Proving idealism: Dharmakīrti

Oxford Handbooks Online Proving Idealism: Dharmakirti a Birgit Kellner The Oxford Handbook of Indian Philosophy Edited by Jonardon Ganeri

i 1 Abstract and Keywords

The e xistence of the external world is a major conte sted issue among Buddhist and Brahmanical thinkers in the logico- epistemological period of Classical Indian philosophy (c.5th-12th century CE). Buddhist philosophers aligned with the idealist Yogacara-Vijnanavada tradition refuted external objects with different methods and arguments. Two

philosophers who contributed significantly to the discussion are Vasubandhu (probably between 350 and 420 CE) and Dharmakirti (between mid-6th and mid-7th c e ntury CE), who was one of the two main figures in the logico-epistemological or pramana school. Vasubandhu's refutation of external objects in his Vimsika Vijnaptimatratasiddhih has been interpreted as an argument from ignorance that external objects do not exist because there is no evidence for their existence . D harmakirti' s main arguments against external objects from Pramanavarttika and Pramanaviniscaya are different.

Investigating them in light of his e liminatio n of arguments from ignorance from his own and original logical theory offers new possibilities for appre ciating his stance on idealism.

Keywords: Idealism, Vijnaptimatrata, Yogacara, Vijnanavada, Vasubandhu, Dharmakirti, Buddhist epistemology and logic, pramana, argument from ignorance, anupalabdhi. Beginning perhaps with the Vimsika Vijnaptimatratasiddhih, the "Proof of Mere-Cognition in Twenty Stanzas,” by Vasubandhu (probably between 350 and 430 CE),1 Indian Buddhist thinke rs have endeavored to philosophically prove that the entire universe is nothing but consciousness or cognition. This idea is expressed in the doctrinal concept of

vijnaptimatra(ta) or "mere-cognition,”2 which had become a hallmark of the Yogacara school of thought by Vasubandhu's time. Mere-cognition is not found in early Yogacara literature —it is notably absent from the Yogacarabhumi —and only gradually makes its way to a central doctrinal concept. There is a long traditio n in the West to construe Yogacara-Vijnanavada—the form of Yogacara that is premised on mere-cognition3—as advo cating (some form of) idealism. Readings of the Vimsika and other texts through the lenses

of representationalism and phenomenalism have also been put forward, especially by North American scholars.4 Some of the most recent interpretations, though not spe cifically targe ting the Vimsika, argue that mere-cognition and associated doctrines are better viewed in terms of Husserlian phenomenology.5 In addition to debating which philosophical positions known from the hi sto ry of philosophy in the West come closest to Yogacara-Vijnanavada, scholars increasingly turn their atte nti o n to the variety in arguments that

Indian and Tibetan Buddhist philosophers offer in favor of mere-cognition and related doctrines.6 New and original arguments were put forward especially by thinke rs affiliate d with the Buddhist tradition of epistemology and logic—also called pramana school—that traces itself back to Dignaga (c.480-540 or slightly earlier) and Dharmakirti (active between the mid-6th and mid-7th c e nturie s CE).7 This school plays a prominent role in Indian philosophical debates until the twe lfth c e ntury, both through sub stantive

contributions and by engaging in the analysis of logical methods. Dharmakirti, in particular, offers an array of arguments against external objects in his two main epistemological treatises, the voluminous Pramanavarttika (PV for short), the "Commentary on the Means of Valid Cognition”, and the more concise Pramanaviniscaya (PVin for short), the "Ascertainment of the Means of Valid Cognition.”8 Even though some of D harmakirti' s arguments are presented in very cryptic form and become fleshed out as arguments against external objects only through later interpretations in his traditio n, it is evident that these are of a quite different nature, force, and flavor than

Vasubandhu's. Dan Arnold, favoring an idealist reading of the Vimsika, observes that Vasubandhu can be taken to establish a “metaphysical ide alism” that invo lve s an “a priori analysis of the adequacy of our concepts,” arguing that "realism about external objects is fundamentally incoherent.”9 Dharmaklrti on the other hand merely establishes a weaker “epistemic idealism”: “what we are imme diate ly aware of must be understood in terms of the intrinsic properties of cognition ... [which] remains compatible with an ontological commitment to really existent external objects.”10

Vasubandhu's Refutation of External Objects in the "Proof of Mere-Cognition in Twenty Stanzas”

Vasubandhu's mereological arguments in Vimsika vv. 11-15 are the main basis of Arnold's reading—"the philosophical heart of Vasubandhu's text”11—and also fe ature most pro mine ntly in other idealist inte rpre tatio ns of the text.12 Vasubandhu here argues that the objects of perception, the "sphere of color-and-form, and the like” (rupadyayatana), can neither exist as single atoms, nor as aggregates being, or consisting of, a plurality of atoms, nor as the wholes (avayavin) that the Vaisesika postulates, which are one, but

not the same as an aggregate of atoms. Vasubandhu first re fu te s atoms and (by implication) aggregates of atoms, since the single atom does not exist. If an atom were to be connected in all six directions with other atoms to form molecules, it would have to have sides or spatial parts. That, however, would contradict its axiomatic indivisibility as the most minute unit of matter. In a second step, Vasubandhu then argues that a division of the objects of our experience into atoms is necessary—the notion of undivided,

spatially extended "wholes” as unitary objects of experience leads to a range of anomalies. At first sight Vasubandhu here seems to tackle the logical possibility of the material world: its division into parts is necessary and must end with indivisible particles, but that division can at the same time not be co he re ntly maintaine d. "Metaphysical idealism,” as Arnold puts it, seems an apt characterization of this position. Many later Indian and Tibetan interpreters like wise read the text as denying the external world.

Some have however argued that even Vasubandhu pursues a logically weaker claim and merely argues that the objects of our experience do not exist outside consciousness. Claus Oetke points out that in both steps of the argument there are several "references to our everyday experience.”13 For instance, Vasubandhu argues that if a material object is one, it would not be possible that it is simultane ously seen by one person and not seen by another, for the two contradictory properties of being seen and not being seen cannot

apply to one and the same object at the same time. Hence the objects of experience must have diffe re nt sides or spatial parts, and their division is necessary.14 When refuting atoms, Vasubandhu claims that we do not see single atoms, and we also do not just see things of the size of a single atom. If the things we see are aggregates of atoms, there must be repulsion between them, otherwise all atoms could have the same spatial position, and we would not see larger things. But if atoms repel each other, then each single

atom must be in contact with several surrounding ones, which in turn entails that is has diffe re nt sides, or spatial parts. As a result it would not be indivisible, which, axiomatically, it would have to be.15 Given these and other references to experience, it might well be more plausible to read Vasubandhu as establishing merely that the things we are experiencing are not atoms, aggregates, or wholes. Such a position would not logically imply that, in Berkeley's words, to exist means to be perceived (esse est percipi). If all that is said is that objects of experience are not external to the mind, there might still be material bodies in the world, and existence would not be wholly mind-dependent.

Given that conflicting readings have been proposed of the Vimsika's position, and that it seems unlikely further textual studies will resolve the matter either way, we have re ce ntly proposed to approach the text diffe re ntly, first and foremost by atte mpting to make sense of the treatise as a whole, and not merely of a part of it, as philosophically important as that part may be.16 Idealism, now, is equivale nt to a negative thesis: it is the negation of the statement that there are uncognized objects, or

objects outside of consciousness. Strategies for proving idealism will therefore likely be mainly concerned with proving a negative statement, with proving that there is no external world independent of our cognitions. If we hypothetically assume that Vasubandhu pursued some strategy in the Vimsika to deny the existence of the external world, elements of such a strategy may well be present in other works where he definitely proved the nonexistence of something. In the ninth chapter of the Abhidharmakosabhasya (AKBh

for short) Vasubandhu sets out to refute a self (atman, pudgala ) . A close reading of that chapter17 suggests that it follows a pattern that can be dubbed an extended argumentum ad ignorantiam, an argument from ignorance: A self does not exist because there is no evidence for its existence—and if it existed, it would somehow reveal itself, make itself known. Evidence might be offered by perception (pratyaksa), inference (anumana) or scripture (agama), which are the three "means of valid cognition” (pramana) that Vasubandhu like other Yogacara authors recognizes, even though unlike Dignaga and Dharmaklrti he did not articulate a full and syste matic pramana theory. Although Vasubandhu is not explicit in this regard, in AKBh IX he nonetheless can be taken to proceed by considering for each means of valid cognition whether it co nstitute s evidence for the existence of a self. In each case, he draws a negative conclusion. The overall conclusion is then that a self does not exist.

The Vimsika, when taken as a whole, can be read as pursuing the same argument strategy for the refutation of the external world.18 The first part of the text (Vimsika vv. 1-7) on this reading argues that reason—inference—cannot establish the existence of physical objects because nothing requires us to postulate them to account for certain facts of our experience: that cognitions are restricted to certain times and places, that some cognitions arise for everyone present and are not limite d to one person, or that food or drink

experienced in waking cognitions produce e ffe cts that can be expected of them, whereas they fail to do so when experienced in dreams. While Vasubandhu's interlocutor insists that these facts require the existence of external objects, Vasubandhu explains them away by appealing to dreams, and also to experiences in the hells that Buddhist doctrine e nvisio ns as one possible realm of rebirth. Dreams are also temporally and spatially specific. We also experience e ffe cts of things we see in dreams, as in the case of

nocturnal pollution. In the hells, spirits of the dead all see rive rs of waters as filled with pus and excrement, due to the same maturatio n of their deeds, their karma, that caused them to be reborn there. The living beings reborn in hell all see hell guardians in certain places and times, and they all see them. But none of these things, the guardians and the mountains and rive rs in the hells, really exist. An interlocutor then argues that the Buddha's teaching of experienced objects—of the “sphere of color-form,

etc.”—establishes the existence of physical objects, including the guardians of hell, outside the mind. This suggests that scripture might be evidence for physical objects, and Vasubandhu next turns to dis cre diting it (Vimsika vv. 8-15). The Buddha spoke of the “spheres” not as his final position, but merely to instruct people in the nonexistence of a self. Ultimate ly he teaches, through mere-cognition, that fa cto rs such as color-form are without essence: that there re ally are no such things. One knows that the Buddha taught the sense-spheres with this inte ntio n, and that they do not exist, because the objects of experience are neither one, nor many—here we encounter the re futatio n of

physical objects that was related above (vv. 11-15). This re futatio n is then subservient to the larger goal of disc re diting scripture as evidence for external objects. But does perception not offer particularly strong evidence for them? If one knows that one just perceived something, is this not evidence that that object exists? How is the cognition that an object was perceived possible without the existence of an object? It is possible, Vasubandhu argues, because it arises in the same way as in dreams. Moreover, the knowledge that one has just perceived something arises as a mental cognition after the perception itself (and also the object) have already disappeared. Memory can also not

establish an external object because one can only remember something that one previously experienced, but Vasubandhu has already demonstrated that one can experience something even in the absence of external objects, as for example in a dream. And such a dre am-like cognition can serve as the basis of a later memory (Vimsika vv. 16-17ab).19 Arguments from ignorance are usually considered as a fallacy in modern textbooks on logic, but they appear to be reasonable where they fu nctio n as a strong abductive argument, as an infe re nce to the best explanation. In certain settings the nonexistence of something provides the best explanation for why there is no e vide nce for it. And

Vasubandhu's argume ntatio n, both in AKBh IX against the self and in the Vimsika against the external world, can be me aningfully interpreted to make this point. In any case, such arguments are being used, in scie ntific reasoning as well as in contemporary philosophy. They were also considered in Indian philosophy,20 and it is not histo ric ally implausible that Vasubandhu would have thought them to be sound.

He does however not clearly state the conclusion that external, physical objects do not exist. This can be interpreted in two diffe re nt ways. First, it is a fairly obvious conclusion to draw, and perhaps Vasubandhu felt that it need not be stated and let his audience take the final step themselves. If there is no evidence whatsoever for the existence of something, does this not e ntail that it does not exist? Second, soteriological considerations might also account for Vasubandhu's re tic e nce at stating the

natural conclusion to his argume ntatio n outright. Vasubandhu claims that mere-cognition is ultimately inconceivable in all its aspects by people like himse lf, since it is not an object of reasoning.21 Some aspects of mere-cognition can only be realized in higher me ditative states, that is, transconceptual me ditatio n or nirvikalpasamadhi . Perhaps such considerations motivated Vasubandhu to re frain from stating his conclusion in unmistakable terms, and to choose an argument that is somewhat indire ct.22 Dharmakirt.i's Refutation of External Objects in the "Commentary on the Means of Valid Cognition” and the “Ascertainment of Means of Valid Cognition” A Theoretical Hierarchy in Dharmakirti's Epistemology

Vasubandhu's Abhidharmakosabhasya and Vimsika are couched in different metaphysical frameworks: the former is written on the presumption that perception is of external objects, whereas the latter on our reading denies their existence. In the Abhidharmakosabhasya, Vasubandhu even gives a discussion of problems related to atomism whose premises are then subjected to criticism in the Vimsika.23 This shift in framework can be explained through biographical factors, assuming that something like the traditional account that Vasubandhu converted to Mahayana (espousing and defending Yogacara-Vijnanavada in particular) in the later stages of his life is historically accurate.

Dharmakirti's oeuvre presents us with a challenge of a different nature. He composed a treatise against solipsism—the Santanantarasiddhi or “Proof of Other Mental Co ntinua”—which shows that he took philosophical implicatio ns of idealism seriously and that he understood the doctrine of mere-cognition to e ntail idealist positions that incur the risk of solipsism.24 Yet, in his two main epistemological treatises, the Pramanavarttika and the Pramanaviniscaya, we find conflicting frameworks presupposed, articulate d, and

defended within one and the same text: both “externalist” and “internalist”25 theories about objects of perception are found side by side. Several areas where D harmakirti puts forward original ideas—the nominalist theory of concept-formation by exclusion (apoha), the proof of universal momentariness, a theory of properties grounded in causality—are largely unaffected by whether or not externalism or internalism are presupposed.26 His theory of cognition rests for the greater part on the unquestioned assumption that

perception is of external objects, so much that one can ide ntify externalism as D harmakirti's default view on perception. As will be shown below, Dharmakirti even supports a particular form of externalism and argues for its superiority over other forms held by his Brahmanical opponents. On a few yet nonetheless conspicuous occasions, he then points out flaws, either in the general idea that cognition is of external objects or of sp e c ific features of the very externalist theory that he himself elsewhere supports. On these occasions he promotes an inte rnalist conception of perception as being more accurate and preferable from a theoretical point of view.

In different ways, Georges Dreyfus, Sara M cClinto ck, and John Dunne have approached the theoretical hie rarchy that is at work here with the he uristic metaphor of a “sliding” or “asce nding” scale of analysis.27 This hie rarchy is info rme d by a soteriological background in the Yogacara-Vijnanavada elaboration of an ideal practitio ne r's progress on the path to liberation, to the effect that more accurate accounts of the nature and structure of consciousness correspond to the nature of mental states thought to be

experienced on more advanced stages on the path. External reality is presupposed by default, in c o nfo rmity with thinking about perception “in the world” (loke) that is also shared by Brahmanical philosophers, up to a point where closer analysis practically forces one to reject its premises. A systematic cognitive disto rtio n, grounded in ignorance (avidya) as a Buddhist root-evil, pre ve nts beings from re aliz ing and experiencing the true nondual nature of consciousness. It makes them project an internal object-form outward and thus impose duality on experience.28 The Buddha taught the “spheres” (ayatana)—external objects of perception—merely in accordance with worldly understanding, setting aside what is ultimate ly true.29

Defending Form-Possession (sakarajnanavada): The First Step to Refute External Objects

It makes little sense to speak of Dharmakirti per se as “externalist” (or some kind of realist) or “inte rnalist” (or some kind of idealist), as he shifts between these frameworks. Moreover, although externalism and internalism are part of conflicting me taphysical frame works—and recognized as being in co ntradictio n—, the sp e c ific externalist and internalist theories that D harmakirti employs also have something in common: the advocacy of cognition's “form-possession” (sakarajnanavada). Even as an externalist, Dharmakirti holds that perception contains the form (akara) or appearance (abhasa) of its object within itse lf. As a momentary mental episode—autonomous because just like Vasubandhu Dharmakirti denies the existence of a self—perception arises from an external object b e aring that object's form.30 In other words, perceptual awareness is connected to the external world through causation and resemblance, according to a model of perception that is often ascribed to the Sautra ntika school of thought, although some characte rizatio ns of Buddhist do ctrinal views indicate that this model was not exclusive to Sautrantikas. It is not unlikely that its close association with Sautra ntika emerged as a result of its prominent place in, and propagation through, Buddhist epistemology.31

In the Sautra ntika model, perception apprehe nds external objects mediated by internal forms that are comparable to sense-data, or—to state it differently—perception directly grasps an internal form, and only indire ctly the external object that caused it. In PV 3.301-319 and the parallel PVin 1 30,9-33,10 (including vv. 34-37) Dharmakirti argues in favor of form-possession first within an externalist framework and thereby establishes that the Sautrantika causation-cum- resemblance model of perceptual awareness is superior to the “fo rmle ss” (nirakara) theories of perception that his Brahmanical opponents have on offer. D harmakirti builds upon and considerably expands on the pithy re marks that Dignaga made in the latter's Pramanasamuccaya(vrtti) 1.8cd-10 (PS(V) for short).32

Dharmakirti's externalist defense of form-possession might at first sight seem to have no relationship to proofs of idealism. When placed in a broader c o nte xt, it can nevertheless be construed as a first step of a more elaborate argumentation against external objects. For Dharmakirti, form-possession is the only proper way to account for the fact that cognitions shows changing objects, for what one might call cognition's object-specificity, or the fact of “re stricte d awareness” (prativedana).33 Exte rnalists on the other hand tended to view the re stricte d and variable nature of cognition to imply, if not to prove, its dependence on an external world. Already in the beginning of the Vimsika, Vasubandhu has an externalist opponent argue that the re strictio n of cognition in place and time—and, by extension, its changing nature—can only be explained through external objects. The Mimamsaka Kumarila, a near contemporary of Dharmakirti, similarly argues that if cognition were without an (external) object (niralambana), then a varie ty of important distinctions in the realms of philosophy, ritual the o ry, and scholarly practice would collapse, including the distinction between valid and invalid cognitions that is surely at the core of any epistemological e nte rprise .34 For Kumarila, it is established for all living beings that something is apprehended having a certain “form”—as blue or yellow, long or short. But it is not evident whether that form belongs to the external object or to cognition.35 His unnamed Buddhist interlocutor argues that form belongs to cognition, while Kumarila takes the opposing viewpoint that it belongs only to the external world.

Kumarila and his Buddhist opponent—whose ide ntity is uncertain—appear to agree that, if form belongs to cognition, this alone might su ffic e to disprove external objects.36 D harmakirti, to recall, takes a more circuitous route and first argues for cognition's form-possession within Sautrantika parameters. What other options might an externalist have for explaining why cognition is of a sp e c ific object such as blue color? One might think that perception is of a particular object due to one of its causes. These include—if we think of visual perception—an external object, a sense-faculty, and according to some of the Brahmanical theories that Dharmakirti is arguing against also the contact between

the sense-faculty and the external object (indriyarthasannikarsa). This contact is, first of all, ill suited to account for what perception shows because it invo lve s the object in its e ntire ty, whereas perception only shows some of that object's qualities. The sense­faculty is also unsuitable, for several reasons. F irst, whatever factor accounts for perception's object-specificity must be such that when that factor changes, perception's object also changes. But when the sense-faculty changes by becoming sharp

or dull, this does not lead to the appearance of a diffe re nt object in perception. To conjure up an example, a person afflicted by presbyopia comes to see things diffe re ntly, but does not end up seeing diffe re nt things. Second, the visual sense-faculty is common to every instance of visual perception, regardless whether it is of blue or yellow color. If, lastly, one were to argue that some other fe ature of cognition were to account for cognition's re strictio n to an object, a feature that is caused by the external object but is not tantamount to “resemblance” in terms of form-possession, one would have to specify just what that fe ature would be. Whatever features of cognition others introduced to link cognition with its object—for

example, it being the “mere seeing of an object” (arthalocanamatra)—are reducible to cognition's form-possession, and cannot be made intelligible without it.

To appreciate this as the first step in an elaborate re futatio n of external objects, recall that Dharmakirti's externalist theory of perception has a representationalist flavor: perception is d ire ctly aware of an object-form that it contains within itself, and that bears a likeness to an external object that caused the perceptual awareness. Through his arguments for cognition's form-possession, Dharmakirti underscored this point. Having led his opponents—all direct realists—to accept that much, he is now in a position to push the point: Why then does one need external re ality in the first place? What is the evidence for it, if all we see are merely appearances in our mind?

The Refutation of External Objects

Accordingly Dharmakirti has a Yogacara-Vijnanavadin raise the critical question: Let us grant that there is restricted awareness, and that it is to be accounted for through cognition's having an object's form. But why should it be of an external object?37 A Sautrantika would of course argue that this is due to causation and resemblance. But upon closer examination the Sautrantika theory turns out to exhibit several flaws. First, perception shows a "coarse" (sthula) form; in the case of visual perception we can assume this basically to mean that perception shows its object as spatially extended. Yet the purported causes of perception are supposed to be a multitude of minute atoms. There is therefore a basic incongruence between what causes perception and what appears in it. Aggregates of atoms, however one conceives of them, might appear, but these are not

ultimate ly real and have no causal capacity.38 And “wholes" are not real, essentially because if, like the Vaisesika, one assumes them to be non-different from their parts a host of anomalies would follow, such as the movement of the whole body when the hand moves.39 In its basic outline D harmakirti' s argument from incongruence takes its clues from Dignaga's "Examination of the Object-Support,” the Alambanapariksa.40 Elsewhere, Dharmakirti presents a refined version of the Sautrantika model where many "conglomerated” (sancita) atoms are the external object of perception because the single atoms have the special quality of being able to produce a perception only when they are together, not in is o latio n. But even in this refined ve rsio n the atoms do not have a spatially e xte nde d appearance, and Dharmakirti ultimately rejects it.41

To the argument from incongruence, D harmakirti adds a further rather oblique argument that can be unpacked as follows. Assume that there are two successive perceptions showing the same object. Both have the same object-form, and the preceding perception is the cause for the subsequent one. The conditions of causation and resemblance are therefore fulfille d—but what has to be considered as the object of the subsequent perception is not something external. In other words, causation and resemblance do not restrict

perception to an external object; an imme diate ly preceding perception—in terms of Abhidharma causal theory, the "imme diate ly preceding homologous condition” (samanantarapratyaya)—might just as well be regarded as perception's object. Elsewhere in the Pramanavarttika, Dharmakirti offers a further and more fundamental critique of the notion of resemblance (sarupya) that underlies the Sautrantika model. If c o g nitio n resembled its object in all respects, it would no longer be c o g nitio n; it would be that object. If it were only partially similar to the object, everything would be aware of everything else. A c o g nitio n of a pot would be that of potsherds because they share the property of "being cognizable” and thus resemble each other partially. Even if it is argued that the c o g nitio n of a pot has the form "pot,” and not "potsherds,” then the problem remains that it would be a c o g nitio n of all pots, and not inde xically linked to the one pot in front of my eyes.42

From such arguments Dharmakirti concludes that cognition, having an object-form within itse lf, is not of anything else; it is only aware of itself. This is the first of the two main claims that are expressed with the no tio n of "self-awareness” or "reflexive awareness” (svasamvedana). At this point, a Sautrantika might insist that even if perception were directly only aware of an object-form it carries within itse lf, that form could still have been produced by an external, physical object. However, D harmakirti quickly adds that a causal explanation for perceptual content on strictly internalist grounds can also be provided: experiences leave traces in a mental stream and become actualized, or "woken up,” by c o g nitive events that occur at a later time; these traces or imprints (vasana) then give rise to specific perceptions. Perception does not causally depend on external objects.43

The argumentation so far can be taken to argue that external objects cannot be perceived (the objects of perception are not external) as a consequence of the co mbinatio n of a particular ato mistic ontology with an account that links perception with its object through causation and resemblance. The argument from incongruence basically claims that this combination produces theoretical problems that cannot be resolved, the samanantarapratyaya argument claims that causation and resemblance do not re strict perception to an external object, and additional arguments o ffe re d elsewhere in the Pramanavarttika demonstrate that "resemblance” is too vague an explanatory concept considering the purpose it is to se rve .

Dharmakirti next o ffe rs what can be regarded as a second set of arguments. These also establish that cognition cannot be of an external object, but they proceed from diffe re nt premises. They are more difficult to re c o nstru ct, because they are partly expressed in a very cryptic form, and partly also have been traditio nally understood as proving other claims apart from cognition's independence of external objects. They also seem to be original; no undisputed precedents for them were ide ntifie d so far. The two main arguments in this set de rive from basic assumptions about the very nature of c o g nitio n, about what it means for mental states to be aware. One such argument, dubbed "samvedana-inference” ("awareness-inference”) by Takashi Iwata,44 takes for its premise an explication of "awareness” (samvedana) as having the nature of "appearing-in-a-

certain-way” (tathaprathana). Because awareness has that nature, cognition cannot possibly be of any other object than itse lf, just like the awareness of a c o g nitio n itself (in virtue of its being aware of an object) is not a case where one mental state is aware of another. This is the second main claim expressed with the no tio n of "reflexive awareness,” the first being that c o g nitio n is only aware of itself. In the awareness-argument, Dharmakirti can be taken to advance an intransitive conception of awareness, co nforming to an established line of criticism directed against transitive accounts that Brahmanical thinke rs, but also Sarvastivada Buddhists, had offered. According to these accounts c o g nitio n is an action undertaken by a subject (a self or a mental state) with respect to an object, with all invo lve d factors assumed to exist in separation

prior to the occurrence of co g nitio n. In fact, the Sautrantika model of perception has been ratio naliz e d as precisely such a critique: As Dignaga and, following him, also Dharmakirti argue, c o g nitio n might appear to perform the activity of apprehending an object, but in reality there is nothing more than a c o g nitio n arising from an object bearing that object's form.45 But in the argument from awareness the additional element is involved that cognition arises from its causes with awareness for its nature. It is this awareness that is explicated in terms of an intransitive "appearing":46 When there is a cognition of an object there is just a cognitive episode arising from its causes

with awareness-qua-appearance for its nature. Although Dharmakirti is rather explicit in deriving from awareness-qua-appearance that cognition cannot be of external objects, a rationale behind this argument is offered only by later thinkers in his tradition. As Santaraksita (725-788) elaborates, cognition arises as completely separate from what is insentient.47 Whatever might be "out there" is insentient (jada). Not having cognition's luminous nature, external objects cannot manifest at all. The argument on its own does not strictly speaking rule out that insentient objects might cause cognitions, but it does entail their radical impe rceptibility—and it does so independently from any view on the atomic c o nstitutio n of the material world, and on its potential causal relation with conscious mental states.

The second main argument that similarly e ntails the radical imperceptibility of the external world is the sahopalambhaniyama-inference, the inference from the necessary joint cognition of blue and its perception to their non­difference (abheda) that Iwata has meticulously documented and studied in its manifold histo rical inte rp re tatio ns :48 blue is not different from its cognition because there can be no cognition of only one of these two without a simultane ous cognition of the form of the other. This explication of the inference is given in Pramanaviniscaya 1.54cd and the prose thereon, which is the most fre que ntly cited version of it. The premise in the inference has two components: The first is that a perception is ne cessarily cognized when an object is perceived. The second is that an object such as blue color is necessarily cognized when its perception is cognized.

A full philosophical appreciation of this intriguing argument, which does not seem to have any parallels in Western philosophy, remains a deside ratum.49 The second component premise seems less co ntrove rsial. When we think of the "cognition" of the perception of blue as an imme diate aware ness that accompanies it, taking on the form "I am aware that I now perceive blue," it may even seem almost trivial. When I am aware that I now perceive blue, I am also aware of blue. Even if the "cognition" were to consist in a later memory of an earlier perception, it seems fairly reasonable to say that when I remember having perceived blue, blue is also cognitively present in my memory. If perception has

an object­form, being aware of perception also means being aware of the object-form that it contains, since any cognition of an awareness will be a cognition of an awareness of something. The first component premise, that perception is necessarily cognized when its object is perceived, depends on reflexive awareness (svasamvedana) insofar as it includes the theorem that conscious mental states are intrinsically and imme diate ly (and intransitive ly) aware of themselves. Accordingly Dharmakirti supports this premise by arguing (a) that perception has to be establishe d—known—in order to perceive its object, and (b) that perception cannot be known in this way only after it has occurred because

that would lead to an infinite regress. The second-order cognition of perception would in virtue of claim (a) have to be cognized by a fu rthe r cognition, and so fo rth. Claim (a), however, remains question-begging. What Dharmakirti is after seems to be what contemporary phenomenology considers a pre-reflective, thin self-consciousness of experience, a way in which we are always in some way acquainted with our experience.50 In a move that finds parallels in the re futatio n of higher-order theories of consciousness in

contemporary philosophy, D harmakirti raises an infinite regress if self-aware ness is not accepted, but he does not offer an explicit rationale for why perception has to be known in order to make known its object.51 The upshot of both these arguments, the awareness-inference and the inference from necessary joint cognition, is that external objects are utte rly impe rceptible in virtue of the nature of awareness itself. If perception is thought of as evidence to establish external objects, by force of these arguments it turns out to be incapable of proving them because it cannot establish them as independent from aware ness. Whatever object perception might reveal is fully dependent on perception itself, and not an independently given external object, material in nature. Discarding Further Evidence for External Objects

Might this much be sufficient to prove the unreality of the external world? First, there might be other evidence for its existence. As for scriptural evidence—the Buddha after all taught the "spheres" as external objects of perception—we have seen above that Dharmakirti considers such teachings to have been o ffe re d merely in conformity with the common unde rstanding among ordinary people (PV 3.219). Even if scripture were to count as evidence in principle—and for D harmakirti the evidential force of scripture is limite d, to say the least—it does not offer evidence for external objects in particular, as the pertinent teachings have to be interpreted on the background of a specific i nte ntio n

that sets aside how things really are. As far as inferential evidence is concerned, one might point to facts that could not be explained without the postulatio n of external objects, thus taking these as the basis of an inference to the best explanatio n. Dharmakirti does not confront the possibility of inferential evidence in a straightfo rward manner and in one particular place, but several of his arguments can be taken to dispute that the external world is the best explanation for certain facts about cognition. The arguments examined above can be taken to claim that the external world is not the best explanation for the restricted nature of perception. Cognition's form-possession, when

accompanied by a causal account that relies only on latent mental traces and causal processes inte rnal to consciousne ss, simply o ffe rs a better explanation. External objects are also not re quire d to account for the difference between valid and invalid cognitions, as Kumarila might claim. Invalid cognitions arise from imprints left behind by disturbed cognitions, and they therefore do not enable one to attain a desired goal, which is what valid cognitions are supposed to do. Valid cognitions on the other hand arise from strong imprints with an uninterrupted connection with desired goals and are therefore re liable .52 Determining cause and e ffe ct, and drawing inferences on the basis of

causal relations, is equally possible without assuming external objects, and this is actually the method preferred by the "wise” (vidusam).53 One might formulate an inference to prove external objects along the following line s: When all other causes for perception are assembled, and perception still does not arise, this implies that an additional cause is needed—and that further cause might well be the external object. But Dharmakirti not only expresses this inference in the hypothetical. He also adds, immediately after

stating it: unless the Vijnanavadin should claim that that additional cause is a special material cause of the cognition, that is, a preceding mental episode in the same mental series.54 The non-arising of perception when a certain number of its causes are present does not conclusively establish that the missing additional cause has to be an external object: it only does so if the possibility of an inte rnal cause is willfully ignored, or set aside. Along these lines, several of Dharmakirti's arguments can be construed as discrediting evidence for external objects, in a fashion that is similar, though not entirely identical, to what we have argued for Vasubandhu. But these arguments are scattered throughout different parts of his works, and they are offered in quite different contexts; one has to piece them together. They do not come anywhe re close to a strategy that, as we have argued for the Vimsika, provides an e ntire treatise with a structure and is indicative of a plan that info rme d its composition.

Concluding Remarks: What Is at Stake in the Idealism Debate

The question thus arises: Did Dharmakirti deliberately choose not to pursue an argument from ignorance as a stand­alone strategy to re fute the external world? Even if he might not have understood Vasubandhu to provide such an argument, he could after all have combined the various scattered arguments against potential evidence for external objects to form a larger structure, but he did not do so. Was this deliberate? And do his arguments show that he maintains a weaker, "epistemic idealism”—in Arnold's terms—that only claims external objects cannot be perceived, while remaining agnostic about their existence?

There is, at least, a possibility that so me thing else might be at stake. Although Vasubandhu worked with a conception of means of valid cognition, none of his preserved works contains a detailed epistemological theory. The theory of inference (anumana), in particular, was considerably advanced by Dignaga and D harmakirti, and there is evidence to suggest that D harmakirti arrive d at certain conclusions that may well have acted as methodological constraints on his refutation of external objects. Dharmakirti is among the first Indian logicians to devote sustained attention to how one can know nonexistence, and to whether nonexistence can be proven infe re ntially. In his logical the o ry, a

separate type of evidence (hetu) is reserved for this purpose: non-appre hension (anupalabdhi). But in working out the scope of "non- appehension,” Dharmakirti offers a variety of arguments against the argumentum ad ignorantiam and effectively e liminate s it as a method to prove the nonexistence of an entire class of entities. One might wish to prove the nonexistence of "remote” (viprakrsta) objects—objects that are distant in time or place, or by their very nature—on the ground that they are not apprehended by any of the

three means of valid cognition, perception, inference, and scripture. That there is no scriptural statement proving the existence of something cannot prove its nonexistence because scripture only teaches what is relevant for a particular purpose. If something is not mentioned in scripture, this cannot conclusively establish that it does not exist. Moreover, remote objects by de finitio n lack the capacity to produce a cognition of themselves—this is what distinguishe s them from objects that are perceptible. Remote

objects are therefore not of such a kind that cognitions can be observed as e ffe cts proving their existence . Since they therefore might exist without giving rise to a cognition of themselves, their non-appre hension through a pramana cannot establish their nonexistence .55 Moreover, a general non-apprehension by all persons cannot prove anything, since it is not known to oneself, nor to anyone else. Only one's own non-appre hension of an object is e vid e nt, and when applying to remote objects, it is subject to the limitations just outlined.56 In its evidential value, non-appre hension is limited to proving that objects that would necessarily produce a perceptual awareness of themselves in a given situation where all other causes for the arising of that awareness are present can be justifiably determined as absent. It can only establish the situationally specific absence of particular objects, not the nonexistence of an e ntire class of objects.

D harmakirti's logical theory, then, does not permit a straightfo rward ontological denial on the basis of an argument from ignorance. A Dharmakirtian might wish to claim that objects that are completely imperceptible are as good as no nexiste nt, but when pushed to prove their no ne xiste nce (s)he would have to resort to other arguments than the ones I have examined here. Such arguments could, of course, be precisely Vasubandhu's mereological arguments against atomism found in Vimsika vv. 11-15. D harmakirti nowhere

seems to make use of these, and we must leave the question open what he may have thought of them. But at least one of his comme ntato rs, Manorathanandin, is well aware that there is a difference between these arguments and Dharmakirti's own. He agrees that D harmakirti's inference from necessary joint cognition merely establishes that there is no means of valid cognition (specifically: perception) proving external objects, but that D harmakirti has thereby not yet established their nonexistence. Mano rathanandin then

goes on to stress that this is entirely sufficient given what a D harmakirtian aims for: "Because [we] establish what [we] intend only as far as [saying] that cognition appears, whe re as the external object does not appear at all, we do not have any regard for negating the external object, which behaves like a[n imperceptible] demon [and] is without a means of valid cognition that proves it. But if the opponent were to strongly insist on negating the [[[external object]]], he should be made to examine the master [[[Vasubandhu's]]] negation of atoms according to whether one supposes that [the external object] has parts or is without parts."57 It is not that Vasubandhu's mereological arguments are formally

flawed, but they prove something that is no longer of value once Dharmakirti's de mo nstratio ns have been considered: they entertain the moot question as to whether an object for whose existence there is no evidence whatsoever actually exists. Whether (some or all) Dharmakirtians apart from Mano rathanandin aim to establish the nonexistence of the external world or deliberately opt for a logically weaker position remains a difficult question, considering that philosophers may, without realizing it, aim to establish claims

that their methods nonetheless are not quite suitable for. For what it is worth, Buddhist epistemologists were taken to advance something stronger than a merely epistemic idealism by their Brahmanical opponents.58 But from Mano rathanandin's vantage point, Dharmakirti's approach to refuting external objects may be taken to signal a shift to a new paradigm within which the type of questions Vasubandhu is understood to ask become simply irrelevant. The project, in other words, of engaging in the kinds of questions to which Vasubandhu provided an answer, becomes philosophically dubious.

Bibliography

Arnold, Dan. "Is Svasamvitti Transcendental? A Tentative Reconstruction Following Santaraksita." Asian Philosophy 15, no. 1 (2005): 77-111.
Arnold, Dan. "Self-Awareness (svasamvitti) and Related Doctrines of Buddhists Following Dignaga: Philosophical Characterizations of Some of the Main Issues." Journal of Indian Philosophy 38 (2010): 323-378.

Coseru, Christian: Perceiving Reality: Consciousness, Intentionality, and Cognition in Buddhist Philosophy. Oxford, etc.: Oxford University Press, 2012.
Kapstein, Matthew. "Mereological Considerations in Vasubandhu's ‘Proof of Idealism'." Idealistic Studies 18, no. 2 (1988): 32-54.
Kellner, Birgit and John Taber: "Studies in Yogacara-Vijnanavada idealism I: The interpretation of Vasubandhu's Vimsika." Asiatische Studien/Etudes Asiatiques 68/3 (2014): 709-756.

Kellner, Birgit: "Proofs of I de alism in Buddhist Epistemology: Dharmakirti's Refutation of External Objects." In Indian Epistemology and Metaphysics, ed. Joerg Tuske. London: Bloomsbury Publishing, forthcoming.

Kobayashi, Hisayasu: "On the development of the argument to prove vijnaptimatrata." In Religion and Logic in Buddhist Philosophical Analysis, ed. Helmut Krasser, Eli Franco, Horst Lasic, Birgit Kellner, pp. 299-308. Wien: Verlag der Osterreichischen Akademie der Wissenschaften, 2011.

Lusthaus, Dan: Buddhist Phenomenology. A Philosophical Investigation of Yogacara Buddhism and the Ch'eng Wei-shih lun. Oxon: RoutledgeCurzon, 2002.
Moriyama, Shinya. "Sense-Data and Akara." In Logic, Navya-Nyaya and Applications, edited by Mihir Chakraborty et al., 205-216. London: College Publications, 2008.
Ratie, Isabelle. "The Dreamer and the Yogin." Bulletin of the School of Oriental and African Studies 73, no. 3 (2010): 437-478.

Ratie, Isabelle, "On the Distinction Between Epistemic and Metaphysical Buddhist Idealisms: A Saiva Perspective." Journal of Indian Philosophy 42 (2014): 353-375.
Saccone, Margherita Serena. "Subhagupta on the Cognitive Process." Journal of Indian Philosophy 42 (2014): 377-399.
Schmithause n, Lambert. On the Problem of the External World in the Ch'eng wei shih lun. Tokyo: The International Institute for Buddhist Stu die s, 2005.
Schmithausen, Lambert. The Genesis of Yogacara-Vijnanavada: Responses and Reflections. Tokyo: International Institute for Buddhist Studies of the International College for Postgraduate Buddhist Studies, 2014.
Taber, John: "Kumarila's Buddhist." Journal of Indian Philosophy 38 (2010): 279-296.
Taber, John: "Philosophical Reflections on the sahopalambhaniyama Argument." In: Proceedings of the Fifth International Dharmakirti Conference, Heidelberg, August 26-30, 2014, ed. Patrick McAllister, Birgit Kellner, Horst Lasic, and Sara McClintock. Wien: Verlag der Osterreichischen Akademie der Wissenschaften, forthcoming.
Tola, Fernando, and Carmen Dragonetti. Being as Consciousness: Yogacara Philosophy of Buddhism. Delhi: Motilal Banarsidass, 2004.
Watson, Alex and Kei Kataoka: "Bhatta Jayanta's Refutation of the Yogacara Buddhist Doctrine of Vijnanavada: Annotated Translation and Analysis." Minami Ajia Kotengaku 5 (2010): 285-352.
Watson, Alex: "Light as an Analogy for C o g nitio n in Buddhist Idealism (Vijnanavada) ." Journal of Indian Philosophy 42 (2014): 401-421.
Author Byline
Birgit Kellner is Professor for Buddhist Studies at the Unive rsity of Heidelberg. She works p rimarily on the histo ry of Indian and Tibetan philosophy, with a focus on Buddhist epistemology and logic.
Notes:
(1) The date of Vasubandhu is controversial. A succinct summary of the problems invo lve d is given in Florin Deleanu, The Chapter on the Mundane Path (Laukikamarga) in the Sravakabhumi: A Trilingual Edition (Sanskrit, Tibet, Chinese), Annotated Translation, and Introductory Study, vol. 2 (Tokyo: The International Institute for Buddhist Studies, 2006), 186-194. Deleanu cautiously proposes to place Vasubandhu's period of activity between 350 and 430.
(2) Translatio ns of vijnapti vary c o nside rab ly, as the term may refer to a mental act b ringing something to awareness, to that which is brought to awareness (a mental representation, manifestation, or image), or to the re sulting state. Texts make full use of this s e ma ntic range, and therefore no simple English translation of the term will match all instances of its use.
(3) "Yogacara-Vijnanavada" is used here as a convenient shorthand for the mere-cognition-based form of Yogacara thought. This term is a neologism; as far as I know, it is not used in Buddhist tra ditio n.
(4) Non-idealist readings are found, among others, in Alex Wayman, "Yogacara and the Buddhist Logicians," Journal of the International Association of Buddhist Studies 2 (1979): 65-78; Bruce Cameron Hall, "The Meaning of Vijnapti in Vasubandhu's Concept of Mind," Journal of the International Association of Buddhist Studies 9, no. 1 (1986): 7-23; Richard P. Hayes, Dignaga on the Interpretation of Signs (Dordrecht: Kluwer Academic Publishers, 1988), chap. 5; Claus Oetke, "Doctrine and Argument in Vijnanavada-Buddhism," Wiener Zeitschrift fur die Kunde Sudasiens 36 (1992): 217­225; Richard King, "Vijnaptimatrata and the Abhidharma Context of Early Yogacara," Asian Philosophy 8 (1998): 5-13; Thomas Kochumuttom, A Buddhist Doctrine of Experience: A New Translation and Interpretation of the Works of Vasubandhu, the Yogacarin (Delhi: Motilal Banarsidass, 2002).
(5) See especially Dan Lusthaus, Buddhist Phenomenology: A Philosophical Investigation of Yogacara Buddhism and the Ch'eng Wei-shih lun (London: RoutledgeCurzon, 2002). Christian Coseru o ffe rs a strong phenomenological reading of Buddhist epistemology in general, based on the Tattvasangraha of Santaraksita (725-788) and Kamalasila's (740-795) commentary on it, the Tattvasangrahapanjika. See his Perceiving Reality: Consciousness, Intentionality, and Cognition in Buddhist Philosophy (New York: Oxford University Press, 2012). Coseru does not discuss proofs of mere-cognition.
(6) Georges B. J. Dreyfus, Recognizing Reality: Dharmakirti's Philosophy and Its Tibetan Interpretations (New York: State University of New York Press, 1997), especially chap. 4 and 25-27; Sara M cClinto ck, “The Role of the ‘Given' in the Classification of Santaraksita and Kamalasila as Svatantrika-M adhyamikas, " in The Svatantrika-Prasangika Distinction: What Difference Does a Difference Make?, ed. Sara McClintock and Georges B. J. Dreyfus, 125-171 (Boston: Wisdom Publications, 2003); John D. Dunne, Foundations of Dharmakirti's Philosophy (Boston: Wisdom Publications, 2004), especially chap. 2; Hisayasu Kobayashi, “On the Development of the Argument to Prove Vijnaptimatrata," in Religion and Logic in Buddhist Philosophical Analysis, ed. Helmut Krasser, Eli Franco, Horst Lasic, and Birgit Kellner, 299-308 (Wien: Verlag der Osterreichischen Akademie der Wissenschaften, 2011); Dan Arnold, "Buddhist Idealism, Epistemic and Otherwise: Thoughts on the Alte rnating Perspectives of Dharmakirti," Sophia 47 (2008): 3-28.
(7) Erich Frauwallner argued for a date of Dharmakirti of c.600-660 ce; see his "Landmarks in the History of Indian Logic," Wiener Zeitschrift fur die Kunde Sud- und Ostasiens 5 (1961): 137-139. Helmut Krasser recently proposed an earlier date around the mid-6th ce ntury; see his "B havive ka, Dharmakirti and Kumarila," in Devadattiyam: Johannes Bronkhorst Felicitation Volume, ed. FranQois Voegeli et al., 535-594 (Bern: Peter Lang, 2012). This would also imply an earlier date for Dignaga, from whom Dharmakirti is separated by one other logician, Isvarasena. Piotr Balcerowicz, "On the Relative Chronology of D harmakirti and Samantabhadra, " Journal of Indian Philosophy (2014), reaches the same conclusion from Jaina sources. As the debate generated by the proposal is ongoing, a softer dating is o ffe re d here that places D harmakirti within a larger range.
(8) References to the text of PV in the following are based on Hiromasa Tosaki's edition of the chapter on perception (pratyaksa), which is the third chapter. See his Bukkyoninshikiron no kenkyu (Tokyo: Daitoshuppansha, 1979 [vol. 1, stanzas 1-319] and 1985 [vol. 2, stanzas 320-539]). The text is still in need of a truly critical edition that takes into account the manuscript material that has become accessible since Tosaki published his edition, and more recent insights into the text's transmissio n in India and Tibet. For methodological observations and a discussion of selected stanzas see B irg it Kellner, "Towards a Critical Edition of Dharmakirti's Pramanavarttka," Wiener Zeitschrift fur die Kunde Sudasiens 52-53 (2009-2010): 161-211. References to the text of PVin—where the chapters relevant for our purposes are the first (on perception) and the second (on infe re nc e )—are based on Ernst Steinkellner's recent edition of the Sanskrit text, see his Dharmakirti's Pramanaviniscaya. Chapters 1 and 2 (Beijing/Vienna: China Tibetology Publishing House/Austrian Academy of Sciences Press, 2007).
(9) Arnold, "Buddhist Idealism," 16-17.
(10) Arnold, "Buddhist Idealism," 15.
(11) Arnold, "Buddhist Idealism," 16.
(12) References to the Vimsika and Vimsikavrtti are based on the widely accessible editio princeps: Sylvain Levi, Vijnaptimatratasiddhi: deux traites de Vasubandhu: Vimsatika et Trimsika (Paris: Librairie Ancienne Honore Champion, 1925). The more recent edition by Piotr Balcerowicz and Monika Nowakowska is based on a reconsideration of the manuscript material, see their "Wasubandhu: 'Dowod na wylqczne istnienie tresci swiadomosci w dwudziestu strofach' (Vasubandhu: Vimsatika—Vijnapti-matrata-siddhi)," Studia Indologiczne 6 (1999): 5-44. This otherwise difficult to access Polish publication is now available online at http://www.orient.uw.edu.pl/buddologia/studiaindologiczne/archive.htm (last accessed March 5, 2015).
(13) Oetke, "Doctrine and Argume nt," 218-219.
(14) Levi, Vijnaptimatratasiddhih, p. 8, lines 15-16, autocommentary on Vimsika 15.
(15) Levi, Vijnaptimatratasiddhih, p. 7, lines 3-11, autocommentary on Vimsika 12.
(16) B irg it Kellner and John Taber, "Studies in Yogacara-Vijnanavada Idealism I: The Interpretation of Vasubandhu's Vimsika," Asiatische Studien/Etudes Asiatiques 68, no. 3 (2014): 709-756.
(17) For details, see Kellner and Taber, "Studies,” 719-725.
(18) Kellner and Taber, "Studies,” 734-749.
(19) The remainder of the text deals with a final series of five objections (Kellner and Taber, "Studies,” 745-747).
(20) Kellner and Taber, "Studies,” 727-734.
(21) Levi, Vijnaptimatratasiddhih, p. 11, lines 1-2, autocommentary on Vimsika 22.
(22) A diffe re nt perspective on how to make inte lligible that a philosopher "aims at” proving something without stating it explicitly (or even being able to state it explicitly) is offered in Oetke, "Doctrine and Argument,” 223-225.
(23) Prahlad Pradhan (revised by Aruna Haldar), Abhidharmakosabhasyam of Vasubandhu (Patna: K. P. Jayaswal Research Institute, 1975), 32, line 12-33, line 7.
(24) The Santanantarasiddhi is in full only preserved in Classical Tibetan translation. An English translatio n is given in Hide no ri Kitagawa, "A Refutation of Solipsism (Annotated Translatio n of the Santanantarasiddhi),” in Indo Koten Ronrigaku no Kenkyu—Jinna no Taikei—(A Study of Classical Indian Logic—A System of Dignaga) (Tokyo: Suzuki Gakujutsu Zaidan, 1965), 405-579. An incomplete Sanskrit manuscript of a commentary is studied in Junjie Chu: "Sanskrit Fragments of the Santanantarasiddhi,” in Religion and Logic in Buddhist Philosophical Analysis, ed. Helmut Krasser et al., 33-42 (Wien: Verlag der Osterreichischen Akademie der Wissenschaften, 2011).
(25) "Externalism” here re fe rs to theories that there is an external world that can be perceived, whereas "inte rnalism” means theories that there is only mind or consciousness, and that experience is to be accounted for only in reliance on a me ntal re alm.
(26) See also Tom J. F. Tille ma ns "Dharmakirti,” Stanford Encyclopedia of Philosophy (Spring 2014 edition), ed. Edward N. Zalta, Available o nline at http://plato.stanford.edu/archives/spr2014/entries/dharmakiirti/.
(27) For D harmakirti, see Dreyfus, Recognizing Reality, chap. 4 (based on Tibetan interpretations); Dunne, Foundations, 53-79 (based on early Indian commentaries); M cClinto ck, "Role.”
(28) PV 3.212, 214, 331, and 431.
(29) PV 3.219.
(30) PV 3.247 = PVin 1.20.
(31) Birgit Kellner, "Cha nging Frames in Buddhist Thought: The Concept of Akara in Abhidharma and in Buddhist Epistemological Analysis,” in "Akara in Buddhist Philosophical and Soteriological Analysis,” ed. Birgit Kellner and Sara M cClinto ck, special issue, Journal of Indian Philosophy 42, no. 2-3 (2014): 282.
(32) The PS(V) are only extant in full in two dive rge nt Tibetan translations. A re co nstru c tio n of the p e rtine nt first chapter on perception, based to a large extent on material from Jinendrabuddhi's eighth-century co mme ntary, the Pramanasamuccayatika, was undertaken by Ernst Steinkellner. This reconstruction is in its latest revision of 2014 available at http://www.ikga.oeaw.ac.at/mediawiki/images/f/f3/Dignaga_PS_1_revision.pdf (last accessed March 8, 2015). Masaaki Hattori's English tra nslatio n of the first chapter remains an invaluable guide; see his Dignaga, On Perception, being the Pratyaksapariccheda of Dignaga's Pramanasamuccaya from the Sanskrit Fragments and the Tibetan Versions (Cambridge, MA: Harvard Unive rsity Press, 1968). A close reading of PS(V) 1.8-12 is given in Birgit Kellner, "Self­Awareness (svasamvedana) in Dignaga's Pramanasamuccaya and -vrtti: A Close Reading,” Journal of Indian Philosophy 38 (2010): 203-231.
(33) This term is used in PV 3.320.
(34) SV niralambanavada 1-3 (John Taber, "Kumarila's Buddhist,” Journal of Indian Philosophy 38 [2010]: 279).
(35) SV sunyavada 5-6 (Taber, "Kumarila's Buddhist,” 283).
(36) SV sunyavada 8-9 (Taber, "Kumarila's Buddhist,” 283).
(37) The following argume ntatio n extends from PV 3.321 to 337. The corresponding section in PVin 1 33,11-36,9 has been rearranged and was also shortened in some places, but enhanced in others.
(38) Note that Dharmaklrti refers to a "deviation” of resemblance in PV 3.322ab (not mentioned in the corresponding section of the Pramanaviniscaya), which the commentator Manorathanandin interprets as an additional argument, basically an appeal to perceptual illusion: there are cases where an object appears that does not resemble anything internal. See Birgit Kellner, "Proofs of Idealism in Buddhist Epistemology: Dharmaklrti's Refutation of External Objects,” in Indian Epistemology and Metaphysics, ed. Joerg Tuske (London: Bloomsbury Publishing, forthcoming).
(39) PVin 1 34,10-35,5.
(40) The Alambanapariksa with its autocommentary is in full only extant in Tibetan translation, although several stanzas have been preserved in Sanskrit through qu o tatio ns in other works. The Tibetan text is edited, together with an annotated German translation, in Erich Frauwallner, "Dignagas Alambanapariksa. Text, Ubersetzung und Erlauterungen,” Wiener Zeitschrift fur die Kunde des Morgenlandes 37, no. 1-2 (1930): 174-194; reprinted in his Kleine Schriften, ed. Gerhard Oberhammer and Ernst Steinkellner, 340-360 (Wiesbade n: Franz Steiner Verlag, 1982).
(41) PV 3.194-224, English translatio n in Dunne, Foundations, 396-411.
(42) PV 3.434, read together with Devendrabuddhi's and Manorathanandin's commentaries. For details and detailed references, see Kellner, "Proofs of Idealism.”
(43) PV 3.336.
(44) Takashi Iwata, Sahopalambhaniyama. Struktur und Entwicklung des Schlusses von der Tatsache, daft Erkenntnis und Gegenstand ausschlieBlich zusammen wahrgenommen werden, auf deren Nichtverschiedenheit, vol. 1 (Stuttgart: Franz Steiner Verlag, 1991), 9-15.
(45) PS(V) 1.8cd, PV 3.308 (parallel in PVin 1 37) and 309.
(46) See Kellner, "Proofs of I de alism, ” for further explication of intransitive vs. transitive notions of awareness.
(47) Embar Krishnamacharya, ed., Tattvasangraha of Santaraksita (Baroda: Central Library, 1926), vv. 2000-2001.
Because Krishnamacharya reserved the verse number 526 for a verse he considered lost in transmissio n (which was not borne out by subsequent rese arch) these are in fact verses 1999-2000.
(48) Iwata, Sahopalambhaniyama.
(49) For pertinent remarks see Taber, "Kumarila's Buddhist,” 292-294, and also his forthcoming paper "Philosophical Reflections on the Sahopalambhaniyama Argument,” in Proceedings of the Fifth International Dharmakirti Conference, Heidelberg, August 26-30, 2014, ed. Patrick McAllister, Birgit Kellner, Horst Lasic, and Sara McClintock (Wien: Verlag der Osterreichischen Akademie der Wissenschaften, forthcoming).
(50) Shaun Gallagher and Dan Zahavi, The Phenomenological Mind, 2nd. ed. (London: Routledge, 2012), 52.
(51) This point has been made in Birgit Kellner, "Self-Awareness (svasamvedana) and Infinite Regresses: A Comparison of Arguments by Dignaga and Dharmaklrti,” Journal of Indian Philosophy 39 (2011): 419-423. It is possible that in PV 3.336, which the commentator Manorathanandin reads as a statement of the sahopalambhaniyama-inference, a weaker form of the argument is presented that is not premised on the theorem of reflexive awareness (Kellner, "Proofs of Idealism”).
(52) PVin 1 43,4-44,6. See Helmut Krasser, "Are Buddhist Pramanavadins Non-Buddhistic? Dignaga and Dharmaklrti on the Impact of Logic and Epistemology on Emancipation,” Horin 11 (2004): 143-144.
(53) PV 3.392-397.
(54) PVin 1 58d, elaborated in PVin 1 43,10-12; see also PV 3.390d-91ab (Krasser, "Are Buddhist Pramanavadins?” 142­143). The inference is construed as a Sautrantika's response to Yogacara criticism by some traditio nal interpreters, while others understand it to express the Sautrantika's view that the external object is only i nfe rre d, and not perceived. See Taiken Kyuma, "On the (Im)perceptibility of External Objects in Dharmaklrti's Epistemology,” in Religion and Logic, ed. Krasser et al., pp. 309-318; p. 314, n. 28.
(55) Pramanavarttikasvavrtti—Dharmakirti's autocommentary on the first chapter of the PV—ad PV 1.199 = 201 (Raniero Gnoli, ed., The Pramanavarttikam of Dharmakirti: The First Chapter with the Autocommentary. Text and Critical Notes [Roma: Istituto Italiano per il Medio ed Estremo Orie nte , 1960]); PVin 2, p. 62.
(56) PVin 2 64,12-14.
(57) Rahula Sankrityayana, ed., Dharmakirti's Pramanavarttika with a Commentary by Manorathanandin. Published as an appendix to Journal of the Bihar and Orissa Research Society 24-26 (1938-1940), 220, lines 17-20. This passage was also discussed by Isabelle Ratie, in "On the Distinction Between Epistemic and Metaphysical Buddhist Idealisms: A Saiva Perspective," Journal of Indian Philosophy 42 (2014): 358, in critical engagement with the interpretation offered in Arnold, "Buddhist Idealism,” 16. I am fo llo wing Ratie's reading of the passage, but I regard some of the possible
inte rpre tatio ns that she discusses as less compatible with D harmakirti's logical theory (see Kellner, "Proofs of Idealism," for further details).
(58) In this context, the refutation of Yogacara-Vijnanavada by the ninth-century Naiyayika Bhatta Jayanta—who strongly relied on Kumarila—has been studied in greatest detail; see Alex Watson and Kei Kataoka, "Bhatta Jayanta's Refutation of the Yogacara Buddhist Doctrine of Vijnanavada: Annotated Translatio n and Analysis, " Minami Ajia Kotengaku 5 (2010): 285-352, and Alex Watson, "Light as an Analogy for Cognition in Buddhist Idealism (Vijnanavada)", Journal of Indian Philosophy 42 (2014): 401-421.
Birgit Kellner
Chair of Buddhist Studies, Cluster of Excellence "Asia and Europe in a Global Context - The Dynamics of Transculturality", University of Heidelberg