Persistence and Cognitive Aention in Udayana’s Ātmatavaviveka
David Nowakowski
Many Buddhist philosophers in India were concerned with defending a rad- ical thesis in metaphysics: nothing which exists can persist; that is, nothing can exist for more than one of the single smallest instants of time. Because on this ac- count everything exists for only a single moment, the view is known as universal momentariness. Udayana, writing in the Nyāya tradition of Brahmanical (so-called Hindu) phi- losophy, devotes the first chapter of his Ātmatavaviveka (“e Inquiry into the Reality of the Soul,” ĀTV), composed around , to the refutation of the mo- mentariness thesis, and to establishing the contrary view that things can and do persist over time. In this paper, I will be examining just one of Udayana’s argu- ments against momentariness, namely that because we are unable to cognitively aend to non-existent things (such as the son of a barren woman, or the horn of a rabbit), it is impossible for the Buddhist to know that everything is momentary. My main concern in this paper is to answer this question: What is the role of considerations about cognitive aention in arguments for or against the per- manence of inanimate objects, in classical Indian philosophy? In the course of answering this question, I will examine related topics concerning the manner in which we cognitively aend to particular objects, and limits of such aention. In section , I explain the terms and concepts required to properly state Uda- yana’s claim. In section , I provide a brief overview of the theory of inferen- tial reasoning in classical India, and situate the Buddhist’s demonstration of mo- mentariness (which Udayana will refute) within that theory. And in section , I examine one of the arguments which Udayana deploys against the inference for momentariness, which turns on our inability to cognitively aend to non-entities.
Unpaing the Claim: Topics and Terms
Udayana makes the following claim, in the course of his refutation: e Buddhist is unable to know the momentariness of everything that exists, because it is im- possible to cognitively aend to non-existent things. To properly understand this
D N — W S Page of claim, we must consider four terms: the Buddhist, momentariness, non-existence, and knowing. In this section, I will explain each of these in turn.
. e Buddhist Udayana’s discussion in the ĀTV is structured, in a style typical of Sanskrit philo- sophical texts, as a conversation between the defender of the correct view (who speaks in Udayana’s own voice) and an objector, who must be disabused of his var- ious false notions. Here, the objector is trying to defend the Buddhist momentari- ness thesis. Various actual philosophers in the so-called “Buddhist epistemologi- cal tradition,” including Dharmakīrti (th c.), Dharmoara (th c.), Jñānaśrīmitra (late th c.), and Ratnakīrti (early th c.) argued for the momentariness thesis on the basis of existence.¹ While the arguments offered by the objector in the ĀTV oen resemble those of these actual Buddhists, Udayana’s text almost never quotes any of these Bud- dhist philosophers directly. e objector, then, is not any particular individual, but rather a generic advocate of Buddhist views. So by the expression “the Bud- dhist,” I will refer to this literary construct, and not to any historical person.
. Momentariness e momentariness thesis seems to be a development of the Buddha’s teaching that “everything is impermanent” (sarvam anityam).² In this statement, the rel- evant distinction is between being permanent (nitya) and impermanent (anitya). To be permanent is to exist forever—a condition Udayana and his fellow Nyāya philosophers aribute to God, universals, space, ether, and atomic substances, among other things. To be impermanent is simply not to exist forever, that is, to cease to exist at some point in time. e Naiyāyikas would cite ordinary, com- posite material substances as prime examples of impermanent things. Universal impermanence, then, is simply the claim that all of the things which allegedly be- long to the first group (God, universals, atoms, etc.), if they exist at all, originated at some time and will cease to exist at some time. e issue of impermanence must be distinguished from the separate, though related, issue of momentariness. An entity is momentary, just in case it exists for
¹at is, “existence” is the reason property (H) in the formal statement of the inference, as I discuss in section below. ²at the momentariness thesis is a later elaboration is evident, inter alia, from the intra- Buddhist debates concerning the view which extended at least until Vasubandhu, writing in the fih century A.D. Alexander von Rospa () provides a lucid account of the early development of the momentariness thesis. D N — W S Page of
P? M? E Yes No X exists forever. . No No X exists for a finite but extended duration. No Yes X exists for only a single instant.
Table : Permanence and Momentariness: ree Alternatives precisely a single instant of time, before immediately ceasing to exist. Momen- tariness is a much stronger claim than mere impermanence. Mere impermanence leaves open the possibility that a thing could persist for some finite but extended interval, while momentariness is the denial of that possibility. Combining these two pairs of terms, we have a three-fold classification, into which every entity X must find a place, as shown in table . We note that momentariness and permanence each logically preclude the other, while non-momentariness and impermanence each leave open both possi- bilities with regard to the other term: Something which is non-momentary might or might not be permanent, and something impermanent might or might not be momentary. e weak claim that “everything is impermanent” (sarvam anityam) is simply that nothing is of type , leaving open the question of which things (if any) are of types and . e stronger claim “everything is momentary” (sarvaṃ kṣaṇikam), advanced by Udayana’s rivals in the Buddhist Yogācara school, is that everything is of type . Udayana himself is commied to the view that nothing is momentary, which is to say, everything which exists belongs either to type or type .³ Various intermediate positions are, of course, also logically possible.
³ere are at least two considerations which show that Udayana is commied to “universal per- sistence.” First, it is implied by Udayana’s critique of the positive concomitance in the Buddhist’s inference from existence in ĀTV I.A (which I discuss in my dissertation). If Udayana acknowl- edged anything which was both existent and momentary, then he would be unable to claim that positive concomitance (anvaya) between existence and momentariness was unestablished; his Bud- dhist rival could simply point to that momentary existent thing as the requisite similar example. Second, universal persistence follows from Udayana’s metaphysics of destruction, which will be involved in the refutation of the Buddhist’s second inference, from the inevitability of destruction, in ĀTV I.. A thumbnail sketch of the argument is as follows: . Causes and their effects must be co-present, at the time of the arising of the effect. . e prior absence of the pot is among the causes of the pot. (Insert any other finitely- persisting entity in the place of the pot.) . e pot is among the causes of this posterior absence/destruction. . e prior and posterior absences of the same thing cannot be co-present. . erefore, the pot must persist for at least two instants of time (with a different one of the two temporal absences being present at each of those two instants). D N — W S Page of
. Non-Existence Before examining Udayana’s arguments about our knowledge of, and reference to, non-existent objects, it will be helpful to clarify what it means for something to be non-existent. Within the ontological framework Udayana inherits from the Vaiśeṣika philosophical tradition, there are two distinct levels on which we could explain the expression “X does not exist” or “X is not” (tan nāsti).⁴ According to the Vaiśeṣika philosophers, every item necessary for giving a complete account of the world is contained within one of seven categories (pa- dārtha): substances, qualities, motions, universals, individuators, inherence, and absences. ese categories may, in turn, be grouped in various ways. I will consider two possibilities for interpreting “X is not” within the Vaiśeṣika ontology inherited by Udayana: the distinction between positive beings and ab- sences, and the distinction between real entities and unreal non-entities. Once these two options are on the table, I will consider the usage of Udayana’s Bud- dhist opponent in this section of the Ātmatavaviveka. e diagram in figure , on page , will help to organise this discussion. e first major division is between absences (the seventh category) on the one hand, and positive beings (the other six categories) on the other. Positive beings, Udayana explains, are “the cognitive objects of cognitions whose cognitive ob- ject is expressible without a negative particle,”⁵ including the ordinary particulars which comprise the first three categories, along with the other metaphysically necessary entities (universals, individuators, and inherence) which make up the next three categories. Absences (abhāva), the cognitions of which are (necessar- ily) expressed with a negative particle,⁶ are genuine, locatible, causally efficacious features of the world, which can be directly perceptible under the appropriate conditions.⁷ Nevertheless, they are intimately connected with the positive beings which comprise the other six categories, in that every absence is “the absence of X,” where X is some specific absentee (pratiyogin). For example, we might find the absence of a pot, or the absence of blue color.
⁴ere is actually a third interpretation of non-existence on the Vaiśeṣika model, namely the lack of a relation to the genuine universal “existence” (saāsambandha). Given the Buddhist’s outright rejection of real universals, this interpretation of “X does not exist” in terms of relation to a universal will play no role in Udayana’s discussion in the Ātmatavaviveka. ⁵LĀ : nañarthaviṣayatvarahitapratyayaviṣayo bhāvaḥ. ⁶LĀ : nañarthaviṣayapratyayaviṣayo ’bhāvaḥ. ⁷Indeed, giving rise to perceptual cognitions is one of the ways in which absences are causally efficacious, and the phenomenal differences between the positive and negative cognitions—e.g., where the former is of the bare floor or the empty doorway simplicter, and the laer is of the absence of a pot on the floor or the lack of a man in the doorway—suggests that there are two distinct effects, which require distinct causal explanations. (And for the Indian realists, like their pre-Humean counterparts in the West, causes are paradigmatically things, not events.) D N — W S Page of
is account of absences as part of the fundamental, objective structure of the world may strike some as a wanton multiplication of entities. To a certain (but limited) extent, this is the price of the Vaiśeṣikas’ direct cognitive realism, which has great ontological parsimony elsewhere, in that it does not require mental du- plicates of physical objects, to function as the direct objects of cognition and as intermediaries between mind and world.⁸ But aer such purely mental objects are ruled out,⁹ the direct realist still requires objects for negative, as well as pos- itive, cognitions. ose cognitive objects, in the case of negative cognitions, are absences.¹⁰ It does turn out, on this account, that we are surrounded by a great many entities which we do not ordinarily observe. Nonetheless, these entities do important philosophical work in accounts of causality and cognition, and even if they are not always observed, they are nonetheless observable under the appro- priate conditions.¹¹ At a higher level of generality, a second major distinction is available to Uda- yana. is is the difference between real entities (vastu) and unreal non-entities (avastu). Here, real entities just include all seven of the Vaiśeṣika categories; that is, the six categories of positive beings together with real absences (abhāva) in their four varieties. ese, and only these, things are said to be nameable (abhi-
⁸e Buddhist epistemologists are happy to accept such intermediate, purely mental cognitive objects, ākāra. is puts familiar pressure on the connection between these mental images and the physical objects in the external world, such that by the time of Jñānaśrīmitra, we find the Buddhist upholding a full-blown idealism. ⁹ere are, on the Nyāya-Vaiśeṣika model, some special cases of cognitions whose cognitive objects happen to mental: most notably, higher-order cognitions, which have as their object either another cognition simplicter, or a property which belongs to another cognition (such as the property of being knowledge, prāmāṇya). But while these cognitive objects happen to be mental, they do not in any way function as intermediaries which convey or represent some other physical object. In these special cases, even the thorough-going realist says that the mental object is all there is. ¹⁰In place of this Vaiśeṣika account of a locus possessing an absence, the alternative employed by the Buddhist epistemologists as well as, arguably, some Mīmāṃsakas (cf. Kellner ) is to analyse putatively negative cognitions as having only the locus as the object. at is, the positively- expressed cognition “ere is a (bare) floor” and the negatively-expressed cognition “ere is no pot on the floor” both have precisely the same object, the floor. e laer cognition is a different way of thinking about the same object; it is not a cognition of a different object. ¹¹An infinite multiplication of higher-order absences (e.g., the absence of the absence of the absence of a pot, which we might symbolise as ~~~X) is ruled out by identifying any first- or second- order absence (~X or ~~X) with the absence of the absence of that higher-order absence itself (~~~X or ~~~~X, respectively). e Naiyāyikas’ logic of absences thus resembles the rules of Intuitionistic logic for negation, according to which any application of the rule of double negation elimination is permied, provided that the resulting expression still begins with at least one negation (so, the Intuitionist inference from ~~~X to ~X is permied, while the inference from ~~X to X is not). e possible combinations which result from having four varieties of absence (as opposed to only a single kind of negation) add some extra complexities, but this is the basic principle. D N — W S Page of Absolute atyantābhāva avastu Posterior Relational pradhvaṃsa (Unreal) non-entities Prior abhāva Absences prāgabhāva Mutual anyonyābhāva vastu Inherence samavāya Real entities () viśeṣāḥ svātmasava Individuators ) sāmānya astitva Universal ( Figure : Levels of (Un)reality in Udayana’s Vaiśeṣika Ontology bhāva Positive beings Motion karman () guṇa ality saāsambandha dravya Substance D N — W S Page of dheya) and cognisable (jñeya).¹² It will also turn out, on Udayana’s view, that all and only those things which are included in the seven categories of real entities have the capacity for causal interactions. Indeed, the seven categories are posited precisely because they are all necessary for providing an account of how and why the world is the way it is. Even absences (abhāva) play a variety of causal roles, accounting for production and destruction of material things, as well as for our negative perceptual aware- nesses (e.g., “ere is no pot here,” or “is man is not Devadaa”). A brief note on translation practices: I will reserve the term “entity” (or occa- sionally “real entity,” for emphasis) for translating vastu, and the corresponding term “non-entity” for translating its negation, avastu (occasionally adding the ad- jective “unreal” for emphasis). “Entity” and “non-entity” will mark exclusively this distinction at the highest level of metaphysical generality. I will use the term “thing,” meanwhile, for terms like artha where the Sanskrit does not carry any precise ontological commitments. e term “object” will likewise be ontologically uncommied, while “cognitive object” will always translate the Sanskrit viṣaya, referring to whatever it is that a cognition is a cognition of. To mark the subdivi- sion of entities, I will use “positive being” and “absence” exclusively to translate bhāva and abhāva, respectively. We see that Udayana has two quite distinct ways for something to not exist. On the intermediate level, there are absences, which do not exist insofar as they are not positive beings (bhāva), but nevertheless are causally efficacious real enti- ties. On the highest level, non-entities do not exist insofar as they are not entities (vastu), and on account of this non-existence are uerly unreal, and devoid of any causal efficacy whatsoever. roughout this discussion, it will be crucial to dis- tinguish () the division of positive beings and absences from () the division of entities and non-entities. A concrete example may help to clarify the distinction. A pot, which actually exists, is a positive entity, and is therefore numbered among the real entities. e absence of that pot is also a real entity, which is present in certain places (but not others), and about which, Udayana is happy to acknowledge, we may have all manner of cognitions. We can think about the absence of the pot, and make various positive and negative predications of it, which will be determinately true or false. e son of a barren woman, by contrast, is a non-entity. It cannot be located anywhere, nor can we coherently think about it, or make predications of it. Critically, because the son of a barren woman is entirely unreal, there is no “absence of the son of a barren woman,” since absences (in their technical sense, as the seventh class of real entities) are dependent upon a real counterpositive:
¹²For nameability in Udayana, cf. LĀ . See also Halbfass : –. D N — W S Page of either a positive being, or another absence. For the Buddhist, to exist just is to stand in causal relations with other things, and to exist as a particular kind of thing just is to stand in some particular causal relations and not other.¹³ So the relevant interpretation of “X does not exist,” or “X is not,” will involve the denial of all causal capacity to the non-existent thing X. is is to follow the second, more general, of Udayana’s two distinctions, since absences do (in the realist’s terms) have causal powers. So to use Udayana’s terms, in this context, to say that X does not exist just is to say that X is a non-entity (avastu). e point of all of this, is that there will be things which (at least on Udayana’s account) can be said to not exist (on the first level), which nevertheless do not help the Buddhist to establish his inference, which requires a non-entity (on the second level).
. Knowing Finally, what is it for the Buddhist to know that everything which exists is momen- tary? Epistemologists in classical India shared a common conceptual framework, in which knowledge (pramā) is a species of cognition (jñāna), and cognitions are events, brought about (like any other event) by some particular causal process at some particular point in time. Knowing, therefore, is best thought of as something occurrent; it is quite unlike the standing beliefs with which contemporary West- ern epistemologists are typically concerned.¹⁴ Perceptual knowledge, for instance, is a cognitive event which occurs for some agent as a causal result of his sense- organs becoming appropriately connected with some object, including (some or all o) that object’s properties. is special connection between sense-organ and object is the epistemic instrument (pramāṇa): the most efficacious causal factor, directly responsible for the perceptual cognition. e central project of Sanskrit epistemologists, then, is to determine what sorts of causes of cognition can count as epistemic instruments, and then to determine whether any given cognition was in fact produced by one of these instruments. All cognitions—even those which, due to some defect in the process of their generation, do not constitute knowledge—have some causal story, and can be dis- tinguished from each other on the basis of differences in the causes which gave rise to them. e issue in this portion of the ĀTV is where these deviant cog- nitions get their content from. at is, what makes it the case that a cognition C has the content that is does (making C the cognition of a blue pot rather than the cognition of a red cloth)? A cognition which is reliably indexed to its object—
¹³cf. PVSV .–.; PVin II k. . ¹⁴I offer a defense of this claim, which includes frequent reference to Udayana, in the second chapter of my dissertation. D N — W S Page of however that comes about—is said to be regular (niyama). Regularity is defined by this reliable indexing, while the explanation for why certain cognitions are regular will appeal to the causal history of those cognitions. Udayana discusses a class of cognitions which are “grounded in the epistemic instruments” (prāmāṇika). is class is more extensive than merely those cogni- tions whose immediate causal instrument was one of the epistemic instruments (and which therefore count as knowledge, pramā), though all of these are in- cluded. Rather, Udayana means to include under this term any cognition whose contents were first delivered to the agent by means of one of the epistemic instru- ments. To take a classic example, when Devadaa has a mistaken apprehension of a long, coiled object in front of him, cognising “this is a snake” in regard to what is actually a rope, Udayana would tell a story on which a process of recalling a previously perceived snake makes available “snake” as part of the content of Deva- daa’s present experience. Had Devadaa never seen (or otherwise had genuine knowledge o) a snake, he would be unable to cognise the rope as being a snake, since that cognitive object would be unavailable to him. is erroneous cognition, then, would still be included among those cognitions which are grounded in the epistemic instruments (prāmāṇika). From all this, we are in a position to conclude that a cognition C is regular, if C is grounded in the epistemic instruments. We will return to issues of regularity in section below.
How to Infer Universal Momentariness
In addition to perception, nearly every philosopher in classical India acknowl- edged inference to be a causal process generative of knowledge, which is to say, inference is one of the epistemic instruments. Inference is thus one of the means by which we come to cognitively aend to something; it is generative of (reliable) cognitive events. In section ., I will quickly outline the basic Indian Buddhist theory of in- ference. Nothing about the account I present here is especially novel, and my cursory exposition will necessarily gloss over many complexities. However, this background is necessary for understanding the Buddhist’s inference for universal momentariness, which I present, according to the general model, in section ..
. Inferential eory in Classical India To greatly oversimplify, an inference relates two properties and three groups of property-possessors. Here, I follow the Indian thinkers in their incredibly broad D N — W S Page of
E D S E P S Target Property the property to be established, as a fire result of the inference H Reason Property the property on the basis of which smoke the target (S) is established PB P Site of the locus/loci where the presence of S mountain Inference is doubtful SP Similarity Class loci where S is known to be kitchen hearth, present red-hot iron, &c. VP Dissimilarity loci where S is known to be absent lake, &c. Class
Table : Elements of an Inference use of “property” (dharma) to refer to anything that can be located in or on some- thing else. at “something else” is the property-possessor (dharmin). e two properties are the target property (sādhya, S), whose presence in cer- tain loci will be established as a result of the inference, and the reason property (hetu/liṅga, H) on the basis of which we can infer the target property. In the stock example, fire is the target property S, inferred on the basis of the reason property (H) smoke. As the example illustrates, “reason” is used here in a logical or epis- temic sense: we know that there is fire because of the reason, smoke. Smoke is obviously not the reason in a causal sense, since it is not the case that the fire in the world exists because of the smoke.¹⁵ Put another way, while the cognition of smoke causes the cognition of fire, the actual smoke does not cause the actual fire. e three groups of property-possessors are defined in terms of our epistemic status as regards the target property, prior to the inference. On the standard Buddhist account, each group must contain a minimum of one member. (e Naiyāyikas relax this requirement somewhat, permiing one of the three to be
¹⁵Strictly speaking, the Nyāya philosophers (including Udayana) define the epistemic instrument of inference as the special consideration of the reason property H. In the stock example, this would mean consideration of smoke insofar as it is a (logical, not causal) reason for fire. e Buddhists define the epistemic instrument somewhat differently, as the cognition of the target property S which directly follows upon the cognition of the reason H. Yet regardless of the details, all parties agree that both properties are necessary to a genuine inference. D N — W S Page of empty.) e three groups of property-possessors are: the site of the inference (pakṣa, P) in which the presence of S is uncertain, the similarity class (sapakṣa, SP) where S is known to be present, and the dissimilarity class (vipakṣa, VP) where S is known to be absent. In the stock example, the site P is standardly a distant mountain, the similarity class SP includes everything we know to possess fire, such as a kitchen hearth and a red-hot lump of iron, and the dissimilarity class VP includes every place where we know to lack the presence of fire, such as a large lake. In theory, every epistemically relevant property-possessor should fit into one of these three classes. In this way, P, SP, and VP are defined relative to the target property S. e inference is able to proceed only if the reason property H is appropriately related to each of the three groups.¹⁶ e required relations are defined in the Triple Conditions:
T (Being a Property of the Site): H is known to be present in every member of P; that is, in every place where the presence of S is uncertain. T (Positive Concomitance): H is known to be present in (at least) some mem- bers of SP; that is, in at least some of the places where S is known to be present. T (Negative Concomitance): H is known to be absent from all members of VP; that is, from every place where S is not present.
Taken together, conditions T and T describe a relation between the two prop- erties H and S known as pervasion (vyāpti). is is a two-place, one-way relation. S pervades H just in case every H-possessor is also an S-possessor. In terms of the stock example, fire pervades smoke, because in every place where smoke is found, fire is also found.¹⁷ In order to be a genuine inferential reason, a property must satisfy all three of these Triple Conditions. For the sake of brevity, I will not consider here either the necessity and sufficiency of the conditions in general, or the appropriate methods of demonstrating, for any particular inference, that they are satisfied.
. e Inference from Existence In the Ātmatavaviveka, Udayana’s Buddhist opponent states the inference from existence, for proving the momentariness of all things, in this way:
¹⁶is is for the Buddhist theory of Dharmakīrti, as revised by Dharmoara. e Naiyāyikas relax the second and third conditions slightly, and add two additional conditions. ¹⁷ere may, of course, be cases like the glowing lump of iron, which possess fire without smoke. D N — W S Page of
Whatever is existent is momentary, like a pot. And these things which are the subject of debate—sound and the like—are existent.¹⁸
In order to be a genuine source of knowledge, the Buddhist’s inference from existence must have the appropriate standard form, and when expressed in that form, must have a genuine inferential reason (rather than one of the three recog- nised varieties of pseudo-reason, hetvābhāsa). at standard inferential form, as we have seen, has five components: the target property to be proven (sā- dhya, S), the inferential reason (hetu, H), the site of the inference (pakṣa, P), an appropriately-exampled similarity class SP (sapakṣa), and an appropriately- exampled dissimilarity class VP (vipakṣa). Here, the first two components—the two properties—are easily identified. e target property S is momentariness, and the inferential reason H is existence. As we saw above, the three groups of property-possessors P, SP, and VP are defined in terms of the epistemic status of the two parties to the debate with regard to the target property S, prior to performing the inference. at means that in this case, SP should include all and only those loci which Udayana and his Buddhist opponent agree to be momentary, while VP should include all and only those loci which they agree to be non-momentary. And as the Buddhist explictly states, the site of the inference, P, is to include all those things whose momentariness is the subject of controversy between himself and Udayana. e trouble, as Udayana quickly observes, is that everything seems to belong to P, leaving SP and VP both empty. at is, the Buddhist takes it that everything is momentary, while Udayana does not acknowledge anything at all to be momentary. Everything which exists “has become the topic of debate,” and is included in P. By the Buddhist’s own lights, the inference will be successful if and only if the inferential reason satisfies all three of the triple conditions: being a property of the locus, positive concomitance, and negative concomitance. As we saw earlier, all three of these conditions are concerned with the relation of the various property- possessors to the inferential reason H, “existence.” e Buddhist must be able to show:
T that all members of P exist, T that some member(s) of SP exist, and T that all members of VP do not exist.
¹⁸ĀTV .–: yat sat tat kṣaṇikaṃ yathā ghaṭaḥ saṃś ca vivādādhyāsitaḥ śabdādhir. Uda- yana’s proximate source for this formulation is most likely Vācaspati Miśra, whose Tātparyaṭīkā (NVT) Udayana knew and wrote a commentary on. e only difference between the two for- mal statements (prayoga) of the inference is Vācaspati’s inclusion of the adverb sarvam, following Dharmakīrti’s usage. (So, NVT .: yat sat, tat sarvaṃ kṣaṇikaṃ, yathā ghaṭaḥ; saṃś ca vivā- dādhyāsitaḥ śabdādhir. D N — W S Page of I E momentariness existence “all these things about which there is debate” [a seed, at each moment of time] non-entities (the hare’s horn, the son of a barren woman, &c.) S E fire smoke mountain kitchen hearth, lump of red-hot iron, &c. lake, &c. Table : Elements of the Inference from Existence D be established, as a result of the inference presence of S is doubtful of which the target property is established loci where S is knownbe to present be absent E S Target Property the property which is to P Site of the Inference locus/loci where the P H Reason PropertyPB the property on the basis SP Similarity Class VP Dissimilarity Class loci where S is known to D N — W S Page of
Since P, SP, and VP are defined relative to momentariness and persistence, the conditions may be understood as:
T All those things whose momentariness is doubtful, have the property “exis- tence.” T Some things (and at least one in particular) which we know to be momentary, have the property “existence.” T All things (and at least one in particular) which we know to be persistent, lack the property “existence.”
Furthermore, as the second list of the triple conditions indicates, in the course of demonstrating T and T, the Buddhist must be able to adduce at least one similar example (sādharmyadṛṣṭānta)—a particular member of SP which is known to possess the property “existence”—and at least one dissimilar example (vaidha- rmyadṛṣṭānta)—a particular member of VP which is known to lack the property “existence.” By his silence on the issue, Udayana grants the rather trivial condi- tion T.¹⁹ Udayana structures his critique of the inference from existence around T and T. For the remainder of this paper, I will set aside Udayana’s concerns about T, and focus exclusively on T, the requirement for negative concomitance.²⁰ Here, the Buddhist objector must find a way of cognising, and talking about, the non- existent things which make up the dissimilarity class VP. Not only must he prove the general principle that whatever is non-momentary is also non-existent, but he must also demonstrate the persistence (i.e., the non-momentariness) of some particular non-existing thing. Only by such a demonstration will T be satisfied.
e Limits of Regularity
In section ., we encountered the principle () A cognition C is regular, if C is grounded in the epistemic instruments. Now, Udayana would like argue for a stronger position, which goes beyond this principle; namely, that there are only two sorts of cognitions: those which are grounded in the epistemic instruments (and are therefore regular), and those which are not regular at all. at is, he will defend
¹⁹T is trivial in regard to this particular inference, only because the property “existence” is so unobjectionable. is first of the triple conditions may, in the context of other inferences, turn out to satisfied only aer substantial philosophical work. ²⁰e discussion of negative concomitance occurs at ĀTV –. D N — W S Page of
() A cognition C is regular, only if C is grounded in the epistemic instru- ments. And from () and (), of course, we have the biconditional () A cognition C is regular, if and only if C is grounded in the epistemic instruments. e Buddhist is happy to accept (), but challenges () by raising the possibility of a third sort of cognitions: those which are not grounded in the epistemic in- struments, but are nevertheless regular in regard to their cognitive objects. Purely mental concepts (vikalpā), he claims, can be reliably aended to even when there is no external object (artha).²¹ In this way, the Buddhist hopes to find the required certainty about non-entities like the hare’s horn. e objector’s claim, then, is the following: ere are (reliable) cognitions which are not founded in the epistemic instruments. is must be distinguished from the weaker claim, that there are cognitions which are themselves not directly generated by any of the epistemic instruments. Udayana can gladly agree to this weaker claim, since he acknowledges a great variety of such cognitions: recollec- tion, doubting awarenesses, hypothetical reasoning, and doubt.While such cog- nitions lack the virtue of being produced by an epistemic instrument, they are nonetheless part of our cognitive life. For the weaker claim to be true, there must be cognitions whose most immediate, preeminent causal instrument (karaṇa) is something other than one of the epistemic instruments (pramāṇa). But in eval- uating the stronger claim, by contrast, we must consider all of the causes while are collectively responsible for a cognition, and not merely the preeminent causal instrument. For the stronger claim to be true, there must be cognitions which do not depend upon the epistemic instruments in any way whatsoever. Udayana denies the stronger claim, holding instead that even our erroneous cognitions are ultimately grounded in the epistemic instruments. On his account, when we cognise something which is not actually present or does not actually exist (say, horns on the head of a rabbit), or misidentify one thing as another (for example, taking mother-of-pearl to be silver), the erroneous cognition is built out of several components, each of which really exists somewhere in the world, and each of which has been made available to us by some antecedent knowing event (pramā).²² So while the preeminent causal factor in erroneous cognition is not an epistemic instrument, prior operation of the epistemic instruments is
²¹ĀTV .–. ²²e horned rabbit is Udayana’s primary example throughout this section. e misapprehension of mother-of-pearl as silver, which is quite popular in the Indian philosophical literature, is not mentioned by Udayana in the first chapter of the ĀTV, though his commentator Bhagīratha does introduce it in explicating this section (cf. ĀTV –). D N — W S Page of necessary for the production of such a cognition, and so that cognition is still said to be grounded in the epistemic instruments (prāmāṇika). In order to prove his stronger claim, then, the Buddhist must provide a case in which even this minimal grounding in the epistemic instruments is not required. e imagined cognition must be generated entirely by mental factors, with no input from the instruments of knowledge. Udayana responds to the Buddhist’s strong claim, that there are reliable cogni- tions of non-entities which are not grounded in the epistemic instruments, in two ways. First, he argues that the Buddhist’s account of error, which underwrites the strong claim, is incorrect (ĀTV –). In what follows, I will examine the var- ious accounts of error under consideration, and the arguments Udayana makes to decide between them. Second, seing aside the first response, he argues that even if the strong claim were true, the Buddhist would still be unable to establish the negative concomitance in the inference from existence (ĀTV –). I consider this second response elsewhere in my dissertation.
. Accounting for Error Udayana offers two reasons for rejecting the Buddhist’s theory of error. First, there would be no cause for cognitions of the sort that the Buddhist postulates. Second, we would face the undesirable consequence that such cognitions (of the hare’s horn, and the like) would be useful for interpersonal communication and the aainment of practical ends, despite their being disconnected from the world of real, causally efficacious things.²³ In arguing for the former reason, the most Udayana can do is to consider in turn each of the major possibilities for what such a cause could be. e second reason is independent of the first, and directly provides a reason for rejecting the Buddhist’s account of error. e section opens when Udayana presents two alternative accounts of erro- neous cognition, asking Should the cognition “hare’s horn” be the mention of what is other- wise, or the mention of what does not exist?²⁴ e first option is the view of Udayana’s own Nyāya school, according to which such an erroneous cognition has the form of “a mention of what is otherwise” (anyathākhyāti). On this view, the cognition arises through a series of four steps. As ever for the Naiyāyikas, the perceptual case is the paradigm; appropriate mod- ifications to the account can easily be made for cognitions related to the other epistemic instruments. I will present the four steps, along with two examples:
²³e summary of the argument comes at the end (ĀTV ). ²⁴ĀTV .–: śaśaviṣāṇam iti jñānam anyathākhyātir vā syāt, asatkhyātir vā. D N — W S Page of one for simple misapprehension, where a cogniser mistakes mother-of-pearl for silver, and one for the case of a composite cognition, where a cogniser takes there to be a horn on the head of a rabbit.²⁵ . First, there is an initial connection between the sense-organ and the object. . is is followed by the apprehension of the object in its generic nature. at is, the cogniser apprehends only the features which would be common to the erroneous cognition, and the related genuine perceptual cognition. In the first example, these features would include the shape and reflective lustre of the object. In the second example, these would include the body of the rabbit, along with two “somethings” sticking up on its head (which are in fact the animal’s ears, but which will later be mistaken for horns). . is is followed by a recollection of something previously perceived (or known by way of another epistemic instrument). In the first case, the cog- niser recalls silver. In the second case, he recalls horns seen elsewhere, perhaps on another animal. . Finally, that recollection triggers an “extraordinary perceptual connection” (alaukikasaṃnikarṣa) with an actual entity existing elsewhere. In the first case, there is extraordinary perception of some silver which exists in an- other place; in the second case, perception of actual horns which exist else- where. e resulting cognition mistakenly connects this actual entity with the object apprehended in step (). For the Naiyāyika, then, the silver or the horns which are cognised are merely absent from a certain place, but exist somewhere else. ey are not uerly non- existent. Without the antecedent awareness of silver, or of horns, there will be nothing for the cogniser to remember in step (), and the erroneous cognition will not be able to arise. e object of the final cognition is not completely imagined, but is a real thing cognised in a way other than how it actually is. us, the name of the view. e distant silver is presented as though it were right here (where the mother-of-pearl actually is). e horns are presented as though they are on the rabbit’s head. e Bhāṭṭa Mīmāṃsa philosophical school advocates for a closely related view, which is mentioned by Udayana in the course of this discussion. is is the ac- count of error as “mention of what is contrary” (viparītakhyāti). e Bhāṭṭa ac- count of error agrees with the first three steps of the Nyāya account presented
²⁵My exposition of these steps closely follows that of Rao (: ), who focuses on the former, simple example. D N — W S Page of above, diverging only at step (). Where the Naiyāyikas posit an extraordinary perception of the erroneous component of the cognition (the silver, in the first example), the Bhāṭṭas claim that only the present locus is perceived, while the erroneous element (the silver) is supplied directly from memory.²⁶ e cogni- tion is erroneous insofar as what is actually a recollection is presented as being a perception. Still, the advocates of both of these views agree on the need (in step ) for recollection, ultimately based on some other genuinely pramāṇic awareness, and so they would (and historically were) happy to make common cause against the Buddhists. Owing the close similarity of the two views, the term viparītakhyāti is used by some Naiyāyikas, such as Jayanta Bhaṭṭa, to refer to their own view.²⁷ So much, then, for the account(s) of error favored by Udayana.
. Refuting the Buddhist’s eory of Error Let us return to Udayana’s question, which opened section . above. e second option, according to which the erroneous cognition is “the mention of what does not exist” (asatkhyāti), is the position of the Buddhist opponent, which Udayana must refute. On this view, the cognitive object does not exist anywhere at all, but is instead purely mental. is entails the denial of steps () and () of the Nyāya and Bhāṭṭa Mīmāṃsa accounts. Udayana has very lile to say in defense of his own account of error. In the present context, he does not need to convince his opponent of every detail of the Nyāya theory. Should the Buddhist be forced to accept any view which in- volves superimposing one real object upon another—whether that occurs through the Naiyāyikas’ extraordinary perception, the Bhāṭṭas’ recollection, or some other method entirely—then the Buddhist has already lost the argument. Such superim- position is directly opposed to the Buddhist’s thesis, that purely mental concepts arise independently of any real object.
²⁶As a group, the Naiyāyikas tend to remain firm in postulating an extraordinary perceptual con- nection, as theoretically cumbersome as it may seem. e motivation becomes more clear, when we recall that the main classificatory division of cognitions is between awareness (anubhava/anubhūti) and recollection (smṛti), where the laer may be defined as “cognition produced solely from the traces of prior cognitions” (SS .–: smṛtiḥ punaḥ pūrvavijñānasaṃskrāramātrajaṃ jñānam u- cyate), while awareness is defined as cognition which is other than recollection (SS .: anubhūtiḥ sā smṛter anyā). Given the requirement (in steps and of the process of error) for some operation of the sense organs in regard to a newly presented object, erroneous pseudo-perception clearly fails to meet the definition of recollection, and therefore must operate by another causal process. Contra the Bhāṭṭas, error is not a recollection, though recollection is among its causes. I discuss this classification of cognitions at greater length in the second chapter of my disserta- tion. ²⁷cf. Rao : . D N — W S Page of
Udayana refutes the Buddhist account of error as the mention of what does not exist (asatkhyātivāda), by showing that no causal story on offer can explain how such erroneous cognitions are produced in a suitably regular, contentful way. ough his refutation may initially resemble a laundry list, Udayana in fact covers all the major possibilities in a fairly systematic fashion. ere are three major alternatives, for what the source of regularity of erroneous cognitions could be:
. through some operation of the sense-organs or other epistemic instruments . through some feature associated with the nature of language . through mental factors, known as vāsanās.
Udayana considers each of these in turn.
.. Pseudo-Perceptual Awareness Udayana begins by ruling out the sense-organs as the cause, since the sense- organs operate only when acted upon by an object. If there is no entity, then there will be no (pseudo-)perceptual awareness.
.. e Power of Language e causal capacity of defective verbal communication is likewise constrained by regular paerns of usage. As Udayana notes, if this were not the case, then upon hearing the word “rabbit’s horn,” we could just as well have a cognition of tortoise hair.²⁸ To prevent this, and allow for language to be the source of our cognitions of non-entities, there must be some source of regularity (niyāmaka), binding certain words to certain objects. Either this source of regularity belongs to the words by their own nature, or else it must be something that human beings give to the words by convention. Dharmakīrti seems to take this laer approach, but in a way that will be unhelpful to his intellectual heirs in the later Buddhist tradition. For him, real things in the world are supposed to be the regulators.²⁹ is does nothing to solve the problem of unreal non-entities. For Udayana’s opponent to succeed, we cannot regulate our terms by the simple baptism of pointing out an external object which is part of our shared perceptual experience and giving it a name. And to posit regularity by assigned convention without such baptism is viciously circular.³⁰ If we escape the regress by assigning meanings to some limited group of terms by baptism, and allow those conventions to inform other terms which have
²⁸ĀTV .–. ²⁹cf. Yoshimizu . ³⁰ĀTV .–. D N — W S Page of no baptismal meaning, then we all superimposing one thing on another, which is just the realists’ (Nyāya and Mīmāṃsa) position.³¹ If the Buddhist takes the other horn of the dilemma, claiming that words have meaning just by their own nature, then we face the undesirable consequence that cognitions would arise even for someone untutored in language, simply by the innate force of the words.³² So much, then, for pseudo-testimony as the cause of cognitions of non-entities.
.. Purely Mental Factors e final option available to the Buddhist is to appeal to purely mental factors (vā- sanā), to account for the generation and regularity of cognitions of non-entities. Such mental factors are standing constituents of a person, responsible (on at least some accounts) for recollection and abstract/generic awareness. e twentieth- century commentator Kṛṣṇamādhava Jhā glosses vāsanā with as saṃskāra, the Naiyāyikas’ technical term for the traces of prior cognitions which generate rec- ollection.³³ Udayana again presents a dilemma: If the Buddhist takes this route, does this mean that the mental factors alone are the cause of such cognitions, or that mental factors together with some other causes produce the cognition? e second option is untenable, insofar as Udayana has just refuted the other possible causal factors. e first option also fails, since the Buddhist cannot ac- count for why these traces, which are always present for an individual, produce their cognitions only some of the time. Appeal to other cooperating causes is unhelpful, since this is merely to move to the other horn of the dilemma. From all of the considerations in this section, we, with Udayana, can conclude that a cognition C is regular, only if C is grounded in the epistemic instruments. No other explanation for that cognitive regularity is available. For this reason, the Buddhist cannot have the cognition of non-entities which he needs in order to have knowledge of the momentariness thesis.
Conclusion
We have seen that the Indian Buddhists’ inference for universal momentariness is inseparable from considerations about cognitive aention. As Udayana shows, in order to know the momentariness thesis by means of the inference from existence, the Buddhist must be able to aend in a regular way to specific non-entities. And
³¹ĀTV .– ³²ĀTV .–: śabdasvābhāvyāt tu niyame vyutpannavad avyutpannasyāpi tathāvidhavikalpo- dayaprasaṅgād iti. ³³. D N — W S Page of as Udayana further argues, because such regular cognitive aention is impossible, the Buddhist is unable to prove (at least by this, most favored inference) the truth of the claim that nothing persists.
Abbreviations
ĀTV Ātmatavaviveka (Udayana) in Dvivedin and Dravida LĀ Lakṣaṇāvalī (Udayana) in Tachikawa NKA Nyāyakusumāñjali (Udayana) in Upādhyāya and Śāstrī NVT Nyāyavārikatātparyaṭīkā (Vācaspatimiśra) in Nyaya-Tarkatirtha and Tarkatirtha NVTP Nyāyavārikatātparyaṭīkāpariśuddhi (Udayana) in akur PVin II Pramāṇaviniścaya (Dharmakīrti), Chapter II, in Steinkellner PVSV Pramāṇavārikasvavṛi (Dharmakīrti), in Gnoli SS Sārasaṁgraha (Varadarāja) in Dvivedin TR Tārkikarakṣā (Varadarāja) in Dvivedin
Bibliography
Dvivedin, Vindhyeśvarī Prasāda, . e Tārkikarakṣā and Sārasaṁgraha of Varadarāja. Varanasi. First published serially in e Pandit, –. ———, and Lakshmana Sastri Dravida, eds. . Udayanācārya: Ātmatavaviveka. Bibliotheca Indica . Calcua: Asiatic Society. Reprinted . Gnoli, Raniero, ed. . e Pramāṇavārikam of Dharmakīrti: e First Chapter with the Au- tocommentary. Serie Orientale Roma . Rome: Istituto Italiano per il Medio ed Estremo Oriente. Halbfass, Wilhelm. . On Being and What ere Is: Classical Vaiśeṣika and the History of Indian Ontology. Albany: State University of New York Press. Kellner, Birgit. . “ere Are No Pots in the Ślokavārika: Kumārila’s Definition of the Abhā- vapramāṇa and Paerns of Negative Cognition in Indian Philosophy.” Journal of the Orien- tal Institute (-): –. Nyaya-Tarkatirtha, Taranatha, and Amarendramohan Tarkatirtha, eds. . Nyāyadarśanam: With Vātsyāyana’s Bhāṣya, Uddyotakara’s Vārika, Vācaspati Miśra’s Tātparyaṭīkā & Viś- vanātha’s Vṛi. New Delhi: Munshiram Manoharlal. First published in the Calcua San- skrit Series, –. Rao, Srinivasa. . Perceptual Error: e Indian eories. Honolulu: University of Hawaiʻi Press. Steinkellner, Ernst, ed. . Dharmakīrti’s Pramāṇaviniścaya: Chapters and . Sanskrit Texts from the Tibetan Autonomous Region . Beijing: China Tibetology Publishing House; Vienna: Austrian Academy of Sciences Press. Tachikawa, Musashi. . e Structure of the World in Udayana’s Realism: A Study of the Lakṣaṇā- valī and the Kiraṇāvalī. Boston: D. Reidel Pub. Co. akur, Anantalal, ed. . Nyāyavārikatātparyapariśuddhi. New Delhi: Indian Council of Philo- sophical Research. Upādhyāya, Padmaprasāda, and Ḍhuṇḍhirāja Śāstrī, eds. . Nyāya Kusumāñjali of Śrī Udaya- nācārya, with Four Commentaries—the Bodhinī, Prakāśa, Prakāśikā (Jalada) and Makaranda by Varadarāja, Vardhamānopādhyāya, Megha akkura and Rucidaopādhyāya. Kashi San- skrit Series . Varanasi: Chaukhambha Sanskrit Sansthan. D N — W S Page of
Von Rospa, Alexander. . e Buddhist Doctrine of Momentariness : a Survey of the Origins and Early Phase of is Doctrine up to Vasubandhu. Alt- und Neu-Indische Studien . Stugart: Steiner. Yoshimizu, Chizuko. . “What makes all that is produced impermanent? e proof of imperma- nence and the theory of causality.” In Religion and Logic in Buddhist Philosophical Analysis. Proceedings of the Fourth International Dharmakīrti Conference, edited by Helmut Krasser, Horst Lasic, Eli Franco, and Birgit Kellner, –. Vienna.