Unfortunately, he does not seem to elaborate thereon. Vācaspati elaborates extensively and discusses absent pots on the floor, the sheer floor and all we know after Kumārila. Why does Uddyotakara not elaborate thereon?

Probably because someone else (who?) in the tradition had mentioned this possibility and so readers would have understood what he meant by the mention of the sixth kind of contact.

Moreover, is Uddyotakara’s viśeṣaṇaviśeṣyabhāva the same as what will be later known as saṃyuktaviśeṣaṇatā ‘the fact of being an attribute of something being in contact [with the sense faculties]’?

At the end of an epistemological discussion in a Sanskrit text, it is standard to discuss the sources of knowledge (pramāṇa) you don’t accept. Long-term memory (smṛti) has most likely been already excluded at the beginning, while discussing the definition of pramāṇa, so that it is not mentioned among the specific candidates.

Within Mīmāṃsā, Kumārila excludes (at the end of the discussion on abhāva) sambhava (inclusion) and aitihya (tradition). Within Nyāya, Gautama excludes these two, as well as two sources accepted by Kumārila, i.e. arthāpatti (cogent evidence) and abhāva (absence).

The discussion starts in NS 2.2.1, where an opponent says: “There are not 4 sources of knowledge, because [also] tradition, inclusion, cogent evidence and absence as sources of knowledge”. The discussion then goes on for several pages. NS 2.2.2 says that aitihya is nothing but linguistic communication (śabda) and inclusion is inference (anumāna). This is what Kumārila also says. What is different is that Gautama says that also arthāpatti and abhāva are nothing but inference.

This is remarkable, because the classical position of Nyāya authors (e.g., Jayanta) about absence is that this is known through sense-perception (not inference!). Only after Kumārila’s objections and through Gaṅgeśa etc. they say that in some cases inference is indeed needed (e.g., when you infer now that a certain person was not in a place you visited earlier today).

NS 2.2.3ff focus on arthāpatti, which is now said to be inconclusive (anaikāntika). NS 2.2.7 focuses on abhāva, which is now said to be not a pramāṇa at all, because there is not a corresponding prameya. Please notice that early Nyāya does not accept abhāva as a separate category, whereas this will be later added as the seventh padārtha.

Uddyotakara explains: abhāva is not a source of knowledge, because there is no content for it. Uddyotakara also adds something which seems new, namely that there is indeed a pramāṇa at stake, but that this has as its content something existing (again, weird, given the status of abhāva as padārtha…).

NS 2.2.8 speaks again in favour of abhāva, insofar as it is used to cognise, among marked things, the ones which are not marked.
This seems to prove that abhāva is indeed a source of knowledge, as explained by Vātsyāyana and Uddyotakara. NS 2.2.9 goes back to the problem that there is no prameya for such a pramāṇa.

Uddyotakara on NS 2.2.12 explains that there are only two types of absence (prior and posterior absence). This is relevant, because Kumārila had discussed absence as being four-fold (thus, Uddyotakara clearly did not know this classification, since he does not even take notice of it). Interestingly, Vācaspati, commenting on Uddyotakara, feels the need to add that the fact that he mentions two does not mean that he refutes the others.

What does this all tell us about early Nyāya until Uddyotakara and absence?

• 1. that Uddyoataka only knows a 2-fold classification of abhāva as prameya.
• 2. that an early position saw abhāva as part of anumāna (this seems to be also Uddyotakara’s position at the end of his commentary on NS 2.2.2).
• 3. that abhāva is considered a useful epistemological category, e.g., to speak about things lacking a certain characteristic.

The above is based on the commentary on NS 2.2.1–12. I will come back to later passages of the NV.

Addendum: Jhā translates abhāva as “antithesis”. Don’t ask me why.

The range of omniscience
A problem (raised by Sudipta Munsi in a comment on this post) connected with the scope of omniscience regards the question of whether an omniscient being also knows all erroneous beliefs. At first sight it might seem that if she does not, she is not completely omniscient and that if she does, she shares also erroneous beliefs, which seems paradoxical. A possible way out consists in claiming that she knows all erroneous beliefs but she attributes them to us. In other words, she knows that I do not know about the place and year of birth of Kumārila, but still correctly knows where and when he was born. Is this solution satisfactorily? Possibly, although this kind of omniscience would lack the first person grasp on how it feels to not know that X or to hold a false belief.

A connected problem regards specifically God’s omniscience: Does God also knows what it is to be in pain? If He does not, He seems to be not omniscient. If He does, He is no longer untouched by sufferance (duḥkha), as claimed in Nyāya and Yoga. In other words, an Īśvara-like God (see below) cannot be said to have experience of duḥkha. His knowledge would nonetheless not be incomplete because duḥkha would be conceived as just a negative entity (the absence of pleasure), which does not need to be separately known. God would be omniscient insofar as He knows all states of affairs, without needing to know also their corresponding absences. By contrast, God as conceived in theistic Vedānta (see below the lines on Viśiṣṭādvaita Vedānta) can even be said to have experience of duḥkha, insofar as He is the inner controller of each conscious being and shares therefore their experience from within.

God’s omniscience
Nyāya authors accept the existence of a God, usually referred to as Īśvara, who can be proved to exist, and develop on this basis a rational theology which accepts His omniscience and omnipotence. They explain that Īśvara `Lord’ needs to be omniscient in order to deploy His functions, which include the re-arrangement of the world after each periodic destruction and the re-assignment of their karman to each living being. Accordingly, God’s omniscience needs to be understood in a robust sense as the knowledge of all present, past and future states of affairs and as completely actualised (against some Buddhist conceptions discussed above). This, however, entails some problem, insofar as the Lord’s knowledge needs to be at any time complete and is in this sense atemporal. But this seems to mean that (a) there is no space for human free will and (b) the Lord knows the world outside of time. He knows, in other words, all states of affairs simultaneously and independent of time. This mirror-like omniscience has been criticised by authors of the Buddhist epistemological school (see Moriyama 2014 and forthcoming).

Śaiva authors, especially of the Pratyabhijñā school, accept both omniscience of yogins and of the Lord/Īśvara. The first one is often referred to in discussions aiming at establishing the omnipresent nature of the Lord as the supreme subject. In fact, how could memory be possible, if there were not a single subjectivity connecting events from a subjective point of view? And how could knowledge be possible, if there were not a fundamental similarity of nature between knower and known things, which does betrays its partaking to the nature of the absolute subject? The Nyāya account of a plurality of subjectivity is rejected insofar as it clashes with cases like the yogins’ ability to access other minds. The yogin, explain Pratyabhijñā authors, knows other minds from within, as the subject of their thoughts, and does not take other minds as an object to be known, since this knowledge would not be a real knowledge of the other mind, which is intrinsically subjective and cannot be reduced to an object. This ability of the yogin depends on the fact that he has recognised his identity with the Lord and can therefore access any mind. The Lord, as the single all-pervading subject, is in fact de facto omniscient and liberation consists in recognising one’s identity with Him (see R. Torella’s studies on yogipratyakṣa in this school).

Vedāntic authors conceive of God as brahman, and therefore as the only absolute reality. In this sense, the brahman is not an additional entity in the world, and the latter only exists because of Him (Dvaita Vedānta), in Him (Viśiṣṭādvaita Vedānta) or does not exist ultimately (Advaita Vedānta). Knowledge is considered in Vedāntic school to be a substance. Advaita Vedāntins resolve the duality which would emerge out of the assumption of brahman and knowledge by stating that brahman consists of cit `consciousness’. This is unintentional, since any content would include duality.

Viśiṣṭādvaita Vedānta authors conceive of God as brahman and at the same time as a personal God. He is therefore the material cause of the world, which is conceived by Viśiṣṭādvaita Vedāntins to exist only as a specification of Him. Like in the case of Advaita Vedānta, knowledge is conceived as a substance. Unlike in Advaita, knowledge is intentional, and has as its content the whole world. The reality of the world is thus guaranteed by its being a specification of the brahman and by its being a content of His knowledge. At the same time, the brahman is conceived of as a personal God, which means that the two above mentioned ways of relating to the world are not mutually exclusive (as it happens to be the case in Spinoza’s pantheism). Rather, knowledge is connected to Him as His characteristic. It is not just one characteristic among many, nor is it connected to the Lord as a quality to its substrate. By contrast, Viśiṣṭādvaita Vedānta authors describe the relation between God and His knowledge as one of indissolubility. The two cannot be experienced the one without the other and, although knowledge is ultimately a substance, it behaves as a characteristic of Him (it is therefore called dharmabhūtajñāna `cognition [behaving] like a characteristic’).

Against omniscience
Basing on the same elements, the authors of the Mīmāṃsā elements altogether deny the possibility of omniscience. They explain that omniscience contradicts our experience, where knowledge always increases but never reaches on outmost limit. Against the argument of repeated exercise, they observe that exercise does not need to be able to reach whatever result. For instance, no matter how much one exercises, one will never be able to jump until the moon. Nor will one’s smell be able to perceive sounds, even after an intense training. Thus, there are intrinsic boundaries to each faculty, including one’s intellect, which cannot directly grasp things, without the mediation of perception, inference and the other instruments of knowledge.
Moreover, no one could judge the omniscience of someone else. Thus, claim the Mīmāṃsā authors, the accounts about the Buddha’s omniscience cannot be trustworthy, since no one but an omniscient can vouch for someone else’s omniscience.

Why do Mīmāṃsakas insist so much on the impossibility of omniscience? From an internal and argumentative perspective, because of their commitment to common experience, which should not be contradicted without a valid reason. From an external and socio-philosophical perspective, because their defence of the Veda depends on its uniqueness as instrument of knowledge for knowing dharma `duty’. It is clear that no other human instrument of knowledge could compete with the Veda, since all human instruments of knowledge can only grasp what there is and not what ought to be. However, if there were an omniscient human or divine being, then they could reasonably compete with the Veda and possibly even falsify it.

The Buddhist arguments against omniscience (see Moriyama 2014 and Moriyama forthcoming) are different, insofar as they object only against the Lord’s omniscience, but accept the Buddha’s one. The difference lies in the fact that the Buddha became omniscient, whereas the Lord is allegedly permanently omniscient. Hence, only in the case of the Lord’s omniscience one encounter paradoxes such as the ones seen above and regarding the incompatibility of temporality and omniscience.

वेदान्तिनां नये तुच्छस्याध्यारोपाविषयत्वात् कथञ्चिच्छब्दमहिम्ना शशशृङ्गपदेन विकल्पात्मकमनोवृत्तौ जातायामलीकत्वस्य विषयत्वमङ्गीक्रियते । तथापि विकल्पस्य ज्ञानत्वानङ्गीकारात् ज्ञानाविषयत्वमलीकस्य सम्भवति । अथापि विकल्पस्य ज्ञानाद्विविच्य प्रदर्शनाय तैः सत्त्वेन प्रतीत्यनर्हम् अलीकम् इत्युच्यते । उक्तानर्हताया अवच्छेदकञ्च किञ्चिद्वक्तव्यम् इति अत एव तन्नये तदेवावच्छेदकं तल्लक्षणं – सर्व्वदेशकालवृत्त्यत्यन्ताभावप्रतियोगित्वे सत्युत्पत्त्यादिशून्यत्वम् – इति सम्भवति ।
अथवा – उक्तप्रतियोगित्वे सति कालासम्बन्धित्वमेवालीकत्वं तदस्तु ।

तार्किकनये तु अलीकस्य ज्ञानसामान्याविषयत्वम् इति तन्नये न विकल्पवृत्तिरङ्गीक्रियते इति प्राप्तम् । अत एव ज्ञानाविषयत्वमेवालीकलक्षणम् । कालासम्बन्धित्वं वा ।

एतेषु सर्व्वेषु पक्षेषु इदं चिन्त्यं यत् –
तत्तन्मते तुच्छस्य यल्लक्षणं ज्ञानाविषयत्वदि तत् किं तुच्छे वर्त्तते न वा । वर्त्तते चेत् तस्यापि स्वरूपं प्राप्तं , नास्ति चेत् कथं तस्य तुच्छत्वम् ।
– इति ;
तुच्छस्यापदार्थत्वेनैव भेदप्रतियोगित्वादिभावधर्म्मानाश्रयत्वे सति पदार्थेषु तद्व्यावृत्तिः कथं सिद्ध्येत । तदसिद्धौ पदार्थानां तुच्छाभेदेनालीकत्वमापतेत् , तत्सिद्धौ च तुच्छे प्रतियोगित्वादिकमङ्गीकर्त्तव्यमापतेत् ।

– इति च ।

(My friend, Sudipta Munsi brought this post from the Bharatiya Vidvat Parisat to my notice and obtained permission from the author to cross-post it on this blog. Except for his name, the learned author, Srimallalitalalita, prefers to remain anonymous.)

• is the example of the presence of horns as leading to “non-horse” an instance of the way apoha works (as with Yoshimizu, which supports in this way his analysis of Dignāga’s procedure as entailing a compositional analysis) or just an example about an inference, which works in a way similar as the apoha, i.e., does not need to exclude elements one by one (as with Kataoka, who thus supports his claim that Dignāga does not need any positive postulation).

More details on each reconstruction can be found below:

Pramāṇasamuccaya and svavṛtti 5.43 (Pind’s reconstruction)

yac coktam <ādyapratyayo> nāstīti, iṣṭisiddhir anāditvāt. [43a]

[…]. yasya tu […] na ca śakyaṃ jātimad vyāptum, na ca […]. yad apy uktaṃ pratyayavṛttir eva nāsti, tad apy ayuktam.

sāmānyena nirākṛteḥ. [43b]

na hi so ’nyāṃ jātiṃ pratidravyam apohate, kiṃ tarhi vyavacchedyavivakṣayaikena sāmānyadharmeṇa. uktaṃ cātra vijātīye ’darśanamātreṇānumānam. tavaiva tv eṣa doṣaḥ. yadi svajātīyavyāptyā <varteta, vyāpyasyānantyaṃ syāt>. tasmād yathā <viṣāṇitvād anaśva ity vacane ’śve viṣāṇitvādarśanena tadvyavacchedānumānam>, na tu <karkādīn> pratyekam apohate, <nāpy ekaikeṣu gavādiṣu vartate. tavāpi vyāvṛttyanuvṛttibuddhimatam>. tathā <cā>tra nyāyaḥ.

Kataoka

tavaiva tv eṣa doṣaḥ. yadi svajātīyavyāptyā <varteta, vyāpyasyānantyaṃ syāt>. tasmād yathā <viṣāṇitvād anaśva ity vacane ’śve viṣāṇitvādarśanena tad- vyavacchedānumānam>, na tu <karkādīn> pratyekam apohate, <nāpy ekaikeṣu gavādiṣu vartate. tavāpi vyāvṛttyanuvṛttibuddhimatam>. tathā <cā>tra nyāyaḥ.

NB: yathā connected with tathā: like it works in the case of inference, so here. Thus, yathā only introduces a diverging example, namely one about inference. atra means “like in the case of inference, so in our case (of apoha)”.

Yoshimizu

tavaiva [Mādhava] tv eṣa doṣaḥ. yadi svajātīyavyāptyā <varteta, vyāpyasyā- nantyaṃ syāt>. tasmād yathā <viṣāṇitvād anaśva ity vacane ’śve viṣāṇitvā- darśanena tadvyavacchedānumānam>, na tu <karkādīn> pratyekam apohate, <nāpy ekaikeṣu gavādiṣu vartate. tavāpi vyāvṛttyanuvṛttibuddhimatam>.

tathā <cā>tra nyāyaḥ.

NB: yathā connected with what precedes, since it is part of a larger quotation of Dignāga’s previous text. tathā out of the quote and disconnected.

The quote is found in Muni Jambuvijaya’s edition of the Jain Dvādaśāra Nayacakra:

yathāha dvādaśaśatikāyām: yad apy uktam aprasaktasya kimartham pratiṣedhaḥ iti naivaitat pratiṣedhamātram ucyate, kintu tasya vastunaḥ kaścid bhāgo ‘rthāntaranivṛttyā loke gamyate yathā viṣaṇitvād anaśva iti.

atra would thus mean, according to Yoshimizu, “here, in this treatise [like in the Dvādaśaśatikā, whence the quote come from]”. This would also explain why in the PS Dignāga did not need to dwell at length on componential analysis, because he could rely on what he had said already in the Dvādaśaśatikā.

What do you think? The yathā-tathā sequence seems appealing, all the more because a tasmāt separates the previous sentence from yathā, unlike in the reused text, but the reused text seems to point to a stricter relation between the yathā-clause and what precedes it.

These are only my reconstructions of Pind’s, Yoshimizu’s and Kataoka’s thought as represented in, respectively, Pind’s PhD thesis, Yoshimizu’s paper discussed here and Kataoka’s papers presented at the last IABS and IDhK conferences. All mistakes are mine. For the first part of my reconstruction, see here.

K. Yoshimizu shows passages of the Mahābhāṣya showing that the denotation of gauḥ is described as involving various elements, such as dewlap, horns, hooves, humpback. According to Yoshimizu, Dignāga maintains that in actual usage language users acknowledge the presence of this elements in order to recognise what is a cow (and their absence in order to recognise what is a non-cow). Yoshimizu says that is process is akin to what contemporary linguists call “componential analysis”. He quotes passages from Dignāga’s PS which apply it even to proper names (since also “Ḍitta” describes a set of qualities, such as being adulterine, having one-eye only, etc.).

Componential analysis cannot work, by contrast, for the “founder” of Bhāṭṭa Mīmāṃsā, Kumārila Bhaṭṭa (who knew Dignāga and criticised his work), since he maintains that the universal “cowness” is directly perceivable and that this is what allows us to recognise a cow before we recognise its dewlap, etc. In this sense, the meaning of a word denotes, for Kumārila, a universal, and can only secondarily be analysed in its sense-components.

This leads Yoshimizu to a further question, namely, how can one perform an injunction, if this referes to a universal? One would never be able to bring either the universal cowness, nor all its instantiations (i.e., all cows) once one has been enjoined to “Bring the cow!”. Fortunately enough, the word “cow” in such a command refers to all individual cows, but one by one (so Kumārila in the TV). How is this possible? Because Kumārila distinguishes two elements in each prescription (what is uddeśyamāna and what is upādīyamāna*), which Yoshimizu equates to what contemporary linguists call “topic” and “comment”. The “comment” adds new information, whereas the “topic” is what we know already about. This part is only needed in order to understand what the comment is about. For instance: “cow” is “comment” and then “topic” in the next two sentences (the example is mine, no responsibility of Yoshimizu in any mistake it may contain):

“That one is my cow” (topic: “That one” (you already know about it since a gesture indicates it); comment: “my cow” (you did not know before the speaker had a cow))

“Bring the cow” (topic: “the cow” (you already know that the speaker has a cow, and which one it is); comment: “Bring [it]!”)

Once the “cow” is made into a “topic”, one knows already its number (in this case, singular) and can identify it easily. Thus, one does no longer need to bring all possible cows sharing the universal “cowness”.

Yoshimizu’s conclusion is that Kumārila leans towards pragmatics (for instance, he implements a topic-comment distinction which takes into account the pragmatic presuppositions implied in a certain linguistic act), whereas Dignāga implicitly presupposes some type of componential analysis.

A dewlap is not always sufficient as a probans to infer a cow

What do you think of the application of contemporary theories to classical Indian philosophy? Do they help or bewilder you?

More on Kataoka’s view of apoha can be read here If you are in Vienna and you want to discuss these topics with Yoshimizu, consider attending this workshop.

*On “topic” and “comment” applied to Mīmāṃsā linguistics one can also read Yoshimizu 2006, where the “comment” is equated to the vidheya

More in detail, he reconstructs how Vācaspati (probably, I would add, since there is always the chance of a shared background of live discussions, what I call “interlanguage”) derived some of his anti-apoha arguments from Sucarita:

1. First, both Sucarita and Vācaspati follow Kumārila in starting their rebuttal of apoha by questioning its locus (āśraya, “reference” in linguistics). If this is a vikalpa ‘unreal conceptual construction’, then how does it come that I keep on understanding the same thing when I hear the word “cow”? If the reference is as ephemorous as an erroneous concept, the meaning of the word should change without interruption. Kataoka’s argument is strengthened by the fact that Jayanta, the other early-but-post-Dharmottara opposer of apoha, uses a different strategy.
2. both Sucarita and Vācaspati note that apoha cannot at the same time express something positive (vidhirūpa) while only being a conceptual construction (kalpita). Here Kataoka is on a less sure ground, given that the problem of the contrast between the negative nature of apoha and the fact that it seemingly expresses positive (vidhirūpa) entities had already been noted by Jayanta (NM, apohadūṣaṇa, sections 2 and 3.1 of Kataoka’s edition), but it is also true that Jayanta does not juxtapose vidhirūpa and kalpita.
3. Last comes an interesting point. Sucarita and Vācaspati agree in describing Dharmottara’s position as descriving the word-meaning as something conceptually constructed and ultimately false (alīka). However, in a later passage Vācaspati uses a more complex term, namely alīkabāhyatva. This might be an evidence of the fact that Vācaspati was following Sucarita (and not the other way round), but incorporated a term which had just been introduced in the debate (possibly by Jñānaśrīmitra, who speaks of āropitabāhyatva).

This last point is convincing as for the relative chronology of Sucarita and Vācaspati, but I must admit that I am not sure I understood what is conceptually at stake in this terminological change. Does it mean that Sucarita is just speaking of the apoha as false, whereas Jñānaśrīmitra speaks of this falsity as appearing as if it were external? If so, then this seems to be just another way to state Dharmottara’s position that apoha is neither internal nor external, but rather an internal construction which appears as if it were external (kaścid āropita ākāraḥ). Jñānaśrīmitra and Vācaspati may have introduced a terminological novelty, but I am not sure whether there was also a conceptual novelty beyond it.

Do you identify a development in the apoha theory after Dharmottara?
For another post on Kumārila’s commentators, see here. On Dharmottara’s position on apoha, see here and here. On apoha you might also enjoy this post.