

Eltschinger 2007 V. Eltschinger, Penser lautorité des Écritures. La polémique de Dhar-makīrti contre la notion brahmanique orthodoxe dun Veda sans auteur. Autour de Pramāṇavārttika I.213268 et Svavṛtti. Vienna 2007.

Eltschinger 2008 V. Eltschinger, Śaṅkaranandana’s Sarvajñasiddhi. A Preliminary Report. In: Manuscripta Buddhica I. Sanskrit Texts from Giuseppe Tuccis Collection Part I, ed. F. Sferra. Rome 2008, 115–155.

Eltschinger 2009 V. Eltschinger, On the Career and the Cognition of Yogins. In: Yogic Perception, Meditation and Altered States of Consciousness, ed. E. Franco and D. Eigner. Vienna 2009, 169–213.

Eltschinger 2014a V. Eltschinger, Buddhist Epistemology as Apologetics. Studies on the History, Self-understanding and Dogmatic Foundations of Late Indian Buddhist Philosophy. Vienna 2014.

Eltschinger 2014b V. Eltschinger, The Four Nobles’ Truths and their 16 Aspects: On the Dogmatic and Soteriological Presuppositions of the Buddhist Epistemologists’ Views on Niścaya. Journal of Indian Philosophy 42 (2014) 249–273.

Eltschinger 2015 V. Eltschinger, Latest News from a Kashmirian “Second Dharmakīrti.” In: Cultural Flows across the Western Himalaya, ed. P. Mc Allister, C. Scherrer-Schaub, and H. Krasser. Vienna 2015, 303–363.

Eltschinger/Krasser/Taber 2012 V. Eltschinger, H. Krasser, and J. Taber, Can the Veda speak? Dharmakīrti against Mīmāṃsā exegetics and Vedic authority. An annotated translation of PVSV 164,24176,16. Vienna 2012.

Franco 2011 E. Franco, Perception of yogis–Some epistemological and metaphysical considerations. In: Religion and Logic in Buddhist Philosophical Analysis. Proceed-ings of the Fourth International Dharmakīrti Conference (Vienna, August 2327, 2005), ed. H. Krasser, H. Lasic, E. Franco, and B. Kellner. Vienna 2011, 81–98.

Franco 2012 E. Franco, Once Again on the Desires of the Buddha. In: Devadattīyam. Johannes Bronkhorst Felicitation Volume, ed. F. Voegeli, V. Eltschinger, D. Feller, M-P. Candotti, B. Diaconescu, and M. Kulkarni. Bern 2012, 229–245.

Frauwallner 1962 E. Frauwallner, Kumārila’s Bṛhaṭṭīkā. Wiener Zeitschrift für die Kunde Südasiens 6 (1962) 78–90.

Frauwallner 1968 E. Frauwallner, Materialien zur ältesten Erkenntnislehre der Karma-mīmāṃsā. Vienna 1968.

Goodman 1989 S.B. Goodman, A Buddhist Proof for Omniscience: TheSarvajñasiddhi” of Ratnakīrti. Unpublished PhD Thesis, Temple University 1989.

Griffiths 1990 P. Griffiths, Omniscience in the Mahāyānasūtrālaṃkāra and its Commen-taries. Indo-Iranian Journal 33 (1990) 85–120.

Harikai 1985 K. Harikai, Kumārila no Shakosun Kan (Kumārila’s views on Śākyamuni Buddha). Journal of the Nippon Buddhist Research Association 50 (1985) 49–65 (in Japanese).

Inami 1996 M. Inami, Puramanavarttika Puramanasiddi sho no kenkyu (5) (*A study of PV II [5]). Shimane Kenritsu Kokusai Tankidaigaku Kiyo 3 (1996) 77–100 (in Japanese).

Jackson 1991 R. Jackson, Dharmakīrti’s Attitude Toward Omniscience. In: Pt. Dalsukh Bhai Malvania Felicitation Volume I, ed. M.A. Dhaky and S. Jain. Varanasi 1991, 230–246.

Kajiyama 1998 Y. Kajiyama, An Introduction to Buddhist Philosophy: an annotated translation of the Tarkabhāṣā of Mokṣākaragupta. Vienna 1998.

Kataoka 2003a K. Kataoka, Kumārila’s Critique of Omniscience. Indo Shisōshi Kenkyū 15 (2003) 35–69.

Kataoka 2003b K. Kataoka, The Mīmāṃsā Definition of pramāṇa as Source of New Information. Journal of Indian Philosophy 31 (2003) 89–103.

Kataoka 20111/20112 K. Kataoka, Kumārila on Truth, Omniscience, and Killing. Part 1: A Critical Edition of Mīmāṃsāślokavārttika ad 1.1.2 (codanāsūtra); Part 2: An Annotated Translation of Mīmāṃsāślokavārttika ad 1.1.2 (codanāsūtra). 2 vols. Vienna 2011.

Katsura 1984 Sh. Katsura, Dharmakīrti’s Theory of Truth. Journal of Indian Philosophy 12 (1984) 215–235.

Kawasaki 1992 Sh. Kawasaki, Issaichi-shiso no kenkyu (*A Study of the Omniscient Being [sarvajña] in Buddhism). Tokyo 1992 (in Japanese).

Krasser 2001 H. Krasser, On Dharmakīrti’s Understanding of pramāṇabhūta and His Def-inition of pramāṇa. Wiener Zeitschrift für die Kunde Südasiens 45 (2001) 173–199.

Krasser 2012 H. Krasser, Bhāviveka, Dharmakīrti and Kumārila. In: Devadattīyam. Jo-hannes Bronkhorst Felicitation Volume, ed. F. Voegeli, V. Eltschinger, D. Feller, M-P. Candotti, B. Diaconescu, and M. Kulkarni. Bern 2012, 535–594.

La Vallée Poussin 1902 L. de La Vallée Poussin, On the Authority of Buddhist Āgamas. Journal of the Royal Asiatic Society 34.2 (1902) 363–376.

Lindtner 2001 Ch. Lindtner, Bhavya on Mīmāṃsā. Mīmāṃsātattvanirṇayāvatāraḥ.

Madras (Chennai) 2001.

Madhyamakahṛdayakārikā See Lindtner 2001.

McClintock 2010 S.L. McClintock, Omniscience and the Rhetoric of Reason. Śāntarakṣi-ta and Kamalaśīla on Rationality, Argumentation and Religious Authority. Boston 2010.

MīSū Mīmāṃsāsūtra. See Frauwallner 1968.

Miyasaka 1972 Y. Miyasaka, Pramāṇavārttika-kārikā (Sanskrit and Tibetan). Acta Indo-logica 2 (1971/72) 1–206.

Moriyama 2014 Sh. Moriyama, Omniscience and Religious Authority. A Study on Prajñā-karaguptas Pramāṇavārttikālaṅkārabhāṣya ad Pramāṇavārttika 2.8–10 and 29–33. Berlin/Zürich 2014.

Much 1991 M.T. Much, Dharmakīrtis Vādanyāya (Teil 1: Sanskrit-Text). Vienna 1991.

Natarajan 1995 K. Natarajan, The Vidhi Viveka of Maṇḍana Miśra: understanding Vedic Injunctions. Delhi 1995.

NB Nyāyabindu (Dharmakīrti). In: Dharmottarapradīpa (being a sub-commentary on Dharmottaras Nyāyabinduṭīkā, a commentary on Dharmakīrtis Nyāyabindu), ed. Dalsukhbhai Malvania. Patna 1971.

NM1/NM2 Nyāyamañjarī of Jayanta Bhaṭṭa, with Ṭippaṇī–Nyāyasaurabha by the editor, ed. K.S. Varadācārya. 2 vols. Mysore 1969 and 1983.

Nyāyabinduṭīkā See NB.

Nyāyavārttikatātparyaṭīkā Nyāyavārttikatātparyaṭīkā of Vācaspatimiśra, ed. Anantalal Thakur. New Delhi 1996.

Nyāyasūtra/Nyāyabhāṣya Gautamīyanyāyadarśana with Bhāṣya of Vātsyāyana, ed. Anan-talal Thakur. New Delhi 1997.

NyK Nyāyakaṇikā. See ViV/ViVK (S).

Ono 2000 M. Ono, Prajñākaraguptas Erklärung der Definition gültiger Erkenntnis (Pramāṇavārttikālaṃkāra zu Pramāṇavārttika II 1–7) (Materialien zur Definiti-on gültiger Erkenntnis in der Tradition Dharmakīrtis 3). Teil 1: Sanskrit-Text und Materialien. Vienna 2000.

Pandey 1972 L. Pandey, The Buddhist Conception of Omniscience. Unpublished PhD Thesis, McMaster University, Hamilton (Ontario) 1972.

Pemwieser 1991 M. Pemwieser, Materialien zur Theorie der yogischen Erkenntnis im Buddhismus. Unpublished M.A. Thesis, University of Vienna 1991.

Potter 2014 K.H. Potter, Philosophy of Pūrva-Mīmāṃsā. Delhi 2014.

Prakīrṇaprakāśa Vākyapadīya of Bhartṛhari with the Commentary of Helārāja. Kāṇḍa III. Vol. 1, ed. K.A. Subrahmania Iyer. Poona 1963.

PS Dignāga’s Pramāṇasamuccaya, Chapter 1. A hypothetical reconstruction of the San-skrit text with the help of the two Tibetan translations on the basis of the hitherto known Sanskrit fragments and the linguistic materials gained from Jinendrabuddhi’s Ṭīkā, ed. E. Steinkellner. Vienna 2005. KGA/PDF/forschung/buddhismuskunde/dignaga_PS_1.pdf, last visited 10-12-2019.

PV 1/PVSV The Pramānavārttikam of Dharmakīrti. The first Chapter with the Autocom-mentary, ed. R. Gnoli. Rome 1960.

PV 2–4 See Miyasaka 1972.

PVA See Moriyama 2014.

PVAS Pramāṇavārtikabhā[ṣ]yam or Vārtikālaṅkāraḥ of Prajñākaragupta (being a com-mentary on Dharmakīrti’s Pramāṇavārtikam), ed. R. Sāṅkrityāyana. Patna 1953.

PVin/PVinK 1–2 Dharmakīrti’s Pramāṇaviniścaya. Chapters 1 and 2, ed. E. Steinkellner. Beijing/Vienna 2007.

Ruegg 1994 D. Seyfort Ruegg, Pramāṇabhūta, *pramāṇa(bhūta)-puruṣa, pratyakṣadhar-man and sākṣātkṛtadharman as Epithets of the ṛṣi, ācārya and tathāgata in Gram-matical, Epistemological and Madhyamaka Texts. Bulletin of the School of Oriental and African Studies, University of London 57.2 (1994) 303–320.

ŚlV Mīmāṃsāślokavārtikam [Mīmāṃsāślokavārtika]. śrīmatkumārilabhaṭṭapādaviracitam (…)śrīmatpārthasāra-thimiśrapraṇītayā nyāyaratnākarākhyayā vyākhyayānugatam, ed. Mānavallyupāhva-tailaṅgarāmaśāstrī. Benares (Kāśī) 1898.

ŚlV (codanā°) See Kataoka 20111.

SSiR Sarvajñasiddhi (Ratnakīrti). In: Ratnakīrtinibandhāvaliḥ (Buddhist Nyāya Works of Ratnakīrti), ed. Anantalal Thakur. Patna 1975, 1–31.

SSiŚ Sarvajñasiddhi (Śaṅkaranandana). See Eltschinger 2008.

Stcherbatsky 1926 Th. Stcherbatsky, Über die Nyāyakaṇikā des Vācaspatimiśra und die indische Lehre vom kategorischen Imperativ. In: Beiträge zur Literaturwissenschaft und Geistesgeschichte Indiens. Festgabe Hermann Jacobi zum 75. Geburtstag, ed. W. Kirfel. Bonn 1926, 369–380.

Stcherbatsky 19301/19302 Th. Stcherbatsky, Buddhist Logic. 2 vols. Leningrad 1930.

Steinkellner 1967 E. Steinkellner, Dharmakīrtis Hetubinduḥ. Teil II: Übersetzung und Anmerkungen. Vienna 1967.

Steinkellner 1977 E. Steinkellner, Jñānaśrīmitra’s Sarvajñasiddhiḥ. In: Prajñāpāramitā and related systems: Studies in honor of Edward Conze, ed. L. Lancaster and L. Gómez. Berkeley 1977, 383–393.

Steinkellner 1978 E. Steinkellner, Yogische Erkenntnis als Problem im Buddhismus. In: Transzendenzerfahrung, Vollzugshorizont des Heils, ed. G. Oberhammer. Vienna 1978, 121–134.

Steinkellner 20131/20132 E. Steinkellner, Dharmakīrtis frühe Logik. Annotierte Überset-zung der logischen Teile von Pramāṇavārttika 1 mit der Vṛtti (Introduction, Überset-zung, Analyse). 2 vols. Tokyo 2013.

Steinkellner/Much 1995 E. Steinkellner and M.T. Much, Texte der erkenntnistheoreti-schen Schule des Buddhismus. Göttingen 1995.

Stern 1988 E. Stern, Vidhivivekaḥ of Maṇḍana Miśraḥ, with commentary Nyāyakaṇikā of Vācaspatimiśraḥ and supercommentaries Juṣadhvaṅkaraṇī and Svaditaṅkaraṇī of Parameśvaraḥ [Parameśvara]. Critical and annotated edition: the pūrvapakṣaḥ (Sanskrit text). Unpublished PhD Thesis, University of Pennsylvania 1988.

SV svavṛtti

Taber 2005 J.A. Taber, A Hindu Critique of Buddhist Epistemology. Kumārila on Percep-tion. TheDetermination of Perceptionchapter of Kumārila Bhaṭṭas Ślokavārttika. Translation and Commentary. London/New York 2005.

Taber 2011 J. Taber, Did Dharmakīrti think the Buddha had desires? In: Religion and Logic in Buddhist Philosophical Analysis. Proceedings of the Fourth International Dharmakīrti Conference (Vienna, August 2327, 2005), ed. H. Krasser, H. Lasic, E. Franco, and B. Kellner. Vienna 2011, 437–448.

Tantravārttika Śrīmajjaiminipraṇīt[aṃ] mīmāṃsādarśan[am]. Vol. 2 (ad MīSū 1.2–2.1), ed. Gaṇeśaśāstrī Ambādāsa Jośī. Poona 1981.

Tātparyaṭīkā Ślokavārttikavyākhyā Tātparyaṭīkā of Umveka Bhaṭṭa, ed. S.K. Rāmanātha Śastrī. Madras 1971.

TS Tattvasaṅgraha of Ācārya Śāntarakṣita, with the commentary Pañjikā of Śrī Kamalaśīla, ed. Dwarikadas Shastri. 2 vols. Benares (Varanasi) 1968.


Vādanyāya See Much 1991.

Vākyapadīya (SV) svavṛtti Vākyapadīya of Bhartṛhari with the Commentaries Vṛtti and Paddhati of Vṛṣabhadeva. Kāṇḍa I, ed. K.A. Subrahmania Iyer. Poona 1966.

Vetter 1964 T. Vetter, Erkenntnisprobleme bei Dharmakīrti. Vienna 1964.

ViV/ViVK M = Vidhivivekaḥ śrīmadācāryamaṇḍanamiśraviracitaḥ pūjyapādaśrīmadvā-caspatimiśranirmitayā nyāyakaṇikākhyayā vyākhyayā samalaṃkṛtaḥ, ed. Mānava-llyupāhvatailaṅgarāmaśāstrī. Benares 1907 (reprint from “The Pandit”).

  G = Vidhi-vivekaḥ of Śrī Maṇḍana Miśra, with the Commentary Nyāyakaṇikā of Vācaspati Miśra, ed. Mahaprabhu Lal Goswami. Benares 1978. S = Stern 1988.

Watanabe 1987 Sh. Watanabe, Śubhagupta’s Sarvajñasiddhikārikā. Journal of the Nari-tasan Institute for Buddhist Studies 10 (1987) 55–74 (in Japanese).

Watanabe 1988 Sh. Watanabe, Tattvasaṃgraha XXVI, kk. 3247–3261 and kk. 3622–3646. Journal of the Naritasan Institute for Buddhist Studies 11 (1988) 501–533 (in Japanese).

Yaita 1985 H. Yaita, On anupalabdhi, annotated translation of Dharmakīrti’s Pramāṇavā-rttikasvavṛtti (I). Taishō Daigaku Daigakuin Kenkyū Ronshū 9 (1985) 216–199.

Footnotes and references:


On the early history of the concept of omniscience in Buddhism, see in particular Pandey 1972, Griffiths 1990 and Kawasaki 1992. For an overview in English, see McClintock 2010: 23–35. As she notes, Buddhist attitudes toward omniscience might have been far more ambivalent than is the case in Jainism, especially in the early period. On this point, see also Jackson 1991: 230–232.


For a synthesis on the question, see Eltschinger (2009), who discusses relevant passages of PS (1.6cd), PVin 1 (27.7–28.8) and PV (3.281–286). Further references in Eltschinger 2009: 191 (n. 94).


By this I do not mean to say that Dharmakīrti, in particular, did not believe in the Buddha’s omniscience, which is obviously not the case. His conception of yogipratyakṣa as a “vision of the [four] Nobles’ Truths” (āryasatyadarśana –see below § 3.5) also implies a form of ‘focused’ omniscience like the one defined in PV 2.29–33. Nevertheless, omniscience did not constitute a major philosophical or religious issue for him as it would for later Buddhist authors. As pointed out by Moriyama (2014: 17), “the concept of omniscience does not occupy a special place in Dharmakīrti’s argument for establishing the Buddha as pramāṇabhūta,” a point that sharply contrasts with later understandings of this attribute of the Buddha. See also Franco 2011: 89 and Eltschinger (2005: 429–434), who explains this state of affairs in terms of a conscious apologetic strategy: “Dharmakīrti aura élaboré une structure doctrinale assez ouverte pour que les docteurs bouddhistes y lisent l’adhésion du maître à l’omniscience, mais assez implicite pour que les pourfendeurs de ce dogme ne puissent lui en faire le reproche” (p. 434). Equally significant is Dharmakīrti’s lack of interest for God’s omniscience in the section of the PV refuting the existence of īśvara (PV 2.8–28). Specialists of Dharmakīrti still disagree as to what his final position regarding total omniscience might have been, in particular in the case of the Buddha. See Jackson 1991: 232–234, Eltschinger 2005: 434 and McClintock 2010: 135–138. The main passage under discussion (PVin 2.55) is however unrelated to the definition of yogic perception formulated in PVin 1 (see preceding note).


On Śāntarakṣita and Kamalaśīla’s discussions of omniscience, see McClintock 2010. On Prajñākara-gupta, see Moriyama 2014 and below § 4. Another early and potentially important document is the (mostly unstudied) Sarvajñasiddhi of Śubhagupta (720–78), preserved only in Tibetan (see Bühnemann 1980: viii and Watanabe 1987).


For a preliminary edition and analysis of Śaṅkaranandana’s Sarvajñasiddhi (kārikās only), see Eltschinger 2008. See also Eltschinger (2015: 323), who mentions, besides the Sarvajñasiddhi, a shorter work called Sarvajñasiddhisaṃkṣepa, still unpublished. A study of fragments of Jñānaśrīmitra’s otherwise lost Sarvajñasiddhi is found in Steinkellner 1977. For an in-depth study of Ratnakīrti’s treatise of the same name, see Bühnemann 1980 and Goodman 1989. On the possible contribution of Jitāri (9th–10th c.?) to this debate, see Bühnemann 1980: viii and Eltschinger 2008: 142.


Kumārila’s date is established in relation to that of Dharmakīrti, of whom he might have been an elder contemporary. The date 600–650 proposed by Kataoka (20112: 112) on the basis of Frauwallner’s dating of Dharmakīrti (600–660) may have to be revised in case the great Buddhist logician should be placed a few decades earlier, as proposed by Krasser (2012).


The numbering of verses in Śāntarakṣita’s Tattvasaṃgraha is that of Swami Dwarikadas Shastri’s edition (Benares, 1968) (= TS), which slightly differs from that of Embar Krishnamacharya’s earlier edition (Baroda, 1926, reprinted in 1988). For a detailed study of Kumārila’s critique of omniscience, see Kataoka 2003a and Kataoka 2011. Equally central to this debate are vv. 26–33 of the pratyakṣa-section of the ŚlV (ad Śābarabhāṣya 1.1.4), dealing with the perception of yogins (translation in Taber 2005: 54–56). Omniscience is again alluded to in ŚlV (saṃbandhākṣepaparihāra) 44ab and 114ab, but these are simple reminiscences of the codanās ection, adding no new arguments.


Only on one occasion does Kumārila allude to omniscience as conceived by the Jains (ŚlV [codanā°] 141–142). No such allusion is found in the BṬ. The question of God’s omniscience is not touched upon by Kumārila, and seems to have been introduced into Mīmāṃsā by Maṇḍana Miśra (see below § 1.3).


Other Mīmāṃsakas of the period might have been interested in the question of the Buddha’s omniscience, but we do not have much evidence for this. The question whether Bhavya/Bhā(va)viveka, in the ninth chapter of his Madhyamakahṛdayakārikās (9.15–16), refers to Kumārila’s views or to those of an earlier Mīmāṃsaka has been raised by Krasser (2012: 559–568), following a remark by Lindtner (2001: 3). I fully agree with Krasser that “one can easily read Bhāviveka as refuting Kumārila” (p. 565), but the passage in question is too brief to say much more. Krasser’s assumption (p. 567) that Bhavya targets an early Mīmāṃsaka different from Kumārila while evoking his opponent’s belief in Jaimini’s omniscience (Madhyamakahṛdayakārikā 9.163 and Tarkajvālā thereon; unavailable in Sanskrit, text and translation: Krasser 2012: 566) requires additional caveats, for it holds only assuming Bhavya is perfectly accurate in his critiques or always relies on a Mīmāṃsaka source, which is far from certain. A further unsolved case is that of the two Mīmāṃsakas Sāmaṭa and Yajñaṭa (see McClintock 2010: 155–156, 225, 356–59). These two enigmatic figures are known exclusively through their mention by Kamalaśīla, who ascribes to them a series of opinions reported by Śāntarakṣita in TS 3246–60 (Japanese translation: Watanabe 1988). See TSP 1020.16–17: sāṃprataṃ sāmaṭayajñaṭayor matena punar api sarvajñadūṣaṇam āha; “Now, he exposes yet another refutation of an omniscient [being], following the view of Sāmaṭa and Yajñaṭa.” Some scholars suspected that these could be forged names, made up in reference to the Sāma-and Yajurveda (Kawasaki 1992: 255), but the fact remains that the opinions associated with them cannot easily be traced to any known mīmāṃsaka text.


For a brief synopsis of the section, see Stern 1988: 28–45. The recent summary of the ViV by Potter (2014: 289–295) unfortunately ignores that important section of the work, as did the small monograph by Natarajan (1995) on which it is based. The only study of that part of the ViV-‘corpus’ so far is the German translation of a fragment of Vācaspati Miśra’s NyK (ad ViV 15) dealing with yogic perception by M. Pemwieser (1991). On this fragment, see also Steinkellner 1978. M. Biardeau’s unpublished French translation of the whole pūrvapakṣa of the ViV and NyK, now kept at the archives of the Collège de France in Paris, naturally includes a translation of ViV 15–25 and the corresponding NyK.


See David 2013: 281, n. 31.


BS 2.27cd–28 (84.9–10).


Although Vācaspati does not mention the Buddha by name, his characterisation of the Omniscient in the NyK (445.5–446.6) has a definite Buddhist ring to it: for Vācaspati, the Omniscient is the “Blessed Doctor” (bhagavān bhiṣaj –445.5), “the Blessed One, whose all-pervasive compassion has become his intimate goal, who has achieved the ultimate degree of detachment, untouched in the least by [main] defilements such as desire or by minor defilements such as excitement or pride” (svārthībhū-taviśvavyāpikāruṇyo bhagavān vairāgyātiśayasaṃpanno mātrayāpi rāgādibhiḥ kleśair upakleśaiś ca madamānādibhir aparāmṛṣṭaḥ [445.10–446.2]). As Stern (1988: 997) rightly points out, the description of the NyK has an almost exact parallel in Vācaspati’s Nyāyavārttikatātparyaṭīkā ad Nyāyasūtra 2.1.68 (384.14–19, especially 384.16–19, which corresponds almost word for word to NyK 446.2–6), dealing with the Naiyāyika definition of the āpta (see following note). There, we see the very same characteristics applied to “the maker of products such as the body, the world, etc.” (tanubhuvanādilakṣaṇasya kāryasya kartā), in other words to God, said to be “the knower of the real nature of all things” (samastavastuta-ttvajña –384.15)–i.e. omniscient–, “untouched by defilements and the store of the maturation of [past] deeds” (kleśakarmavipākāśayāparāmṛṣṭa –384.15) and “extremely compassionate” (paramakāruṇika –384.15). This striking similarity between the two descriptions might have motivated the assumption by Stcherbatsky (1926) that Maṇḍana as well, in the ViV, is discussing the omniscience of ī śvara. I find it unlikely, though, that Vācaspati should say in the NyK that God “achieved the ultimate degree of detachment” (vairāgyātiśayasaṃpanna –NyK 445.10–446.1), a quality which, as far as I can see, only fits a human being. This, together with the insertion immediately thereafter in the NyK of a long defence of omniscience directly quoting Buddhist sources (NyK 447.2–458.7), makes me think that Vācaspati, at least, is really speaking here of the Buddha.


ViV 15 (introduction): anuvidheyo niyoktā bhūtānāṃ sākṣātkṛtānuśravikaśreyaḥsādhanoa hitakāmaḥ sarvajñaḥ (S 445.1–448.1 [≈ M 110.2–4/G 78.3–4]). a °anuśravika° S G: °anubhavika° M. This description, in itself, has nothing specifically Buddhist. Of the four main characteristics of the omniscient outlined in this passage (niyoktṛ[tva], sākṣātkṛta[]sādhana[tva], hitakāma, sarvajña[tva]), the first directly follows from the context of the ViV (see below § 2.1). The second and third may, of course, refer to the first two epithets of the Buddha in the famous opening verse of Dignāga’s Pramāṇasamuccaya (PS 1.1a): pramāṇabhūta (“authoritative”) and jagaddhitaiṣin (“seeking the benefit of the world”). But Maṇḍana’s formulation of the second attribute–sākṣātkṛtānuśravikaśreyaḥsādhana (“having directly perceived the means for realizing the Supreme Good as prescribed in the Scriptures”)–also recalls the expression sākṣātkṛtadharma(n) (“having directly perceived the dharma(n)”) used to qualify Vedic ṛṣis in the Nirukta (1.20), in Bhartṛhari’s Vākyapadīya (1.5–SV 24.2) and in Bhavabhūti’s Uttararāmacarita (see Ruegg 1994: 307–308). As is well-known, the quality of being sākṣātkṛtadharma(n) is also part of Vātsyāyana/Pakṣilasvāmin’s definition of an āpta (“reliable speaker”) in the Bhāṣya ad Nyāyasūtra 1.1.7 and 2.1.68 (14.4 and 96.16). Pakṣilasvāmin’s commentary in sūtra 2.1.68 also mentions two additional qualities of the āptabhūtadayā (“compassion for beings”) and yathābhūtārthacikhyāpayiṣā (“desire to communicate about a real object”) (96.17)–, the first of which could very well be hinted at by Maṇḍana while speaking of the Omniscient’s compassion toward others. Thus, although several external clues plead for an identification of Maṇḍana’s omniscient being with the Buddha, one must keep in mind that his description remains quite unspecific and corresponds to what most philosophical traditions of his time would have expected of a reliable teacher, human or divine.


As Stern (1988: 28) rightly remarks, the section referred to by Vācaspati as bauddhasarvajñaparīkṣā (“examination of the Omniscient [as conceived] by the Buddhists”) on two occasions in the NyK (612.7–8 and 634.7, ad ViV 17) is certainly ViV 15.


Several of them are mentioned in Vācaspati’s commentary: “someone with a whiff of Prābhākara[-doctrine]” (kaś cit prābhākaragandhī –NyK 570.10, ad ViV 16), Naiyāyikas (NyK 606.7–8, ad ViV 17; 679.9, ad ViV 21), “upholders of [the doctrine] of Svayaṃbhu [i.e. Patañjali’s Yoga]” (svāyaṃbhuvaḥ –NyK 627.14, ad ViV 21). These identifications however testify to Vācaspati’s attempt to read in the ViV a refutation of various ‘doctrines,’ while Maṇḍana’s progression is dialectic much more than doxographic, so they should be taken with much caution.


See Moriyama (2014: 37), who also notes Maṇḍana’s influence on immediately later Buddhist thinkers on that topic, in particular on Prajñākaragupta.


See Kataoka (20112: 11 and n. 8), who reports the view of Harikai (1985: 63) on a possible evolution in Kumārila’s attitude towards Buddhism (less ‘logical,’ more ‘emotional’) between the ŚlV and the TV. On this topic, see also Eltschinger 2014a: 66.


See La Vallée Poussin 1902, Eltschinger 2007: 38–46 and Eltschinger 2014a: 66–70.


See ŚlV (codanā°) 110cd–111, BṬ (= TS 3127), Kataoka 2003a: 42–43 and Kataoka 20112: 320–324. On Kumārila’s understanding of dharma, see ŚlV (codanā°) 13–14, translated in Kataoka 20112: 206–209. As explained in these verses, actions, substances and qualities are not considered to be dharma(s) in themselves, but only in so far as they are conducive to an expected result.


Modern readers of Dharmakīrti’s statements on omniscience in PV 2.32–33 have rarely noticed how close he actually stands to Kumārila on this point. Thus I would not necessarily interpret these verses of the PV as a ‘response’ from Dharmakīrti to Kumārila (as suggested by Kataoka [20112: 321]), but rather as an essential agreement of both thinkers on the main point at stake: what matters is not the possibility for a human being to know everything, but his capacity to access ethically/soteriologically relevant matters by means of perception. As pointed out by Kataoka (20112: 321), the distinction between sarvajña and dharmajña is introduced by Kumārila only in the BṬ. Its use by Maṇḍana in the prose introduction to ViVK 25 (S 733.4) might therefore point to his familiarity with Kumārila’s lost work.


ViV 15 (introduction)–S 445.1 (translated above § 1.3).


See ViV 12–14, David 2017 and David (forthcoming).


ViV 14: pravartanāmātraṃ niyogaḥ, sa ca liṅādyartha iti pravartito’hama iti pratipattiḥ śabdāt. karta-vyatāvagamas tu niyoktur anuvidheyatvāt, anuvidheyaniyoge mamedaṃ kartavyam ity adhyavasāyāt, itaratra tu pravartanāmātrapratīteḥ. uktaṃ ca: kartavyatāviṣayo hib niyogaḥ, na niyogaḥ kartavyatām āha (S 441.5–442.3 [≈ M 108.5–109.4/G 77.3–6]). a pravartito’ham S; pravartito’ham atra M G; b hi S: Ø M G.


A more literal translation would be “the awareness of an obligation.” As Vācaspati makes clear (NyK 442.9–10), the difference between a mere instigation (pravartanāmātra) and a proper obligation (karta-vyatā) is that only the second can be the cause of an activity (pravṛttihetu) for a rational agent.


Bṛhatī 1.1.2 (38.8–9).


ViV 15: yady apy ātmapratyakṣanivṛttir viprakarṣavatām abhāvaṃ vyabhicarati, sarvapratyakṣanivṛttir asiddhāsarvadṛśaḥ; sakalapramāṇanivṛttyā ca nārthābhāvasiddhiḥ, avyāpakanivṛttāva avyāpyanivṛtter aniyamāt, avyāpakatvaṃ ca, ahetutvāt, tanmātrapratibandhābhāvācb ca, anyathā sarvasya sarvadarśi-tvaprasaṅgaḥ, aviśeṣātc; vacanādayaś ca yady apy avirodhād anivartakāḥ, tathāpi kāraṇanivṛttyā kāryanivṛttipratīteḥ pramāṇānupapattyā tatkāryāyāḥ sarvārtheṣu saṃvido’bhāvam anumimīmahe (S 459.1–461.2 [≈ M 115.1–116.2/G 81.4–82.6]). a °nivṛttau S: °nivṛttyā M G: °nivṛttyā ca Mss (S); b °pratibandha° S: °anubandha° M G; c aviśeṣāt S: aviśeṣāc ca M G.


Literally: the “cessation” or “non-activity,” as opposed to pravṛtti (“activity”). The term is used by Dharmakīrti in a similar context in expressions like pramāṇatrayanivṛtti (“the absence of the three means of knowledge”–PVSV 102.1), also found in Maṇḍana’s text. See Yaita (1985: 215): “the cessation of the three means of knowledge;” Steinkellner (20131: 81) “das Auffallen der drei Erkenntnismittel;” Eltschinger/Krasser/Taber (2012: 9, n. 9): “the silence of the three means of valid cognition.” This last translation is preferable, in my opinion, as the English term “cessation” implies previous activity, which is not always the case of Dharmakīrti and Maṇḍana’s use of the word nivṛtti. See for instance Maṇḍana’s (or rather, his opponent’s) definition of pleasure (sukha) as “the absence of pain” (duḥkhanivṛtti) in the Brahmasiddhi (BS 1.1 [1.17]), which does not imply the previous existence of pain. I opt for a plainer translation (“absence”) only to avoid confusion between a proper use of the word “silence” (in the case of Scripture) and a metaphorical one (in the case of perception and inference).


According to Vācaspati (NyK 459.6–7), “distance” (viprakarṣa) is threefold: in space (deśa), time (kāla) and nature (svabhāva). The same tripartition is found in chapter 2 of Dharmakīrti’s Pramāṇaviniścaya (PVin 64.9–10) and in the Nyāyabindu (NB 2.27); the Pramāṇavārttika (PVSV 102.6–7, ad PV 1.199) only mentions deśa and kāla.


Both published editions of the ViV suggest that asarvadṛśaḥ should be read with what follows, not with what precedes, a solution also adopted by M. Biardeau in her unpublished translation: “(…) [quoi]que l’absence de perception de tous ne soit pas établie, que l’on ne puisse établir la non-existence d’un objet par l’absence de tous les moyens de perception chez quelquun qui nest pas omniscient (…)” (Ms. f. 123r). Although neither Stern’s edition nor Vācaspati’s commentary plead for either solution, I find it easier to link the genitive with what precedes, since the position of ca would otherwise be odd. The following argument (2) is directly borrowed from Dharmakīrti, as we shall see (§ 2.3), and stands perfectly well on its own. On the other hand, it makes sense to argue that only an omniscient would be able to establish a universal absence of perception concerning a given object.


In other words: the inference of the absence of a pervaded (vyāpya) property from the absence of the pervading (vyāpaka) property is valid, as when we conclude to the absence of smoke (= vyāpya) on the hill from the absence of fire (= vyāpaka): vyāpakābhāva → vyāpyābhāva (“a → b” = “valid inference from a to b”). But this inference would not be valid if both properties were not in a relation of pervasion (vyāpti), or if the pervasion was the other way round. Thus the inference of the absence of fire (= vyāpaka) on the hill from the absence of smoke (= vyāpya) is not valid, for there are exceptions to this conclusion. In the present case, the question raised is whether one can correctly conclude to the absence in the world of an omniscient being–the object to be known (prameya)–from the absence of a means to know (pramāṇa) such a being (pramāṇābhāvaprameyābhāva?). It follows from what precedes that the inference is valid only if pramāṇa and prameya stand in a relation of pervasion (vyāpti), in other words if we could correctly conclude to the existence of a means of knowledge from the existence of the object of knowledge (prameya → pramāṇa?). Maṇḍana will now show that this is not the case, by excluding the two only possible grounds for pervasion or invariable concomitance in Dharmakīrti’s system: causality (tadutpatti) and identity (tādātmya) (argument 2AI).


See NyK 459.11–460.1: ahetutvād akāraṇatvāt.


The compound tanmātrapratibandha is equivalent to the expression tanmātrānubandha/°saṃbandha used by Dharmakīrti on several occasions to speak of the relation underlying the use of an “essential [inferential] reason” (svabhāvahetu). See for instance PVSV 6.26, 17.20, 18.1 and 18.21 (°anubandha), PVSV 16.28 (°saṃbandha). See also NyK 460.3–5: hetudharmamātrānubandhī hi sādhyadharmas tasya vyāpakaḥ, yathā vṛkṣatvaṃ śiṃśapātvasya; “For when the property to be proved (a) merely depends on the property which is the [inferential] reason (b), it (= a) pervades the other (= b); for instance, the quality of being a tree [pervades in this manner] the quality of being a śiṃśapā.” The reading °anubandha° (instead of °pratibandha°), found in all Mss of the ViV and in some Mss of the NyK (see Stern 1988: 1023–1024), is therefore equally plausible.


That is: if there was an essential connection between the existence of the object to be known (prameya) and that of a means for knowing it (pramāṇa).


For a possible paraphrase of this difficult argument by Prajñākaragupta, see below § 4.5.


On kāraṇānupalabdhi, see NB 2.39: kāraṇānupalabdhir yathā nātra dhūmaḥ, vahnyabhāvād iti; “[Es-tablishment of the non-existence of the effect through] non-apprehension of [its] cause is for instance: ‘Here, there is no smoke, for there is no fire’” (p. 135). In his commentary, Dharmottara remarks that this particular use of anupalabdhi is restricted to cases where “the effect, even if it existed, would not be perceived” (kāryaṃ sad apy adṛśyam bhavatiNyāyabinduṭīkā 136.1), which is obviously the case of an omniscient being. Although neither Maṇḍana nor Vācaspati uses exactly the term kāraṇānupalabdhi, its equivalent kāraṇānupalambha is found in a pūrvapakṣa of Ratnakīrti’s Sarvajñasiddhi (SSiR 7.18), which presents a reasoning identical in substance to Maṇḍana’s. The parallel between these two passages would certainly require further exploration.


This passage has been translated twice, into English by Yaita (1985) and, more recently, into German by Steinkellner (20131 and 20132). I am much indebted to the latter’s translation and rich annotation of this text. Strictly speaking, Dharmakīrti’s discussion of anupalabdhi is much longer, finding its conclusion only with PV 1.339 (thus practically with the end of the first chapter), including also his lengthy digression on the authority of Scripture (āgama). See Eltschinger/Krasser/Taber 2012: 9, n. 9. I am essentially concerned here with the initial part of this section.


Dharmakīrti remains quite vague in PV(SV) 1.198–204 about objects whose existence cannot be disqualified by mere silence of Scripture, like “particular things (…) which are far away in time and space” (deśakālavyavahitāḥ [] dravyaviśeṣāḥ –PVSV 102.6–7). He is more precise about objects whose non-existence cannot be proved by the mere fact that we cannot infer them: “a mind free of passion” (viraktaṃ cetas –PVSV 103.4), “a particular deity” (devatāviśeṣaibid.), “the capacity of intentions [relative to] gifts and refraining from violent action to cause happiness” (dānahiṃsāviraticetanānām abhyudayahetutā –PVSV 103.5; my translation of the compound in the genitive relies entirely on its interpretation by Yaita [1985: 213] and Steinkellner [20131: 84]). The closest approximation we find in the Pramāṇavārttika to Maṇḍana’s reasoning is found in PV(SV) 1.311 (I thank V. Eltschinger for drawing my attention to this important passage). In that portion of the SV, which forms a sort of ‘coda’ to his discussion on mantras, Dharmakīrti discusses possible objections against the assumption of an “extraordinary person” (puruṣātiśaya) who could be the author of mantras on the basis of his “humanity” (puṃstva), a property which must itself be inferred from his possessing an intellect (buddhi), senses (indriya) and speech (vacana) (see PVSV 164.15). His main response is clearly similar to Maṇḍana’s principal argument: na hy atīndriyeṣv ataddarśinaḥ pratikṣepaḥ saṃbhavati, satām apy eṣam ajñānāt; “Those who do not see supersensible [objects] cannot confute (prati-kṣip) them, for even if they exist, they will not know them” (PVSV 164.17–18 [I do not translate hi]).


NB 3.69–71: yathāsarvajñaḥ kaś cid vivakṣitaḥ puruṣo rāgādimān veti sādhye vaktṛtvādiko dharmaḥ saṃdigdhavipakṣavyāvṛttikaḥ, sarvajño vaktā nopalabhyata ity evaṃprakārasyānupalambhasyādṛśyā-tmaviṣayatvena saṃdehahetutvāt. tato’sarvajñaviparyayād vaktṛtvāder vyāvṛttiḥ saṃdigdhā. vaktṛ-tvasarvajñatvayor virodhābhāvāc ca yaḥ sarvajñaḥ sa vaktā na bhavatīty adarśane’pi na sidhyati, saṃdehāt; “If what must be proved is, for instance, that a certain intended person is non-omniscient (asarvajña), or is passionate, etc. (rāgādimant), one can doubt that a property like being a speaker (vaktṛtva), etc. [establishing that conclusion] is absent from the negative instance [i.e. an omniscient person]. For a non-apprehension like ‘we do not apprehend any omniscient speaker’ is the cause of a [mere] doubt, for it is about [an object] whose nature cannot be seen (adṛśyātman). Therefore, the absence of [the property of] being a speaker, etc. in [someone] other than a non-omniscient [i.e. in an omniscient person] is subject to doubt. Moreover, since being a speaker and being omniscient are not in contradiction (virodha), [the rule that] ‘whoever is omniscient does not speak’ is not established, even when we do not see [anybody who is omniscient and speaks], for this is subject to doubt.” The exemple of the property sarvajña(tva) is also used in NB 3.94–95, 3.125 and 3.130, with the same kind of implications.


In a suggestive note of his recent study of PVSV (Steinkellner 20132: 45–48, n. 49), E. Steinkellner proposes to link the development of the theory of anupalabdhi in Dharmakīrti’s *Hetuprakaraṇa to the debate on omniscience as known to us in particular through the works of Kumārila. See Steinkellner 2013: “(…) ein wichtiger Anstoß für die kräftige Entwicklung der Lehre von der negativen Erkenntnis durch Kumārila und stärker noch durch Dharmakīrti [ist] in der bei Kumārila sichtbar werdenden Polemik der Mīmāṃsā gegen die Ansicht von der Existenz eines Allwissenden (sarvajña) zu identifizieren” (p. 46). The ViV provides, in a way, a powerful confirmation of this insight, as do the passage of the Nyāyabindu translated above (n. 40) and the statement from Kamalaśīla’s Nyāyabindupūrvapakṣasaṃkṣipta quoted by Steinkellner (p. 47). I also fully agree that most of Kumārila’s arguments against omniscience do not hold against Dharmakīrti’s elucidation of anupalabdhi (see below § 2.3–4). But one may also wonder how far the question of omniscience was present to Dharmakīrti’s mind from the very beginning and, if it was, why he never mentions it in the Pramāṇavārttika. It is not impossible that Maṇḍana’s text reflects and elaborates on developments which are characteristic of Dharmakīrti’s later work.


All kārikā-numbers refer to the codanā-section (ad MīSū 1.1.2).


A useful synopsis of the section is found in Kataoka 20111: xliv and Kataoka 20112: 182–184. For a systematic comparison of the ŚlV and BṬ, see Kataoka 2003a: 38 and Kataoka 20112: 38.


ŚlV (codanā°) 117–118: sarvajño dṛśyate tāvan nedānīm asmadādibhiḥ / nirākaraṇavac chakyā na cāsīd iti kalpanā // na cāgamena sarvajñas, tadīye’nyonyasaṃśrayāt / narāntarapraṇītasya prāmāṇyaṃ gamyate katham //. Translation: Kataoka 20112: 332–336 (slightly modified).


See the quite impressive list of quotations of these verses in Kataoka 20111: 29–30. The fact that an omniscient being “is not seen now” (na [] idānīṃ dṛśyate) is also, as we remember, the basic argument of Bhavya’s pūrvapakṣa in the Mīmāṃsā-section of his Madhyamakahṛdayakārikā (15ab) (on Bhavya’s relation with Kumārila, see above n. 10).


One could read an echo of Kumārila’s argument in ViV 19, where Maṇḍana underlines that an omniscient being can neither be perceived, as he is “not within the reach of the senses” (indriyāṇāṃ na gocaraḥ [ViVK 19b]), nor inferred, for the very same reason (ata eva [ViVK 19c]). Yet we should also pay attention to the fact that Maṇḍana thereby only wants to prove our ignorance (ajñāna [S 686.2]) of an omniscient being, while its non-existence or intrinsic impossibility (anupapatti) is considered sufficiently proved by the preceding section (ViV 15–18). Kumārila, on the other hand, evaluates which supposition (kalpanā)–that of the Veda’s authorlessness or that of an omniscient being–is more dispendious (see ŚlV [codana°] 116). His reasoning is therefore against the probability for the Omniscient’s existence, while Maṇḍana’s claim is only, on an epistemic level, about our knowledge of that person.


On these two types of anupalabdhi, see Steinkellner (1967: 157–158), who proposes to distinguish between non-apprehension in general (“Nicht-Beobachtung im Allgemeinen”) and non-apprehension in particular (“Nicht-Beobachtung im Besonderen”) in the Pramāṇavārttika. Additional remarks on this distinction are found in Steinkellner 20132: 44.


See PVSV 101.19–20: na () asattāsādhanī (read °sādhanī instead of °sādhānī ); PVSV 103.10: ata eva saṃśayo’stu.


See PVSV 101.11: anupalabhyamāno na san nāsan, satām api svabhāvādiviprakarṣāt kadā cid anupa-lambhāt; PV 1.202a: sadasanniścayaphalā [anupalabdhiḥ]; PVSV 103.1–2: yasya kasya cit svabhāvo nopalabhyate deśādiviprakarṣāt, na sa tadanupalambhamātreṇāsan nāma.


See PVSV 101.17–18: asattāyām () pramāṇam.


Although Dharmakīrti does not mention these two cases in the PVSV, the distinction is found in some of his later works, as pointed out by Stern (1988: 1023). See for instance NB 2.27 ([] deśakālasva-bhāvaviprakṛṣṭeṣv artheṣv ātmapratyakṣanivṛtter abhāvaniścayābhāvāt) and Vādanyāya 10.12–14 (na hy anumānādinivṛttir abhāvaṃ gamayati, vyabhicārāt, na sarvapratyakṣanivṛttiḥ, asiddheḥ, nātmapra-tyakṣāviśeṣanivṛttir api viprakṛṣṭeṣu).


PVSV 102.10: na ca te pramāṇatrayanivṛttāv api na santīti śakyante vyavasātum. The three means of knowledge (pramāṇatraya) are enumerated in PV 1.199. This point, which is developed in the whole SV on this verse, is reiterated at the very end of the first chapter of the PV. See PV 1.339cd and PVSV 176.11–12: tenāsanniścayaphalānupalabdhir na sidhyati // tasmān na pramāṇatrayanivṛttāv api viprakṛṣṭeṣv abhāvaniścayaḥ; “[PV:] Therefore, it is not established that non-apprehension results in the ascertainment [of something] as non-existent. [SV:] Therefore, the non-existence of [things] beyond the reach [of ordinary cognition] cannot be ascertained even if all three means of valid cognition [should] fail to operate” (read asanniścaya° instead of asaṃniścaya°; translation: Eltschinger/Krasser/Taber 2012: 76–77; I modify “non-perception” into “non-apprehension”).


The same kind of refutation also forms the basis of immediately later Buddhist defences of omniscience, like that of Śāntarakṣita and Kamalaśīla. See McClintock 2010: 165–187. Śāntarakṣita’s use of Dharmakīrti’s analysis of anupalabdhimātra as producive of a mere doubt (saṃśaya) is also clear. See TS 3300–3301, translated in McClintock 2010: 186. I find it quite remarkable that Kumārila who, according to the now (almost) consensual hypothesis of Frauwallner (1962), wrote the BṬ partly as a response to Dharmakīrti’s *Hetuprakaraṇa, does not modify at all his strategy in what could be his last great work. On the contrary, far from renouncing his proof of the Omniscient’s non-existence by mere lack of evidence, Kumārila brings it to its perfection in the BṬ by adding to the examination of perception (pratyakṣa), inference (anumāna) and speech (śabda) carried out in the ŚlV a consideration of comparison (upamāna) and presumption (arthāpatti) as well. See BṬ (= TS 3214–3228).


As we have seen (§ 2.3), this constitutes the third part of Kumārila’s argument in that section.


ŚlV (codanā°) 137: rāgādirahite cāsmin nirvyāpāre vyavasthite / deśanānyapraṇītaiva syād ṛte pratya-vekṣaṇāt //; “And when he is established as having no action because he lacks desire and so on, [his] teaching could only have been composed by others without having [directly] observed [anything].” Translation: Kataoka 20112: 366–369.


See Kataoka 2003a: 55–63, Kataoka 20112: 366–369 (nn. 425–426). Doubts about this identification have been expressed by J. Taber (see Eltschinger/Taber/Krasser 2012: 119–120, n. 3) since Kumārila, unlike Dharmakīrti, does not mention the Buddha’s speech (vacana) but only his operation (vyāpāra). Though I agree that only a quote could settle the matter, I find the objection hardly convincing for, as already pointed out by Steinkellner (20132: 85), one fails to see to which ‘operation’ Kumārila could possibly allude if not to the Buddha’s teaching (deśanā), which would otherwise have to be promulgated by others (anyapraṇīta). Further arguments in favour of Kataoka’s identification on the basis of the structure of PVSV have been voiced by Steinkellner (20132: 84–87), who carefully concludes that Dharmakīrti must be attacking, if not Kumārila himself, at least some Mīmāṃsaka position concordant with that of Kumārila.


PVSV 9.7–8 (ad PV 1.12): yathā rakto bravīti, tathā virakto’pīti vacanamātrād apratipattiḥ. Translation: Franco 2012: 231.


Although Vācaspati tells us nothing of the value of °ādi, the expression vacanādi may correspond to the compound spandavacanādi (“movement, speech, etc.”) found at the beginning of the SV on PV 1.12 (PVSV 9.3), or else to the three properties of humanity enumerated in PV(SV) 1.311 (senses, mind and speech). See above n. 39. The parallel passage in NB 3.71 only mentions vaktṛtva (see our translation of this passage above, n. 40). Another possibility is that Maṇḍana alludes here to the contradiction between contemplation (dhyāna) and teaching (upadeśana) underlined by Kumārila in the BṬ (= TS 3237–3239). It is unclear, in that case, how he intended to solve this apparent contradiction.


BS 1.1: na hīcchāmātraṃ rāgaḥ. avidyākṣiptam abhūtaguṇābhiniveśaṃ rāgam ācakṣate. tattvadarśanavaimalyāt tu cetasaḥ prasādo’bhirucir abhīcchā na rāgapakṣe vyavasthāpyate, yathā saṃsārāsā-ratātattvadarśananiṣpanno nodvegas tato dveṣapakṣe (3.17–20). I slightly modify the translation of this passage by Taber (2011: 443), who rightly points out the importance of this parallel for a correct interpretation of Dharmakīrti’s concept of rāga in PV 1.12 and the corresponding SV.


Dharmakīrti’s definition of rāga in PVSV 9.5–6 is exactly similar if we except the mention, instead of “nescience” (avidyā), of a list of objects of defilement typical of Buddhist thought, corresponding to the four “aspects” of the first āryasatya (anityatā, duḥkha, anātmatā, śūnyatā –see Eltschinger 2014b): nityasukhātmātmīyadarśanākṣiptaṃ sāsravadharmaviṣayaṃ cetaso’bhiṣvaṅgaṃ rāgam āhuḥ; “They call ‘passion’ the attachment of the mind, which has the defiled elements of existence as an object, which is caused by seeing [erroneously, what is impermanent, suffering, not the self, and does not belong to the self as] permanent, pleasure, the self and what belongs to the self.” Translation: Franco 2012: 231 (I modify “desire” into “passion,” to suit the convention followed here).


PV 1.201cd and PVSV 103.16–18: kārye tu kārakājñānam abhāvasyaiva sādhakam // svabhāvābhāve sādhye tadanupalambha evāpramāṇam ucyate. kārakānupalambhas tu pramāṇam eva. na hy asti saṃbhavo yad asati kāraṇe kāryaṃ syāt.


See PVSV 103.1–2: yasya kasya cit svabhāvo nopalabhyate deśādiviprakarṣān na sa tadanupalambhamātreṇāsan nāma, yathoktaṃ prāk. On the identification of the passage alluded to by Dharmakīrti (PVSV 101.11 or 102.11–12?), see Steinkellner 20132: 273 (n. 543).


The expression kārakānupalambha used in the PVSV is of course equivalent to the compound kāraṇā-nupalabdhi found in Dharmakīrti’s later works (see above n. 37).


This point is well made by Yaita (1985: 202, n. 65).


On puruṣātiśaya / sātiśayo naraḥ, see BṬ (= TS 3161/3159) and PV(SV) 1.311. The expression darśanātiśaya is used, for instance, in Jayanta Bhaṭṭa’s Nyāyamañjarī (NM1: 268.3) to define the cognition of yogins. A similar use of atiśaya applied to cognition is found in the ŚlV (codanā° –114) and BṬ (= TS 3386).


Interestingly, this seems to be precisely the point where the problem of omniscience is taken up by Śaṅkaranandana in his Sarvajñasiddhi, possibly written in the 9th century. See SSiŚ 2: jñātakāraṇasad-bhāvā saṃbhaviny anyakāryavat / sarvārthaviṣayā saṃvit, sa hi saṃbhava ucyate; “A consciousness encompassing all objects, like [any] other effect, is possible (saṃbhavin) [only] if the actual exis-tence of its cause (kāraṇasadbhāva) is known; for this is what [we] call ‘possibility’ (saṃbhava).” The corresponding prose portion, a preliminary edition of which is established by Eltschinger (2008: 140–141), is still too obscure (to me, at least) to provide any reliable information about the intellectual background of this stanza. The proximity between Maṇḍana and Śaṅkaranandana’s way of speaking of omniscience (sarvārtheṣu saṃvid [Maṇḍana]/sarvārthaviṣayā saṃvid [Śaṅkaranandana]) need not be significant, but is nevertheless striking. The same kind of consideration is also found in Ratnakīrti’s Sarvajñasiddhi: nanu kāraṇānupalambhād eva sarvajñatāpratiṣedhaḥ sidhyati; “[Objection:] but, the negation of omniscience is established by the non-apprehension of its cause (kāraṇānupalambha)!” (SSiR 7.19–20). See also Moriyama 2014: 64 (n. 29).


As noted by Moriyama (2014: 60–61), the quotation of scriptural passages mentioning the Buddha’s “Divine Eye” (divyacakṣus) by commentators on ŚlV (codanā°) 112–115 and by Kamalaśīla while commenting on a verse of the BṬ (ad TS 3159–3160) in TSP 999.12–13 is certainly not done by chance. It is indeed likely that Kumārila had this or a similar notion in mind while discussing this topic.


ŚlV (codanā°) 112–114: ekena tu pramāṇena sarvajño yena kalpyate / nūnaṃ sa cakṣuṣā sarvān rasādīn pratipadyate // (…) yatrāpy atiśayo dṛṣṭaḥ sa svārthānatilaṅghanāt / dūrasūkṣmādidṛṣṭau syān na rūpe śrotravṛttitā //; “However, if [you] postulate that he knows everything through a single means of valid cognition, he surely grasps all tastes, etc. with the eye! Even when superiority of a particular [pramāṇa] is seen, in so far as [a sense] does not go beyond [its] own object, that [superiority] can [happen] in perceiving things that are remote, subtle, etc., [but] it is not the case that the ear should grasp color.” Translation: Kataoka 20112: 324–328 (modified). Cf. BṬ (TS 3157–3158), translated in Kataoka 20112: 328–329 (n. 368).


Although this last part of the argument is not voiced by Kumārila, it seems nevertheless a natural consequence of his statements about the senses and their limited domain. It is found explicitly in Kamalaśīla’s commentary on TS 3157. See TSP 998.7–9: na caitac chakyate vaktum mā bhūd ekena jñānena yugapad aśeṣārthasya grahaṇam, anekena bhaviṣyatīti, yato yugapad anekavijñānāsaṃbhavāt; “And you cannot say [the following:] ‘Maybe it is impossible to grasp all objects at the same time (yugapad) in a single cognition (ekena jñānena), but this can happen in several [cognitions] (anekena),’ for it is impossible that several cognitions [should take place] at the same time.”


To the already long list of quotes of ŚlV (codanā°) 112–114 enumerated by Kataoka (20111: 27–29), I can only add the (somewhat unexpected) quotation of v. 112ab in Helārāja’s Prakīrṇaprakāśa (vol. 1, p. 54.17–ad Vākyapadīya 3.1.46; I thank Vincenzo Vergiani for drawing my attention to this passage). Helārāja’s response to Kumārila is quite unique in that the 10th-century Kashmiri grammarian directly contests the Mīmāṃsaka’s claim that the domains of the senses are mutually impenetrable, and does so on the basis of some hitherto unidentified Scripture (āgama): tac ca teṣāṃ śiṣṭānāṃ jñānaṃ sarvendriyam, pratiniyamānapekṣaṇāt. sarvajñā hīndriyāntareṇāpīndriyāntaravyāpāraṃ kurvanti, tathā cāgamaḥ nedānīm indriyair eva paśyanti, ghrāṇataḥ śabdaṃ śṛṇoti, pṛṣṭho rūpāṇi paśyati, apy aṅgulyagreṇa sarvendriyārthān upalabhyate; “And this knowledge of the Learned Ones (śiṣṭa) is [produced] by all the senses (sarvendriya), for omniscient [beings] accomplish the operation of a sense even by means of another, as it is said in the [following] Scripture: ‘Now they do not see only by the senses. [In that state,] one hears sounds by [the organ of] smell, sees forms [even] in [one’s] back. More than that! One grasps all sensory objects even with the tip of a finger!’” (Prakīrṇaprakāśa 54.1–5). The boldness of Helārāja’s statement appears by comparing it, for instance, to Jayanta Bhaṭṭa’s much milder response to Kumārila: rasādigrāhīṇy api yoginām indriyāṇi cakṣurvad atiśayavanty eveti na rasādiṣu cakṣurvyāpāraḥ parikalpyate; “The senses by which yogins grasp smells, etc. are also eminent, just like [their] eyes, so there is no need to assume an operation of the eye towards smells, etc.” (NM1 270.1–2). It is impossible to decide if Maṇḍana positively rejected Kumārila’s claim (and in that case, on which basis), or simply considered it irrelevant to the present debate. I find it unlikely, in any case, that he ignored it.


ViV 15: na pratyakṣaṃ cakṣurādijanma tāvat sarvārtheṣu, teṣāṃ viṣayaniyamāt. kiṃ cid eva hi vartamā-naṃ saṃbaddhaṃ ca tadviṣayaḥ, na ca sarve’rthās tathāa. yady api cānekavidhaprameyasaṃvedanād rūpato niyamo durnirūpaḥ, paṭumandatādibhedataś ca, nānādeśaparimāṇārthaparicchedād deśa-taḥb saṃbandhaniyamaḥ, parimāṇato niyamaś ca, kālatas tu nirūpyate, prāpyakāritve’vartamānasya prāptyabhāvāt, aprāpyakāritve’py arthasāmarthyasavyapekṣatvāt, jñānotpattāv ajātanivṛttayor anupā-khyeyatvād asāmarthyāt. tadāc cātiśayānād api cakṣurādayaḥ svaviṣayam eva deśataḥ parimāṇataḥ saṃkhyāto vādhikaṃ bodhayeyuḥ, nāviṣayam ajātam ativṛttam. tad uktam vidyamānopalambhanatvād iti (S 461.2–468.1 [≈ M 116.2–119.1/G 82.6–84.3]). a tathā S: Ø M G; b deśataḥ S: na deśataḥ M G; c tadā S: tathā M G; d atiśayānā(ḥ) S: atiśayānām M G. The text is trunked in M (118.4).


On this well-known divergence, see Chatterjee 1978: 138–141 and Bhatt 1989: 174–177. It opposes thinkers (including Mīmāṃsakas, Naiyāyikas and Sāṃkhya-philosophers) who think that the senses must “reach” (pra-√āp) their object in order to produce sensation and others (notably Buddhists) who estimate that sensation can happen even while senses and object are at a distance.


Maṇḍana’s conception of past and future as “inexpressible [in terms of being and non-being]” (anupākhyeya) might be borrowed from Bhartṛhari’s Vākyapadīya. See David (forthcoming).


Maṇḍana’s equation between being and being present forms the topic of the third study in this series. See David (forthcoming).


ŚlV (codanā°) 115ac: bhaviṣyati na dṛṣṭaṃ ca pratyakṣasya manāg api / sāmarthyam; “It is never seen that perception has even a bit (manāg api) of capacity with regard to a thing in the future.” Translation: Kataoka 20112: 329. Unsurprisingly, Kumārila reads in MīSū 1.1.4, especially in the reason vidyamānopalambhanatvāt, an implicit attack against the possibility of yogic perception. See ŚlV (pratyakṣa°) 26–27ab (translated in Taber 2005: 54).


The earliest echo of this shift in the Mīmāṃsakas’ attitude towards omniscience in a Buddhist text might be found in Kamalaśīla’s Pañjikā on TS 3156–3157 (= BṬ), where the learned Buddhist scholar takes into consideration two hypotheses about the nature of the “complete knowledge of all things” (sakalavastuparijñāna): sensory cognition (indriyajñāna) and mental cognition (manovijñāna). See TSP 997.20–998.21. It is possible that Kamalaśīla’s comments on these stanzas should be read as an effort to integrate Maṇḍana’s arguments into Śāntarakṣita’s mainly Kumārilan framework. His examination of mental cognition (TSP 998.18–21), in particular, with his insistence on the mind’s “dependence” (pāratantrya) on the senses, clearly reminds one of Maṇḍana (see below § 3.4). The case of Prajñākaragupta will be dealt with further on (§ 4).


On these four types, see for instance NB 1.7–11.


See PV 3.281ab: prāg uktaṃ yogināṃ jñānaṃ teṣāṃ tad bhāvanāmayam (Translation: Eltschinger 2009: 192). On the ‘causal’ interpretation of bhāvanāmaya, see Eltschinger 2007: 85–86, n. 58.


This is, at least, the assumption made by Umbeka Bhaṭṭa (8th c.?), the oldest commentator on the ŚlV, who begins his commentary on ŚlV (codanā°) 115 with the following objection: nanu heyopādeya-grāhakasya vijñānasyāsty ekaṃ kāraṇaṃ bhāvanā, kim ucyate kāraṇānupalabdhyā kāryābhāva iti? bhāvanājanyapratyakṣaṃ dharmādharmagrāhakatvena tair iṣṭam!; “[Objection:] but, there is [indeed] a cause for the cognition that grasps what is to be abandoned and what is to be appropriated, [namely] meditation (bhāvanā)! Why do [you] say that ‘the effect does not exist since one does not grasp [its] cause’? They [= Buddhists] maintain that perception born from meditation (bhāvanājanyaṃ pratyakṣam) is what grasps merit and demerit!” (Tātparyaṭīkā 74.7–8; translation: Moriyama 2014: 64 [modified]). The phrase kāraṇānupalabdhyā kāryābhāvaḥ recalls Maṇḍana’s formulation in ViV 15 (kāraṇanivṛttyā kāryanivṛtti[ḥ] [S 460.2–461.1]), and it is possible that Umbeka, who is also the author of a commentary on Maṇḍana’s Bhāvanāviveka, makes here an approximative quote of the ViV.


The possibility that dreams (svapna) manifest the mind’s capacity to grasp external objects independently of the senses–a hypothesis Maṇḍana eventually rules out–is the topic of a separate investigation in ViV 16 (S 583.2–590.3). This enquiry, carried out essentially with non-Buddhist arguments and only loosely related to the question of omniscience, need not concern us here.


As is well-known, Dharmakīrti’s second characterization of “valid knowledge” (pramāṇa) in PV 2.5a defines it as the “manifestation of an unknown object” (ajñātārthaprakāśa). On this definition, see Katsura 1984, Krasser 2001: 185–190 and Kataoka 2003b. Further references in Krasser 2001: 184–185 (n. 45).


See PV 3.239–244: pūrvānubhūtagrahaṇe mānasasyāpramāṇatā / adṛṣṭagrahaṇe’ndhāder api syād arthadarśanam // () tasmād indriyavijñānānantarapratyayodbhavaḥ / mano’nyam eva gṛhṇāti viṣa-yaṃ nāndhadṛk tataḥ // svārthānvayārthāpekṣaiva hetur indriyajā matiḥ / tato’nyagrahaṇe’py asya niyatagrāhyatā matā //; “If mental [perception] grasps [an object] that has been experienced before (pūrvānubhūta), then it is not a means of valid knowledge (apramāṇatā); if it grasps something [en-tirely] unseen (adṛṣṭa), then a vision of the object would occur also to a blind man, etc. (…) Therefore, the mind [= mental cognition] (manas), born from the immediate cause (anantarapratyaya) that is a sensory cognition (indriyavijñāna), grasps an entirely different object (anyam eva viṣayam) [with respect to that sensation], so that [the undesired consequence that is] the vision [of the object] by a blind man does not [occur]. [Still,] the sensory cognition (indriyajā matiḥ) that is [its] cause (hetu) is entirely dependent (°apekṣaiva) on an object (artha) related to its own object [as its immediate cause] (svārthānvaya°); so, even though it grasps something different, [we] consider that it grasps [only] a delimited object (niyatagrāhyatā).” See also Vetter 1964: 40 and PVin 1.19 (19.1–7). A thorough account of Dharmakīrti’s theory of mental cognition is given by Vācaspati in the NyK (471.2–473.8); the passage has been translated into English by Stcherbatsky (19302: 318–320).


ViV 15: mānasam api pratyakṣam, bahir manaso’svatantratvāt, svātantrye’ndhabādhirādyabhāvaprasaṅgāt. tathā hi: pratyakṣe rūpādivedane tac cakṣurādiparatantram tanniyamād eva niyatam, anyathāndhādyabhāvaprasaṅgād ity uktam. tajjasamanantarapratyayasahakāryapekṣaṇād yadi nā-ndhādyabhāvaḥ sarvathā na pāratantryaṃa nivartate, tadviṣayajaviṣayatvātb, anyathendriyāntara-jasahakāriṇocpi pravṛtteḥ sa evāndhādyabhāvaḥ (S 468.1–474.2 [≈ M 119.1/G 84.3–85.5]). a na pāratantryam S: pāratantryaṃ na M G; b tadviṣayajaviṣayatvāt S: tadviṣayatvāt M G; c indriyāntaraja° M G: indriyāntara° S. The order of the sentences differs widely between S and both printed editions (M/G); I do not reproduce these variants here.


See NyK: mānasam api pratyakṣaṃ na sarvaṃ bodhayati (468.13).


Cf. ViVK 15d: paratantraṃ bahir manaḥ; “Regarding external [objects], the mind is dependent (paratantra) [on the senses]” (S 458.3 [= M 114.3/G 81.3]).


As we have seen before, the essential limitation of the senses, in Maṇḍana’s view, is their incapacity to grasp objects in the past or future.


Cf. PV 3.243ab (translated above, n. 82), PVin 1.19 (19.5–7) and NB 1.9. The latter’s definition reads as follows: svaviṣayānantaraviṣayasahakāriṇendriyajñānena samanantarapratyayena janitaṃ tan manovijñānam: “[M]ental perception is the product of a sense perception which forms its immediately preceding homogeneous cause, and which cooperates with the immediately succeeding facsimile [i.e. the second moment] of its proper object.” Translation: Kajiyama 1998: 45.


That is: the objective moment (kṣaṇa) which forms the content (viṣaya) of mental awareness of a patch of blue (K2), for instance, is not identical with the preceding objective moment (K1), the content of sensation. Yet both moments are not unrelated, since K1 is also the “homogeneous and immediate cause” (samanantarapratyaya) of K2. In order for mental cognition to take place, then, we need two things: a (sensory) cognition which is its samanantarapratyaya –or, in Vācaspati’s more oecumenic terms, its “material cause” (upādāna –NyK 472.2)–and an objective correlate (sahakārin) which is the immediate product of the objective moment (K1) grasped by that sensation. Thus, although sensation and mental awareness have different contents, they are nevertheless indissolubly intertwined.


I exceptionally disagree with Stern’s choice to read °indriyāntarasahakāriṇo, and prefer the reading °indriyāntarajasahakāriṇo transmitted in his Ms. B, in his own version of the NyK (474.5) and also chosen by both published editions of the ViV (M 120.5/G 85.4). Vācaspati’s interpretation of the compound indriyāntaraja° as referring to the cognition (indriyāntarajavijñāna°) which is the material cause (or samanantarapratyaya) of mental awareness clearly supports that interpretation.


ViV 15: na kva cida anāntare manaḥ sākṣāt svatantraṃ pravartate (S 541.3–4 [≈ M 140.6–141.1/G 100.3]). a na kva cid S: na kva cid api M G. Cf. ŚlV (pratyakṣa°) 160cd: pravṛttiḥ sukhaduḥkhādau kevalasyaiva dṛśyate; “(…) a functioning of the mind by itself is observed in respect to pleasure, pain, etc.” (text and translation: Taber 2005: 158 and 114).


ŚlV (pratyakṣa°) 26ac: atītānāgate’py arthe sūkṣme vyavahite’pi ca / pratyakṣam yoginām iṣṭaṃ kaiścin muktātmanām api // (text: Taber 2005: 152; translation: Taber 2005: 54 [adapted]).


On abhyāsa and its interpretation in Buddhist texts, see Eltschinger 2009: 184 (n. 57).


On the “vision of the [four] Nobles’ Truths” (āryasatyadarśana) as the specific goal of yogic perception for Dharmakīrti, see PVin 1.28 (27.11) and Eltschinger 2014b: 250–251. As convincingly shown by Eltschinger (2009: 199–200), that vision corresponds, in Dharmakīrti’s view, to the cognition of the Buddhist practitioner after the “revolution of the basis” (āśrayaparivṛtti) has taken place.


See PV 3.282 (= PVin 1.29). Translations: Eltschinger 2009: 193, Franco 2011: 83.


See PVin 1.4ab (pratyakṣaṃ kalpanāpoḍham abhrāntam), NB 1.4 (tatra pratyakṣaṃ kalpanāpoḍham abhrāntam). See also Steinkellner 1978: 126, Eltschinger 2009: 196–197 and Franco 2011: 86–88.


Maṇḍana alludes here to his own tripartition of Brahman-knowledge in the beginning of the Niyogakāṇḍa: “knowledge born from speech” (śabdāt pratipattiḥ), knowledge “consisting of its continuous fixation” (tatsaṃtānavatī [pratipattiḥ]) and knowledge “consisting of a direct apprehension” (sākṣātkārarūpā [pratipattiḥ]). See BS 74.10–13. Meditation (bhāvanā) as understood by Dharmakīrti corresponds, of course, to the second of these three stages.


BS 2.101–105ab (introduction): dvitīyapratipattiviṣayo hi na tattvāvabodhaviṣayaḥ, kiṃ tu tadabhyāsa- viṣayaḥ (115.2–3–I do not translate hi). Śaṅkhapāṇi’s commentary (Brahmasiddhivyākhyā 239.10–11) makes the precision that the difference between both types of knowledge (pratipatti) is equivalent to that between valid knowledge (pramāṇa) and memory (smṛti).


ViV 15: bhāvanāmayam apia śrutānumitaviṣayam akasmād bhāvanāyogād āgamānumānaparatantraṃ pūrvajñānabhūtārthatvaviparyayānuvidhānāt sāpekṣatvād apramāṇam (S 542.1–555.2 [≈ M 147.1–3/G 104.1–3]). a bhāvanāmayam api S: bhāvanāmayam api vijñānam M G.


See PVin 1.28: yoginām api śrutamayena jñānenārthān gṛhītvā yukticintāmayena vyavasthāpya bhāva-yatāṃ tanniṣpattau yat spaṣṭāvabhāsi bhayādāv iva tad avikalpakam avitathaviṣayaṃ pramāṇaṃ pratyakṣam; “Having first grasped objects through a cognition born from listening [to the treatises] and [then] ascertained [them] through a [cognition] born of reflecting [upon them] by means of rational enquiry, yogins cultivate [those objects]. The [cognition] which, at the completion of this [cultivation], appears as vividly as in such cases as fear, etc. and [at the same time] is non-conceptual [and] has a true object, is also a means of valid knowledge, [namely] perception” (27.9–11–translation: Eltschinger 2009: 198 [modified]). The hypothesis of a direct response to Dharmakīrti is indirectly supported by the paraphrase of PVin 1.28 in the corresponding portion of the NyK. Interestingly, Vācaspati does not speak in general of the cognition of yogins but specifically of that of the Buddha (tathāgata), and also relates Dharmakīrti’s remarks to the question of omniscience: śrutamayena vijñānena samasta-vastuviṣayaṃ nairātmyādi gṛhītvā yuktimayena ca bhūtatām asya vyavasthāpyāsakṛccetoniveśanarū-pabhāvanāprakarṣaparyantajanma pratyakṣaṃ vijñānam anavayavenānātmādirūpaviśvālambanaṃ karatalāravindaviṣayam ivātiviśadaṃ bhāvayiṣyati tathāgataḥ; “Having first grasped the absence of a Self, etc., which concerns all beings, by means of a cognition born from listening [to the treatises] and [then] established their reality by means of [a cognition] born from reasoning, the Tathāgata was able to effectuate (bhāvayiṣyati) a perceptual cognition born of [His] intense meditation, consisting of a repeated presentation to [His] mind [of the objects he reflected upon]. [That vision] had for its objective correlate everything without exception possessing the property of being selfless, etc. and was as entirely clear as [the vision] of a lotus on the palm of one’s hand” (S 544.6–545.4).


Recall that novelty or “manifestation of an unknown object” (ajñātārthaprakāśa) is one of the definitions given by Dharmakīrti himself of “valid knowledge” in PV 2.5a. See above n. 81.


Bṛhadāraṇyakopaniṣad 2.4.5: ātmā vā are draṣṭavyaḥ śrotavyo mantavyo nididhyāsitavyaḥ; “Truly, it is the Self that must be seen, heard, thought, meditated upon.” As is well-known, this passage is taken by the later Vedāntic tradition to enunciate the three stages of the knowledge of Brahman, starting with its “audition” (śravaṇa) in the Scripture, developing through “reflection” (manana) and “contemplation” (nididhyāsana), and eventually leading to “direct perception” (sākṣātkāra).


See Steinkellner 1978: 127.


Apart from the PVA, possible echoes of Maṇḍana’s arguments have been identified so far in the works of Kamalaśīla (see above n. 76), Śaṅkaranandana (see n. 66) and Ratnakīrti (see nn. 37 and 66).


I am quoting here the recent edition of the passage by Moriyama (2014: 168–179) (= PVA), which corresponds to PVAS 50.17–52.25. The numbering of kārikās is identical in both editions.


According to R. Jackson (1991: 235–236), Prajñākaragupta’s “conflating omniscience and authorita-tiveness” constitutes a decisive innovation of the Vārttikālaṃkāra with respect to earlier commentaries on Dharmakīrti’s work. See also Franco (2011: 90, n. 44) and Moriyama (2014: 19–26), who reach the same conclusion, the latter by an in-depth study of PVA ad PV 2.1–7. Interestingly, omniscience will be regarded by some later Tibetan commentators as the fundamental topic, not only of those kārikās, but of the whole second chapter. See Jackson (1991: 232 and 236), who mentions in particular the case of rGyal tshab (15th c.). The latter’s indebtedness to Prajñākaragupta (which of course need not be direct) is evident from the passage of his commentary on PV 2 translated in Jackson 1991: 241, which is little more than a paraphrase of PVA ad PV 2.29.


Apart from a small portion of the text (PVA 168.9–10 and PVA 2.367–369), a rather bold adaptation of an argument originally found in PV 1.335, all arguments of Prajñākaragupta’s pūrvapakṣin can be traced back to earlier Mīmāṃsā works (see table below, § 4.6). Yamāri’s tentative identification of Prajñākaragupta’s opponent as a materialist (tshu rol mdzes pa pa, Skt. *cārvāka?), on which Moriyama (2014: 244, n. 5) already expressed serious doubts, can therefore be entirely discarded.


As he convincingly shows (pp. 58–59), the objection given in PVA 2.359 that an omniscient would also experience the taste of impure things (aśuci), which is not found in the ŚlV, is certainly borrowed from the BṬ (= TS 3144). Even if some parallels he draws between the PVA and the ŚlV might be contested (especially in the case of PVA 2.358 and 2.359cd, as we shall see, but also in other cases like PVA 2.362ab and 2.364, in which Prajñākaragupta might equally refer to the BṬ), the identification of ŚlV (codanā°) 137 as the source of PVA 2.365, already proposed by Jayanta (see Moriyama 2014: 248, n. 25), is in turn quite convincing. The parallel passage of the BṬ (= TS 3238–40) might indeed be alluded to by Prajñākaragupta through his use of the expression vikalparahita (“devoid of conceptual knowledge”), but the kārikā shows no evident formal similarity with that part of Kumārila’s late work, while it is very close in wording to the verse of the ŚlV.


The possibility of Maṇḍana’s influence on Prajñākaragupta in this pūrvapakṣa is briefly considered by Moriyama (2014: 63–65), who does not however engage in a systematic comparison of both texts.


The whole text of PV 2.29 runs as follows: prāmānyaṃ ca parokṣārthajñānaṃ tatsādhanasya ca (em.; tat sādhanasya Ed) / abhāvān nāsty anuṣṭhānam iti ke cit pracakṣate //; “And the reliability [of any religious authority] consists in [His/its] knowledge of objects beyond the senses, but because there is no [possible] means to complete it, there is no [successful] practice in conformity with [its teaching]. Thus claim certain [Mīmāṃsakas].” Translation: Moriyama 2014: 244.


See PVA ad PV 2.29: yas tāvad asarvajña eva sarvajño bhavati, tasya parokṣārthaparijñāne ko hetuḥ? na khalv īdṛśaṃ kim api kāraṇam upalakṣitam, yato’nuṣṭhānāt sarvavedanaṃ bhavati () nāpi tanniścaye hetur asti; “First of all, if someone who is not omniscient becomes omniscient, what is the cause (hetu) of his knowledge of objects that are beyond the senses? For sure, [you] cannot point out any such cause (kāraṇa) from which, through practice, the knowledge of all [things] would arise (…); nor is there any cause of [our] certitude (niścaye hetuḥ) that such a [being exists]” (168.5–9). Translation: Moriyama 2014: 244 (modified).


See Inami 1996: 87, n. 5 and Moriyama 2014: 56.


See ViV 15: () na, tasyānupapatter ajñānāc ca; “No, for [an omniscient being] cannot be accounted for, and because we would have no way to know [Him]” (S 445.1–458.1 [= M 114.1/G 81.1]); ViV 19: evaṃ tāvad anupapattiḥ, ajñānama api; “Thus [it has been shown], first of all, that [an omniscient being] cannot be accounted for; now [we will see that] there is also no knowledge [of such a being]” (S 686.1–2 [≈ M 204.5/G 145.2]). a ajñānam S: jñānam M G.


Tibetan text quoted in Moriyama (2014: 59, n. 12). Since the original Sanskrit of Yamāri’s Supariśuddhī (on which see Steinkellner/Much 1995: xx) is still unpublished, it is not possible to check whether oddities of this classification are due to Yamāri or to his Tibetan translator. In any case, Maṇḍana’s name is not included in the list of authorities identified by M. Ono (2000: xxiv) in Yamāri’s commentary, so it is quite possible that Prajñākaragupta’s commentator did not know the ViV. My very limited knowledge of Tibetan did not allow me to consult Jayanta’s slightly earlier commentary.


This seems to be, in substance, the option chosen by Moriyama (2014: 59–62), who solves the difficulty by reading Prajñākaragupta’s argument as a reaction to Dharmakīrti’s newly elaborated theory of yogic perception. I am not quite convinced by this explanation since Dharmakīrti, like Dignāga before him, clearly distinguishes between mental and yogic perception, so there would be no point for Prajñākaragupta to discuss mental perception specifically if that were really the point at stake.


See PVA 168.11–12: () indriyajñānasya saṃnihitaviṣayasya darśanāt; “because one observes that a sensory cognition has its object in [its] proximity.” Translation: Moriyama 2014: 245.


See PVA 2.360cd: svatantraṃ tu manojñānaṃ naiva kena cid iṣyate //; “On the other hand, a mental cognition [that is] independent [of sense faculties] is never observed by anyone.” Translation: Moriyama 2014: 246. Cf. ViVK 15d: paratantraṃ bahir manaḥ // (S 458.3–translated above, n. 85).


PVA 2.361–363: abhyāsāt spaṣṭatā tasya na sarvaviṣayā bhavet / āgamāśritatve tasya bhrāntatāpi prasajyate // 361 // anumānaprasiddhaṃ tu vastu sarvaṃ na labhyate / tato na sarvaviṣayā bhāvanā sarvavit katham // 362 // śāstrādyābhyāsataḥ śāstraprabhṛty evāvagacchatu / sākalyavedanaṃ tasya kuta evāgamiṣyati // 363 // (p. 170).


My translation of the passage differs only punctually from that of Moriyama (2014: 246–247).


Even though this interpretation is clearly very tentative, such a solution would avoid the disturbing redundancy of v. 363ab with respect to vv. 361–362. The hypothesis of an implicit reference to Kumārila’s BṬ (= TS 3163), upheld by Moriyama (2014: 77), would be another way out of this difficulty, but I must say I cannot easily read in Prajñākaragupta’s half-verse Kumārila’s objection that excellence or superiority (atiśaya) in the knowledge of a treatise (śāstra) does not entail excellence in the knowledge of another treatise (śāstrāntara): evaṃ śāstravicāreṣu dṛśyate’tiśayo mahān / na tu śāstrāntarajñānaṃ tanmātreṇaiva labhyate // “Thus we notice [in some people] a great superiority in the knowledge of treatises, but this is not sufficient [to establish their] knowledge of other treatises.”


The pūrvapakṣin’s ground for refusing access to the totality of being to inference and Scripture is in itself far from clear. It is almost certain, as rightly pointed out by Moriyama (2014: 59–60, n. 14), that he discards inference in v. 362ab on the basis of Kumārila’s remark in ŚlV (codanā°) 115cd that inference and similar pramāṇas cannot grasp objects in the future (bhaviṣyant). His argument against Scripture in v. 361cd is in turn quite obscure, and I am not at all convinced by Yamāri’s recourse to the (typically Buddhist) argument of an “absence of connection [of speech] with external objects” (phyi rol gyi don dang’brel med pa –Tibetan text quoted by Moriyama [2014: 246, n. 18]), which I find very unlikely to come from the mouth of a mīmāṃsaka opponent. Though many interpretations of this half-verse are possible, I think it would make more sense for the pūrvapakṣin to underline, while speaking of the “erroneousness” (bhrāntatā) of verbal cognition, its conceptual character. The opponent would then reject Dharmakīrti’s claim that a conceptual (i.e. erroneous) cognition arising from the audition of Buddhist Scriptures could lead by its mere repetition to a non-conceptual (i.e. non-erroneous) cognition.


The Sanskrit text of PVA 2.358 reads as follows: indriyārthāviśeṣe’pi yadi sarvavidudbhavaḥ / sarvajña eva sarvaḥ syād indriyārthāviśeṣataḥ // (p. 170).


Recall that such a relation would allow us, in the hypothesis considered by Maṇḍana, to infer the absence of the object to be known (prameyābhāva)–i.e., an omniscient being–from the absence of a pramāṇa establishing its existence, as when we infer for instance the absence of smoke (= vyāpya) from the absence of fire (= vyāpaka).


ViV 15: () tanmātrapratibandhābhāvāc ca, anyathā sarvasya sarvadarśitvaprasaṅgaḥ, aviśeṣāt (S 459.4–460.1–translated above, § 2.2). Supposing Prajñākaragupta is indeed alluding to that argu-ment, it is possible that he interprets aviśeṣa in ViV 15 as well as the absence of difference between the senses and the object (indriyārthāviśeṣa). This intepretation would differ from Vācaspati’s under-standing of that term as referring to the absence of difference between us and the Omniscient. See NyK 460.7–8: aviśeṣād asmadādīnāṃ bhavadabhimatena saha sarvavidā. My translation of the passage (above, § 2.2) follows this last interpretation.


PVA 2.359ab: aśucyādirasāsvādasaṃgamaś cānivāritaḥ /; “And [for an omniscient being who perceives everything through the sense faculties], the connection with the experience of tasting an impure [thing], etc. is unavoidable.” Translation: Moriyama 2014: 245.


As rightly pointed out by Moriyama (2014: 246).


BS 2.106cd–107: nanu prapañcaśūnyasyādvaitasya brahmarūpasya jñeyābhāvād īśitavyābhāvāc ca vijñānam aiśvaryaṃ cānupapannam, tatra sarvajñaḥ, sarveśvara ity api śrutī samādheya eva.–naitat sāram, yato neśitavyakṛtam īśvaratvam, jñeyakṛtaṃ vā jñātṛtvam, kiṃ tu siddhena jñānarūpeṇa si-ddhayā ceśanaśaktyā jñeyam avāpnoti, īśitavyaṃ ca viniyuṅkte praśāsti ca, prakāśadāhavat. siddhena hi prakāśarūpeṇa prakāśyaṃ prakāśayati vivasvān, na tu prakāśyādhīnaṃ tasya prakāśarūpam, dāhyā-dhīnā vāgner dāhaśaktiḥ. tathā ca taccaitanyenaiva kṛtsnasya prapañcasyāvabhāsanāt tasya bhāsā sarvam idaṃ vibhāti, nānyo’to’sti draṣṭetyādiśruteḥ sarvajñatvam (127.5–13).


Kaṭhopaniṣad 5.15.


Bṛhadāraṇyakopaniṣad 3.7.23.


See also Ānandapūrṇa’s comment: jāḍyapāratantryābhāvopalakṣitaṃ rūpaṃ jñānādi, na tu saviṣayam; “Knowledge and [sovereignty] are natures marked by the absence of insensibility and dependence, they do not imply [the existence of] an object” (Bhāvaśuddhi 446.11).

Like what you read? Consider supporting this website: