The ORE of Religion will be available for subscription in late September. Speak to your Oxford representative or contact us to find out more.

Show Summary Details

Page of

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, RELIGION ( (c) Oxford University Press USA, 2016. All Rights Reserved. Personal use only; commercial use is strictly prohibited. Please see applicable Privacy Policy and Legal Notice (for details see Privacy Policy).

date: 23 August 2017

The Philosophical Works and Influence of Dignāga and Dharmakīrti

Summary and Keywords

The Indian Buddhist philosophers Dignāga (c. 480–540 ce) and Dharmakīrti (c. 600–660 ce) decisively influenced the course not only of Buddhist philosophy, but of Indian philosophy more generally. Having inherited an earlier philosophical tradition (the one advanced in the Buddhist Abhidharma literature) that had been largely intramural in character, Dignāga and Dharmakīrti fundamentally transformed Buddhist philosophy by advancing basically similar commitments with arguments meant to be persuasive across party lines. In doing this, they influentially theorized a family of concepts largely shared by all Indian philosophers writing in Sanskrit—a family centering on the concept of pramāṇa, which denotes a reliable way of knowing or epistemic “criterion” (as one might translate the word)—in ways that facilitated an unprecedented extent of debate among Indian philosophers of all sorts. The resultant growth in the sophistication of philosophical traditions is one of the most salient features of the mature period of classical Indian philosophy. Though there are significant differences between them, Dignāga and Dharmakīrti commonly argued in terms of a broadly empiricist sort of epistemology; this was advantageous insofar as that involves premises to which they might readily win assent, while nonetheless being conducive to the philosophical idealism they both finally upheld. Committed as they were to the basically empiricist notion that only perceptibles are finally real, both thinkers affirmed versions of the innovative sort of nominalism first introduced by Dignāga (and significantly revised by Dharmakīrti): the elusive apoha (“exclusion”) theory of meaning, which represents one of the Buddhist tradition’s signal contributions to the history of Indian philosophy. While some of Dignāga’s works were translated into Chinese and thus became influential in East Asia, none of Dharmakīrti’s was; in both India and Tibet, however, Dharmakīrti effectively eclipsed his predecessor. For generations of subsequent Indian philosophers, Dharmakīrti practically epitomized “the Buddhist position” in matters philosophical, and his works figure to this day as central to most Tibetan monastic curricula.

Keywords: epistemology, empiricism, logic, idealism, Sautrāntika, Yogācāra, nominalism, pramāṇa, apoha, svasaṃvitti

From Abhidharma to Logic and Epistemology

Following some later Indian commentators, doxographical traditions that were central to the structuring of Tibetan monastic curricula represented Dignāga and Dharmakīrti as commonly exemplifying the “Sautrāntika” school of thought that originates in the Abhidharma period; both were also said, however, to embrace the idealist Yogācāra school, which represents their definitive views. Notwithstanding the possibly misleading character of such doxographical terms, this pair of characterizations succinctly captures the decisive philosophical change Dignāga and Dharmakīrti ushered in, as well as the kinds of arguments they typically made for idealism. Tibetan doxographers helpfully distinguished, in regard to the first of these doxographical terms, between two subtypes of Sautrāntikas—those who follow scripture (āgama), and those who follow reasoning (yukti)—and took Dignāga and Dharmakīrti to represent the latter; the Ābhidharmika Vasubandhu (fl. c. 360 ce) was in these terms chief among exemplars of Sautrāntikas who follow scripture. Vasubandhu’s Abhidharmakośabhāṣya, an influential “summa” of the Abhidharma traditions of thought preserved in Sanskrit, typifies the earlier tradition of Buddhist philosophy that Dignāga and Dharmakīrti so influentially transformed. While texts such as Vasubandhu’s can reasonably be understood as crystallizing early Buddhist attempts at formulating a basic ontology, the Abhidharma literature was largely driven by basically exegetical considerations. Thus, Abhidharma’s project of enumerating and characterizing the “ultimately existent” (paramārthasat) constituents of reality—called dharmas in this literature—had its impetus in the challenging hermeneutical task of systematizing the many lists of categories that proliferated in the Sūtra literature traditionally attributed to the Buddha. Much of the discussion in this literature accordingly turns on questions of scriptural interpretation and is replete with arguments to the effect that “if X is not the case, the Buddha would not (as he did) have said such-and-such.” The resultant systematization of categories aimed to provide an ultimately true description of what really exists, and the Abhidharma literature was thus understood as providing a definitive account of how Buddhist categories could be used to elaborate an exhaustively impersonal account of reality—an account, that is, consistent with the cardinal Buddhist doctrine that there are no real selves (anātmavāda). While this surely amounts to the elaboration of a complete ontology, the exegetical impetus of the project made it a basically intramural one; few who did not already take it as axiomatic that adequacy to Buddhist scriptures is a relevant constraint were apt to be persuaded by many of the arguments of this literature. This is the sense, then, in which thinkers like Vasubandhu were aptly characterized as “Sautrāntikas who follow scripture.”1

That Dignāga and Dharmakīrti should also have been called “Sautrāntikas” makes sense given the basic commitments and orientation they share with the mainstream of the Abhidharma literature. Like thinkers in that tradition, Dignāga and Dharmakīrti, too, exhibited confidence in the possibility of specifying and characterizing what kinds of things count as ultimately existent, and in the possibility of showing how all our epistemic practices could be re-described in terms of the impersonal, momentary events that alone count as such. (This is in contrast especially to the Madhyamaka school of thought that begins with the philosopher Nāgārjuna, who argued that it cannot coherently be thought that our epistemic practices ever reach anything ultimately real. The guiding conviction of proponents of Madhyamaka was thus that the kinds of explanatory categories posited in the Abhidharma literature invariably turn out themselves to admit of the same kind of analysis that shows selves to be unreal.) Despite, however, the orientation they thus shared with thinkers like Vasubandhu, Dignāga and Dharmakīrti were characterized by Tibetan doxographers as “Sautrāntikas who follow reasoning.” This aptly reflects the fundamentally different way in which they went about arguing for such views; for in contrast to the largely intramural arguments of the Abhidharma literature, the stock-in-trade of Dignāga and Dharmakīrti was epistemological arguments informed by systematic theorization of the logical form of valid inferences—arguments that could at least in principle be persuasive across party lines.

Their thoroughgoing emphasis on logical consistency led, moreover, to a radicalization of the kind of ontology characteristic of Abhidharma. Consistent with their radically nominalist bent (which itself follows from a deep suspicion of the kind of conceptual thought that necessarily involves abstractions), Dignāga and Dharmakīrti recognized that the Ābhidharmika tradition’s enumeration of dharmas amounts to the enumeration of kinds of things—the enumeration, that is, of types, of which there could be innumerable tokens. (According to Vasubandhu’s Abhidharmakośabhāṣyam, for example, there are seventy-five dharmas; that does not mean, of course, that he argued for a universe containing exactly seventy-five existents, but rather that there occur seventy-five distinct kinds of events.) Against this, Dignāga and Dharmakīrti held that ontological bedrock could really consist only in the unique particulars encountered in perception. In place of the Abhidharma literature’s complex lists of ontologically basic kinds of things, then, Dignāga and Dharmakīrti suggest a much more austere ontology according to which only perceptible particulars finally count as “ultimately existent” (paramārthasat). The austere character of the resultant picture, moreover, comes fully into view only when it is appreciated that what really count as perceptible particulars must, in light of their uncompromising emphasis on the nonconceptual character of perception, be other than we typically suppose; for given Buddhists’ reductionist view that temporally enduring wholes are not ultimately real, the kinds of ordinary objects conventionally understood as given to perception represent, in fact, the constructive work of conceptual thought. When this is taken into account, it seems Dignāga and Dharmakīrti must mean that only momentary mental events finally count as “perceptible”; it is insofar as something like this is indeed their considered view that Dignāga and Dharmakīrti were reasonably taken by the doxographical traditions as finally upholding the Yogācāra school of Buddhist idealism. Insofar, however, as they were chiefly concerned to advance maximally persuasive arguments, both thinkers characteristically exploited more intuitively plausible ideas; their route to idealism, then, goes through empiricism, in the sense that their arguments typically worked by showing what follows simply from the analysis of perception’s role among our epistemic practices. Their carefully formulated arguments thus remain largely epistemological in key, and they seldom explicitly declare for any particular metaphysical or ontological conclusions (they did not, for example, explicitly affirm many of the doctrinal categories associated with Yogācāra)—an approach that sometimes makes it hard to discern what they finally believed, and that warrants the many modern characterizations of both thinkers as chiefly exemplifying a methodologically defined school of “Buddhist Logic” or “Buddhist Epistemology.”2

Their epistemological predilections found expression in terms informed by sophisticated attention to the form of valid inferences. In this regard, Dignāga’s brief Hetucakraḍamaru (“Drum of the Cycle of Reasons”) formalized valid argument forms, concisely presenting what Dignāga took to be all possible relations between the terms of any formally stated inference. Abstracting from the content of any particular argument, this text can clearly be characterized as elaborating a basic table of logic. In particular, Dignāga considered all of the various ways in which the three terms of an inference—a reason, the locus in which that is instantiated, and the conclusion warranted thereby—could be related, thus providing a content-neutral way to characterize not only the form of valid inferences, but also the various ways in which these can be fallacious. As with Indian logic more generally, Dignāga’s remains an account of inductively valid inferences; the role played by reference to examples therein is among the considerations that seemingly preclude the idea of deductive validity. (On a canonical illustration, reference to an example figures thus: “There is fire on the mountain, because there is smoke, as we know from seeing these things together in a kitchen.”) Chief among Dharmakīrti’s revisions of his predecessor’s project, however, was his attempt to theorize inferences with something more like deductive certainty, though there is considerable dispute about whether that is an apt way to characterize either his aim or the result.3

Dignāga’s sensitivity to such considerations shows up in his magnum opus, the Pramāṇasamuccaya (“Compendium of Epistemic Criteria”), which refined epistemological terms of art, attested since the early literature of the Brahmanical Nyāya school of thought, in ways conducive to characteristically Buddhist conclusions. Chiefly, while many Brahmanical schools of thought affirmed that language or the testimony of tradition ought to be reckoned among reliable epistemic criteria (as, that is, pramāṇas), Dignāga argued that only perception (pratyakṣa) and inference (anumāna) have this status; all other ways of knowing, he argued, are reducible to one of these. Of particular importance here was his emphasis on the essentially nonconceptual character of perception. While this is an intuitively plausible idea—surely the difference between perceiving a tree and (say) imagining one is that in the former case alone, one comes up against something in the world—the point becomes more radical in light of the unusually thoroughgoing reductionism characteristic of Buddhism. Given, then, the Buddhist premise that temporally enduring wholes are essentially conceptual fictions, the ordinary objects typically taken as disclosed in perception cannot finally be what counts as “perceived”; rather, it must be only momentary sense data that count as really perceptible. Indeed, Dignāga gives reason to think he finally intends the view that we are immediately, nonconceptually acquainted only with the fleeting occurrence of our own mental events. This is why his epistemology ultimately recommends the kind of idealism characteristic of Yogācāra. (Dignāga more concisely argued for idealism in his other principal work, the Ālambanaparīkṣā, or “Critical Investigation of Percepts.”) Such a view readily makes sense as supporting fundamental Buddhist commitments; above all, this recommends the conclusion that only fleeting sensations are finally real—the thought that such sensations must be the states of a self (ātma) stands revealed as an inferential belief that is finally unwarranted.4

Dharmakīrti, who is traditionally represented as Dignāga’s grand-disciple, framed his most extensive work—the Pramāṇavārttika (“Critical Commentary on Epistemic Criteria”)—as a commentary on Dignāga’s magnum opus. (Dharmakīrti’s principal works also include the Pramāṇaviniścaya, or “Ascertainment of Epistemic Criteria,” and the Nyāyabindu, “An Epitome of Philosophy.”) Benefiting from intervening Brahmanical critiques of Dignāga, Dharmakīrti greatly elaborated and revised Dignāga’s thought, advancing what many Indian philosophers (Buddhist and non-Buddhist alike) would take to be the definitive arguments for characteristically Buddhist positions; indeed, Dharmakīrti’s influence in India effectively eclipsed Dignāga’s (although only works by the latter were ever translated into Chinese). Chief among Dharmakīrti’s innovations was a strong emphasis on causal efficacy as the criterion of the real—an emphasis expressed in an oft-quoted passage that frames the point in terms of the Buddhist idea of two levels of truth: “Whatever has the capacity for causal efficacy is ultimately existent (paramārthasat); everything else is conventionally existent (saṃvṛtisat).” These two kinds of things respectively consist, he adds, in unique particulars (svalakṣaṇa) and abstractions (sāmānyalakṣaṇa); these are the objects of (respectively) perception and inference.5 Once again, it is an intuitively plausible idea that perception is to be distinguished by its thus being causally describable; surely a salient difference between perceiving a tree and imagining one is that in the former case alone one has a cognition that is actually caused by what it is about. Yet again, though, the point becomes more radical in light of the uncompromising reductionism of Buddhist philosophers; for given the Buddhist premise that only momentary things are ultimately real—and given, moreover, that only ultimately real things are capable of causal efficacy—we are again driven toward the conclusion that perception must finally be understood under something other than a conventional description. This dialectic typifies the generally epistemic arguments that Dharmakīrti, in particular, clearly advanced in support of idealist conclusions: starting from the intuitively plausible view that perception, uniquely among ways of knowing, can be described in causal terms, Dharmakīrti thus argued that what is caused by our perceptual encounters with the world must be mental “images” (ākāras); only these, he could then argue, are the direct objects of awareness, and it can only be inferentially that we suppose these to have been caused by the elements of an external world. To assent to this much is already tantamount to conceding that Dharmakīrti’s idealism wins the day.6

Consistent with the idealism advanced by both thinkers, Dharmakīrti influentially elaborated Dignāga’s thought that we are perceptually (which is to say immediately, nonconceptually) aware, finally, only of the occurrence of our own mental events. In this regard, Dharmakīrti’s arguments for the perceptual character of “reflexive awareness” (svasaṃvitti)—the perceptual character, that is, of any cognition’s awareness of itself as given along with whatever is represented therein—came to be taken as one of the tradition’s definitive arguments for the idealist view that only cognition itself is indubitably known. Dharmakīrti argued, in particular, that anything at all that can be an object of awareness can be known only along with the cognition to which it is present; cognition itself must, to that extent, be understood as explanatorily basic, and it is only inferentially that one can suppose that the content of awareness represents anything external thereto. Here, too, we can see that Dharmakīrti chiefly argues epistemologically for a view that recommends the idealist conclusion that only mental items finally exist. Insofar, however, as his arguments remain basically epistemological, Dharmakīrti is often content to rest with the conclusion that it must be allowed that reference to an external world is just optional. The stance Dharmakīrti typically adopts is reflected in another of the later doxographical tradition’s characterizations of these thinkers: “proponents of the doctrine that external objects can only be inferred” (bāhyārthānumeyavāda). This is eminently consistent with the basically empiricist orientation that Dharmakīrti epitomizes; classical empiricists like John Locke similarly argued, based on an intuitively plausible causal theory of perception, that we are immediately acquainted only with mental representations. (It was left to Bishop Berkeley to argue that if one holds such a view, one is effectively committed to idealism.) There are, however, moments in the works of both Dignāga and Dharmakīrti where they offer something more like metaphysical arguments to the effect that only mental events can exist; more on their stronger arguments below.7

The difference that Dharmakīrti’s emphasis on causal efficacy makes shows up in his attempt to address perceived deficiencies in Dignāga’s clearly inductive account of inferences.8 This emphasis shows up, as well, in Dharmakīrti’s very different (albeit complementary) elaboration of the “exclusion” (apoha) theory of meaning first introduced by Dignāga. Though it has been common for scholars both traditional and modern to associate Dignāga and Dharmakīrti as commonly exemplifying a unitary school of thought, it is perhaps especially with respect to the apoha doctrine that some important differences between them can be readily appreciated. The apoha doctrine is typically represented as a position in first-millennium Indian debates regarding the ontological status of universals (paradigmatically, linguistic referents), which Buddhists characteristically held to be ultimately unreal. (In contrast, many Brahmanical schools of thought—chiefly the Mīmāṃsā school, whose constitutive concern was with the interpretation of a Vedic corpus taken to represent the most significant pramāṇa—strongly affirmed the essential reality of linguistic referents. Indeed, the 5th-century grammarian Bhartṛhari held that linguistic items are finally the most real of all existents.) While it is apt to represent the apoha doctrine as thus advancing a kind of nominalism regarding universals, it should be understood that the scope of this elusive doctrine exceeds that; particularly as developed by Dharmakīrti, the doctrine represents a way to explain, in the absence of real universals, how all conceptual thought is constructed, and why, despite its ultimate unreality, such conceptual content can nevertheless facilitate our commerce with the world. The doctrine originates with Dignāga, though, as integral to his theorization of the inferential relations that figure in the kind of reasoning theorized in his Hetucakraḍamaru. Let us, then, begin a more fine-grained look at some of the principal doctrines and arguments of these thinkers by starting with Dignāga’s introduction of the idea that inferential relations can be exhaustively characterized in terms of “exclusions” (apoha).

The Apoha Doctrine: Inferential Relations and a Theory of Mental Content

According to a typical thumbnail sketch, the apoha theory affirms that the referents of kind-terms (cow, on a canonical example) are not really existent features of reality; there is no really existent property like being a cow, but it is possible to construct such ideas entirely through a process of exclusion. The referent of a word like cow, on this view, is to be understood as really arrived at just by excluding whatever is not a non-cow. On the face of it, this appears to be circular, since it would seem that one could conceptually exclude all non-cows only if one already knows what a “cow” is—and that is just what the doctrine supposedly explains. To be sure, many of the objections reasonably leveled at the doctrine by the Brahmanical interlocutors of Dignāga and Dharmakīrti involve variations on this charge, which does have something to it. That there is nevertheless a profound and interesting insight here can be brought out, however, if we begin by appreciating that Dignāga is particularly theorizing what may be called inferential relations.9

Consider, then, a canonical example of inferential relations: the relation between being an oak and being a tree. (Another canonical example, favored in many basic Tibetan debate primers, is the relation between being blue and being a color.) Dignāga’s guiding thought is that every concept divides the world into two mutually exclusive classes: the class of everything that comes under that concept, and the complementary class of everything else. When the matter is thus understood, the question to be answered will always simply be: at what level of generality has this division been made? Dignāga’s insight is that the answer to this question can be expressed (and accordingly, that relative conceptual determinacy can be explained) entirely in negative terms—which is to say, without having to specify any really existent universals. The relative determinacy of concepts like “tree” and “oak,” then, can be precisely expressed without reference to timeless properties like being a tree or the set of all trees; rather, the term “tree” is contentful just insofar as it excludes from its purview everything there is that does not properly come under the concept “tree.” By itself, this example seems to involve the kind of question-begging circularity already noted. Dignāga’s idea, though, chiefly has to do with understanding the relative determinacy of concepts that invariably have their place in hierarchies of superordinate and subordinate categories. At the highest level of generality in such a hierarchy, a concept divides the world into two classes of great indeterminacy—as, for example, when we entertain the concept being existent, which excludes so little as to be of little practical use in most contexts. Starting at this level of generality, concepts are to be understood as having their place in branching subsets that become more narrowly circumscribed just insofar as they exclude more. Our example, then—the idea that the term “tree” excludes everything that does not come under the concept tree—is not really informative until we appreciate Dignāga’s point that the relatively greater determinacy of a subordinate concept in the same branching hierarchy (“oak,” for example) consists precisely in the fact that it excludes more than the superordinate concept “tree.” The concept oak is narrower than the concept tree, then, just because oak excludes everything in the world that is not a tree, plus all trees that are not oaks. The greater determinacy of the concept “oak” is thus a function of its excluding a larger domain, the scope of which had already been narrowed by the immediately superordinate category in an ascending hierarchy of increasing generality.

While it is in the nature of concepts never to achieve the concreteness that characterizes spatio-temporally determinate objects—no matter how fine grained our concepts become, their approach to the world of concrete particulars is necessarily asymptotic, finally giving way to moments where one can only point at this thing, right here—the most determinate concepts will always be those with the greatest exclusion ranges. Among the things that recommend Dignāga’s idea as an account of inferential relations is that this picture of branching hierarchies elegantly captures the asymmetry that essentially characterizes inferential relations; this makes good sense, that is, of why inferences are only warranted if, as it were, they move in the right direction in such a hierarchy. We are, then, entitled to infer that because something is an oak it is therefore a tree just because the concept oak excludes everything that tree excludes, as well as all other trees; we are not, however, entitled to claim that because something is a tree it must therefore be an oak, since not all trees are oaks. “This is an oak” has a more precise meaning than “this is a tree” just because everything that is incompatible with “this is a tree” is incompatible with “this is an oak,” but not the other way around. This gives us a way to specify negatively the different scope of concepts, whose relative determinacy is expressible simply in terms of their exclusion of all the other members of a branching hierarchy of concepts; the concept tree excludes things like chariots but not things like “oak” or “maple,” while “oak” excludes not only chariots (and everything else in the world that is not a tree), but additionally all trees that are not oaks.

With regard to Indian debates about the ontological status of linguistic universals, the point is that it is thus possible to account for the relative contentfulness of concepts without having to specify in what, precisely, their content “really” consists; we can account for the inferential asymmetry that defines the place of any concept (say, tree) in a larger scheme without having to specify either the intension (“being tall and leafy”) or the extension (“the set of all woody perennial plants”) of the concept. This is a desideratum, for Buddhists, since nothing one could positively specify in this way is really real in anything like the same way as particular trees; one can chop down a particular tree and use it for fuel, but one can do nothing of the sort with the abstract idea being a tree. Dignāga’s apoha doctrine, then, is a strictly formal account of the inferentially asymmetrical relations that track the relative scope of concepts.

Still, one may reasonably object that the doctrine as so far elaborated by Dignāga cannot tell us everything we need to understand in order to make sense of why it is that despite the ultimate unreality of linguistic items, the use of language nevertheless facilitates our getting around in a world of concrete particulars; for it seems we still have no way of knowing which particulars are rightly brought under any specific concept. If, in terms suggested by Frege, Dignāga has thus given a good account of the relative sense of concepts, we are still entitled to ask for a corresponding theory of reference. That is just what Dharmakīrti’s complementary elaboration of the apoha doctrine can be taken to add to the picture so far developed following Dignāga. And, given Dharmakīrti’s characteristic emphasis on causal efficacy as the criterion of the real, it is as we should expect that his significantly different development of the doctrine centrally involves reference to causal efficacy. On Dharmakīrti’s version of the doctrine, then, the idea is that the basis for our excluding whatever does not interest us at the moment is just our explanatory or practical interests—and these, he argued, can be described in terms of the effects we aim to bring about.10

If, for example, we are interested in ameliorating a fever, we may find it useful to refer to certain herbs known for their fever-reducing effects; given this, the most useful concept will be one that excludes all herbs that do not have such an effect. To be sure, this might again be thought to beg the question, since it could be objected that capacity to produce such and such an effect amounts to precisely the kind of abstraction that’s here supposed to be explained; after all, doesn’t possession of such a capacity amount to a defining characteristic of a general type, of which what is sought in this case would thus be a token? Dharmakīrti’s thought seems to be that if usage of the concept leads us to pick some herb, it can only be as a particular that it produces the desired effect, since only particulars have causal efficacy. Against that, one might object that while success in producing the desired effect may indeed count as evidence that one had rightly used the concept in question, it leaves unaddressed the conceptually prior question of what was understood such that our practices could in the first place culminate in our arriving at the desired particular; while it must indeed be some particular herb that finally produces the desired effect, it seems we are still owed an account of what we understood by the concept medicinal herb that could lead us to any such particular in the first place. It seems, then, that variations on the circularity objection may still loom.11

That worry might, however, be mitigated by the fact that Dharmakīrti seems, at the end of the day, to have had in mind a rather narrower sense of the relevant “capacity to produce an effect”; in particular, he seems to have meant that what is excluded, on any occasion of concept-use, is whatever does not produce an effect that is phenomenally the same as what is intended. More precisely, what is excluded is whatever does not produce that effect which is a cognition of the expected sort; the only “sameness” that characterizes the capacity in question, then, attaches to the judgments that concern what is perceptually encountered in the course of using the concept, not to the supposed referents of our concepts. This means, to be sure, that there is after all something invariant across uses of a concept; however, the relevant sameness is really not a property of the particulars in question, but rather of the perceptual judgments produced thereby. Dharmakīrti thus means to argue not only that the construction of kinds is relative to our explanatory or practical interests (what we exclude, when conceiving some herb, will vary depending on whether it is fever reduction that is wanted, or the flavoring of a stew), but also that the “sameness” that constitutes the kind in question exists only (as it were) in our heads. There are, in other words, no real, mind-independent similarities that could warrant the idea that our concepts correspond to natural kinds; rather, there are only phenomenal similarities that guide our attention to particulars, facilitating our sorting of things in one way or another depending entirely on our interests.

Among other things, Dharmakīrti’s causal application of the apoha idea explains why he could think that the inferential relations theorized by Dignāga can be characterized as relations of “identity” (tādātmya). This is as Dharmakīrti says in a concise text (the Sambandhaparīkṣā, or “Critical Investigation of Relations”) that theorizes what he took to be the only two kinds of relations there can be: causal relations (which have to do with one thing’s “arising from another,” or tad-utpatti), and identity (tādātmya) relations. These kinds of relations closely track the principal kinds of reasons that can warrant judgments: those consisting, respectively, in effects (kārya-hetu), as when one infers from the presence of smoke that there must be fire; and those consisting in a conceptual “essence,” or svabhāva-hetu, as when one knows that because something is an oak, it is therefore also a tree. The idea that the latter is an identity relation might seem to fly in the face of Dignāga’s recognition that inferential relations are essentially characterized by asymmetry; if the relation between being a tree and being an oak is one of identity, how then can we make sense of the fact that inferential relations between these concepts work in one direction but not in the other? (It is true that if something is an oak, then it is also a tree, but it is not true that if something is a tree, then it is an oak; a relation of identity, however, would seem to entail the truth of the latter, as well.) In fact, Dharmakīrti accepts Dignāga’s basic idea regarding inferential asymmetry; we must bear in mind, though, that Dharmakīrti’s development of the apoha doctrine addresses a different question than had oriented Dignāga’s elaboration, albeit a question complementary to Dignāga’s. One can, then, retain Dignāga’s thought that hierarchically branching concepts become more determinate insofar as they exclude more (and with this, the idea that inferential relations essentially involve the kind of asymmetry we have noted)—and at the same time appreciate the sense it makes for Dharmakīrti to think there is nonetheless a kind of identity evident in the application of concepts like tree and oak. Dharmakīrti’s point is that whether we find it relevant to apply the concept tree or the concept oak, either way it will be because we are presented with the same particular object having the same causal capacities. In other words, it is as a function of our conceptual interests (and not of the particulars we encounter) that we find it relevant to attend to it at one level of generality or another; whether it is most useful to pick something out as a tree or as an oak, though, either way it is just the same object that is producing our perceptual awareness thereof.12

In the respective elaborations of apoha doctrine by Dignāga and Dharmakīrti, then, we thus find two different but complementary ways to explain the construction of conceptual content entirely in terms of “exclusions.” For Dignāga, the eminently logical point of the doctrine is to make sense of the asymmetry that essentially characterizes inferential relations; his claim is that the relative determinacy of concepts can be characterized entirely in terms of the fact that concepts become more precise just insofar as they exclude more than the overarching concepts that subsume them. This makes good sense of why tree is a narrower concept than existent, and of why oak is narrower still—and it makes sense of this in such a way as to explain why inferences are valid in one direction but not the other. Dignāga’s development of this idea may, however, be vulnerable to the objection that this all makes sense only if we already know which particulars are rightly brought under any such concept—and to the extent that is just what a theory of meaning is supposed to explain, his version of the doctrine may be thought to beg the question. Dharmakīrti’s complementary development of the doctrine aims, though, to address just that problem, as Dharmakīrti instead advances the eminently epistemic point that our practical success in bringing particulars under concepts can be explained in terms of our excluding whatever does not produce the effects we desire on any occasion of concept-use. While Dharmakīrti’s version of the doctrine, too, may be thought vulnerable to the objection that there is a vicious circularity in play—this because the idea of “sameness of effect” would seem to be tantamount to the idea of a shared defining characteristic—it is clear at least that his approach undermines realist claims (typical of Mīmāṃsā and other Brahmanical schools of thought) to the effect that the referents of words correspond to really existent natural kinds. Against such views, Dharmakīrti would have us recognize that the content of concepts is thoroughly relative to our explanatory and practical interests, and that the only “sameness” that is anywhere in view is to be found simply in our judgments.

While there is much to recommend both these versions of the apoha doctrine—indeed, it can be argued that some of this anticipates (among other things) some contemporary cognitive-scientific approaches to the analysis of conceptual thought13—the non-Buddhist interlocutors of Dignāga and Dharmakīrti were never persuaded that these influential Buddhist philosophers could answer objections to the effect that the doctrine presupposes just the sort of thing it claims to explain. The doctrine continued, however, to be refined by later Buddhist thinkers who took their bearings from Dharmakīrti, and there can be no doubt of the increasing subtlety with which apoha theory was elaborated; chief among later exponents of the doctrine were Śāntarakṣita (d. 788 ce), Jñānaśrīmitra (fl. late 10th century), and Ratnakīrti (fl. early 11th century).14 Where one comes down on the ultimate tenability of the doctrine may well have to do with how promising one takes the basic epistemological commitments of Dignāga and Dharmakīrti to be; for particularly in Dharmakīrti’s elaboration of it, the apoha doctrine turns out to relate closely to problems entailed by both thinkers’ peculiarly strong emphasis on the constitutively nonconceptual character of perception. Let us turn, then, to a discussion of the characteristic epistemological contributions of Dignāga and Dharmakīrti.

Epistemology: Nonconceptual Perception and Its Idealist Implications

One way to understand particularly Dharmakīrti’s deployment of the apoha doctrine is as addressing a problem that arises given his emphasis (shared with Dignāga) on the essentially nonconceptual character of perception. The problem is this: if one shares with Dignāga and Dharmakīrti the basically empiricist conviction that perception (pratyakṣa) represents a privileged kind of cognition; if (with Dharmakīrti) one emphasizes that its privileged status is owing to the fact that perceptual awareness is, uniquely, caused by what it is of; and if, as both thinkers urged, conceptual awareness is in contrast held to be constitutively misleading insofar as it involves ultimately unreal abstractions—if all these commitments are held, it becomes a well-nigh intractable question how (or even whether) perceptual awareness can be a constraint on (or in any other way related to) the kind of conceptual cognition that alone counts as knowing. It seems, that is, that one can “know” only that such-and-such is the case (that it is hot outside, that this tree is a maple, that everything is impermanent)—and the content of any such that-clause must be a complex state of affairs under some description. That just is to say, however, that it is only as conceptualized that anything can count as the content of an episode of knowing. The problem with taking perceptual awareness to be constitutively nonconceptual, then, is that it becomes hard to see how what is present to perception could ever figure in what is present to the kind of discursive thought that alone counts as genuinely epistemic.15 In that case, though, how could perception ever count as an “epistemic criterion” (pramāṇa)? How could the privileged status that goes with perception’s being caused by its object be transferred to the kind of discursive awareness that, by Dignāga and Dharmakīrti’s lights, is not causally describable in this way? This can be understood as chief among the problems that Dharmakīrti’s version of the apoha doctrine addresses. On this way of thinking about apoha, the point is that conceptual content can, after all, be described as constructed from what is given to perceptual awareness; for on Dharmakīrti’s development of the doctrine, as we have seen, the conceptual process of “exclusion” explains how concepts can work even given that we cognitively encounter only the unique particulars given to perception.

Whether or not the apoha doctrine (or anything else) can resolve the issue, it is surely a central part of the project of Dignāga and Dharmakīrti to emphasize the nonconceptual character of perception. This is, we have noted, an intuitively plausible premise shared by most empiricists, who can be understood to take their bearings from the recognition that perception is, as Kant puts it, a faculty of “receptivity”—that perceptual awareness, in other words, is largely a function of the world’s impinging upon our sensory faculties. This characterization of perception not only has, though, good empiricist bona fides, but also reflects what can be recognized as a core commitment of most Buddhist thought: the idea that conceptual thought is constitutively misleading. Indeed, there is an important sense in which conceptual thought is, for Buddhists in general, precisely the problem to be overcome by the practice of the Buddhist path. It is typical of Buddhist philosophers to urge that what a Buddha experiences is an immediate, nonconceptual (nirvikalpaka) awareness of the truths taught by the Buddhist tradition—chiefly, the truth that selves are not really real. This idea partly reflects, no doubt, a recognition, on the part of a basically gnostic tradition—a tradition, that is, that takes religious transformation to result from rightly understanding something—that there is a profound difference between merely understanding and assenting to various propositional claims, and understanding in such a way as to be thoroughly changed by what is known. Buddhist philosophers surely mean to emphasize, then, that the transformative gnosis they have in mind consists in something much more than simply understanding and reproducing the kinds of arguments they are in the business of making. Still, the emphasis on nonconceptual awareness goes deeper than that; the really important point, surely, is that habitual attachment to the idea of one’s self represents the paradigm case of a specious conceptual thought. Concepts, that is, are essentially unreal abstractions comparable to the kinds of unitary wholes Buddhists are always concerned to refute—and the idea that our fleeting sensations must be states or parts of the “whole” that is one’s self represents the most pernicious example of conceptual thought, the one that Buddhist philosophers always have in view.16 The characteristically Buddhist emphasis on the desirability of taming conceptual thought goes closely together, then, with the idea that if we could but eliminate our habituated tendency to project unreal wholes, we would see that what really exists is just momentary sensations, and that there is no “self” over and above these.

So, the epistemology of Dignāga and Dharmakīrti advances eminently empiricist insights, but does so in large part because these are conducive to the realization of the truth most central to the Buddhist tradition. Whatever we say of their motives, though, there can be no doubt that their emphasis on the nonconceptual character of perception drives much of what is to be found in the philosophical projects of Dignāga and Dharmakīrti. Dignāga introduces this idea in the first chapter of his Pramāṇasamuccaya, which is dedicated to characterizing what is clearly first among pramāṇas for Dignāga: pratyakṣa, or “perception.”17 It is in verse three of the opening chapter that he characterizes perception as “devoid of conceptual construction” (kalpanāpoḍha). He immediately explains that “conceptual construction” here means “association with names and classes and so forth.” Concerned that that characterization defines perception as lacking only overt uses of language (as though one could count as having a conceptual thought only if one explicitly utters or entertains some linguistic item), Dharmakīrti subsequently revises Dignāga on this point, emphasizing that conceptual construction denotes any thought that is so much as suitable for association with linguistic items—any thought whose general or abstracted character represents the kind of thing that can in principle figure as linguistic referent. This qualification is important since without it, pre-linguistic and non-linguistic beings would not exemplify conceptual thought—and it is integral to a Buddhist account that all beings who are not yet “awakened” (buddha) are habitually misled by conceptual thought.

The 8th-century commentator Dharmottara thus explains this point with respect to one of the places where Dharmakīrti advances it:

In this regard, some thoughts have phenomenal content actually associated with expression—for example, the conception of an object which is a jar, on the part of one by whom the relevant linguistic convention is known, has as its phenomenal content an object associated with the word ‘jar.’ But some thoughts, even though not actually associated with expression, nevertheless have phenomenal content that is suitable for association with expression—for example, the conception had by a child by whom the relevant linguistic convention is not known. In this regard, if Dharmakīrti’s statement said only “conception has a phenomenal content associated with expression,” the conception had by someone by whom the relevant linguistic convention is unknown would not be included; but when there is reference to ‘suitable,’ the latter is also included. Even if the conception had by a new-born baby does not have phenomenal content actually associated with expression, a baby nevertheless has phenomenal content that is suitable for association with expression.18

Considering the question of how we can know, of anyone incapable of giving overt expression to their thoughts, that they nevertheless have such thoughts, Dharmottara notes that a newborn baby’s purposeful behavior is intelligible only with reference to thought that is “conceptual” in the requisite sense. A newborn’s seeking and settling on its mother’s breast, for example, makes sense only if we understand the baby as recognizing something it has previously experienced as sating its hunger.19 And recognition constitutively involves memory, which in turn involves reference to something not actually present—the hallmark of conceptual thought. Perception, in contrast, constitutively involves only actually present existents, which is why it represents a uniquely indubitable kind of awareness.

It seems, however, that even if it is allowed that the objects of perception are actually present, one might still doubt whether they really exist just as they appear; that a perceptible object (unlike a conceptual abstraction) is really present does not guarantee that it is rightly perceived. Indeed, this is as Dignāga and Dharmakīrti themselves would emphasize insofar as they commonly affirmed a basically representationalist epistemology. It follows, that is, especially from Dharmakīrti’s basically causal account of perception that what we are immediately aware of is just mental representations—“aspects” or “images” (ākāra) comparable to the “sense data” of the classical British empiricists. This is why Dignāga and Dharmakīrti were (as noted above) represented by doxographers not as direct realists, but as “proponents of the doctrine that external objects can only be inferred”; one can at most infer that things “out there” in the world are as represented in any occurrent cognition. Here, then, it becomes especially important to appreciate that despite what is suggested by the English translation equivalent “perception,” the word pratyakṣa, as Dignāga emphasized early in his account of the matter, does not refer only to sensory awareness; what really distinguishes perception is just that it is nonconceptual, and Dignāga argued from the beginning that it is not only sensory perception that qualifies as such. He argued that the kind of meditative awareness cultivated by skilled yogis counts as perceptual, but also (and more basically) that what he called “reflexive awareness” or “self-awareness” (svasaṃvitti) does, too—and in fact, there is reason to think he finally meant that only the latter kind of awareness is genuinely and nonconceptual.20

Dignāga’s idea here is that it is in the nature of every cognition to have two basically different kinds of “aspect” (ākāra): one that represents what the cognition is of (an aspect that is, say, a tree or a cow), and one that represents the cognition’s own awareness of itself as cognizing that. The latter he referred to as a cognition’s “self-awareness,” which effectively denotes something like the subjectivity of any awareness—the fact, that is, that no matter what any cognition is of, it is always integral to cognition that there is some felt way that that content seems for the cognition’s subject. Dignāga’s arguments for this aspect of cognition involve an appeal to the phenomenologically distinctive character of memory. Thus, he held that what distinguishes any first-order awareness from a subsequent recollection thereof is that in memory, one is aware not only of whatever it is that one had previously experienced, but also of oneself as having experienced that—and, he argued, the latter fact could be available to memory only if the same fact had been somehow present as part of the initial experience.21 (One cannot be said to remember something not initially experienced.) This is the idea, then, that integral to any awareness is its subject’s own awareness of the occurrence thereof; whenever one is aware that (say) there are cows and trees in a meadow, one is, ipso facto, also aware that one is having an awareness of that. And with this, we really have a plausible candidate for a uniquely indubitable sort of awareness; for while one can indeed doubt whether anything “out there” in the world really is as it seems, one cannot coherently doubt that that is how it seems.

Dignāga’s development of this idea is, however, rather indeterminate, and there is much that is unclear about it. In Dharmakīrti’s hands, however, the doctrine of svasaṃvitti more clearly becomes a doctrine to the effect that it is in the nature of cognitions to be self-intimating. This idea is expressed in an argument from Dharmakīrti that became one of the tradition’s principal arguments for the doctrine—the so-called sahopalambhaniyama argument, which urges that anything at all that one can know is always characterized by the “constraint” (niyama) that one can be aware of it only “together with the apprehension” thereof (saha-upalambha). This amounts to the familiarly idealist point that (as F. H. Bradley similarly argued in the late 19th century) “you cannot find fact unless in unity with sentience.”22 As understood by the Buddhists Dignāga and Dharmakīrti, the point cannot, of course, be that one is always aware of one’s self as the center of awareness; rather, they clearly mean to argue just that cognition itself is explanatorily basic—as Dharmakīrti puts it in the Pramāṇaviniścaya (a locus classicus for this argument), “there is not awareness of an object simply in virtue of there being an object; rather, [there is awareness of an object] by virtue of there being an awareness thereof.” That awareness occurs, then, is the self-evident basis of anything else we can know; hence (per Pramāṇaviniścaya 1.54cd), “the perception of objects does not make sense for one whose apprehension thereof is itself imperceptible.”23

So, while any cognition’s really being of an existent object can coherently be doubted, there being such a cognition cannot itself be thought to require demonstration. Of one’s contemplation of the Pythagorean theorem, for example, it makes sense to ask how or whether one knows it is true (“Because I learned it in school,” one might say, or, “Because I can give you the proof”); it does not, in contrast, make any sense to ask how you know simply that that is what you are thinking of. One is, rather, just immediately aware of whatever it is one is thinking; nothing at all could be present, Dharmakīrti argues, to someone whose own apprehension of things was not itself “perceptible” this way. It is just this self-intimating character of cognitions that Dharmakīrti means by “self-awareness” (svasaṃvitti), and it is easy to see how he and Dignāga could think this counts as a sort of perceptual awareness—for that is just to say it is an immediate, nonconceptual awareness. Recalling again the unusually thoroughgoing reductionism that is characteristic of Buddhist thought, we can see, too, how this may indeed be the only finally “perceptual” sort; for given the Buddhist premise that temporally enduring wholes are essentially conceptual fictions, even a seemingly perceptual awareness to the effect simply that “there is a tree” turns out already to be conceptual. What is immediately, indubitably known, in such a case, is simply that there is occurring an awareness to that effect. Dignāga says as much upon first introducing the idea of svasaṃvitti; having first characterized that as essentially perceptual, then, he anticipates this challenge: “If self-awareness (evident in things like our affective responses) is perception, then conceptual cognitions are, too.” Insofar, that is, as all cognitions are characterized by svasaṃvitti, it seems that even conceptual cognitions would turn out to be, after all, “perceptual”—but that, it seems, would effectively undermine the distinction (so important for Dignāga) between perceptual and conceptual thought. Addressing this worry, Dignāga embraces the point, but explains (at Pramāṇasamuccaya 1.7ab) why it is not a problem: “That’s true; even conceptual thought is admitted [as being perceptual] in terms of self-awareness”—but, he continues, “not with respect to its content (artha), because of the conceptual construction of that.”24 He thus emphasizes that in svasaṃvitti we are perceptually (which is just to say nonconceptually) aware of our own (conceptual) acts of judging; it is not, however, with respect to the content of such acts that this immediacy obtains—what is immediately apprehended is only the occurrence of judgments as particular mental events. Following through on the idea that we are in this sense indubitably acquainted only with the very occurrence of our mental events, Dharmakīrti and (more clearly) some of his later commentators emphasized that svasaṃvitti alone is a pāramārthikapramāṇa—only self-awareness, that is, counts as ultimately a pramāṇa, and everything else that is said by Dignāga and Dharmakīrti to count as “pramāṇa” is really just conventionally so.

Once again, then, intuitively plausible premises that Dignāga and Dharmakīrti share with classical empiricists—that perception represents the most basic kind of awareness, and this because perception, unlike conceptual fancies, can (Dharmakīrti emphasized) be described as caused by real objects—turn out to recommend the idealist conclusion that only the occurrence of cognition itself is indubitably known. To that extent, Dignāga and Dharmakīrti are aptly characterized as finally having meant to advance the Yogācāra school of Buddhist idealism. Nevertheless, it is typical of both thinkers to limit their arguments to an epistemological key, and generally to prescind from explicitly metaphysical or ontological conclusions; to that extent, it makes sense that traditional doxographical texts also represented them as “Sautrāntikas who follow reasoning”—a characterization that captures the modus operandi of both thinkers, who mostly restrict themselves to arguments involving the kinds of basically empiricist premises that were apt to win wide assent. Neither Dignāga nor Dharmakīrti made much reference to any specifically Yogācāra doctrines (they do not, e.g., refer to ālayavijñāna, or to the trisvabhāva doctrine), and the fuller elaboration of a complete idealist metaphysics was largely left to other thinkers in the tradition. Much as Bishop Berkeley took John Locke to have shown, then, Dignāga and Dharmakīrti mostly argued that if one accepts a basically empiricist account of perception, it turns out that one is already effectively committed to idealism.

It is, however, important to note an important exception to the generally epistemological modus operandi these thinkers shared; for Dignāga’s concise Ālambanaparīkṣā (“Critical Investigation of Percepts”) advances an argument with affinities to the essentially metaphysical argument for idealism that is most famously associated with Vasubandhu—the argument, in particular, that physical objects cannot coherently be conceived insofar as no coherent account of atoms can be given.25 Dignāga’s related argument concerns the intelligibility of the concept of a “percept” (ālambana), which returns us to Ābhidharmika categories. Thus, of the many sophisticated taxonomies of mental events and factors posited by Ābhidharmika philosophers, one of the most basic enumerates kinds of “causal conditions” (pratyaya) that figure in the occurrence of every moment of consciousness. Authors of the Abhidharma literature generally agreed there are four such conditions, a moment of (say) seeing some autumn trees must have as its causes: (1) a properly functioning ocular sense faculty (this is the adhipati-pratyaya, or “predominant condition”); (2) a previous moment of ocular experience (the samanantara-pratyaya, or “immediately preceding condition”—a category that explains how a series of fleeting moments of experience can seem, phenomenologically, to be a continuous flow; the reason we do not experience each new moment of seeing as having just popped into being is that the moments occur in continuous series); (3) a collection of other causes (the hetupratyaya, or “causal conditions which are causes,” where hetu refers to another list of causes); and (4) the autumn trees themselves, insofar as they are among the causes of my seeing them. The latter is the ālambana-pratyaya, or “percept condition.”

Dignāga’s brief “Critical Investigation of Percepts” analyzes this last category, which was defined by Buddhist philosophers as satisfying two conditions; thus, the “percept” denotes that one among the causes of a cognition which is at the same time what the cognition is of. Dignāga concisely argues that no physical object could satisfy both these conditions, and that it therefore cannot be physical objects that are present to awareness; this is because the kinds of things we typically take cognitions to be of cannot, in principle, be among the causes of cognitions. The argument is simple, and presupposes these premises: to be real, for most Buddhists as for most empiricists, just is to be capable of causally interacting with other existents; anything without this capacity does not ultimately exist. But on a Buddhist account, only irreducible things count as ultimately real. This means that only atomic sensible particulars (fleeting occurrences of shape and color and solidity and whatnot) could cause any cognition, since only these are ultimately real. The problem, though, is that it is not sensible “atoms” that are given in experience; rather, our experience typically concerns what J. L. Austin memorably referred to as “medium-sized dry goods.” In terms, then, of the Buddhist category that Dignāga is analyzing, the problem is that the kinds of things that can meet the causal condition (atoms, momentary sensa) do not meet the content condition; for experience is manifestly not of such things. Conversely, the kinds of things that do meet the content condition (the temporally enduring wholes that show up in experience) do not, on a Buddhist view, ultimately exist—which is just to say they cannot cause anything. Dignāga concludes from this that nothing with a basically atomic structure—which is to say, no really existent external objects—can be present to awareness.

Dignāga proposes, though, that an idealist account can circumvent the problem he thus elaborates. Thus, he takes the foregoing considerations to recommend concluding that only something intrinsic to cognition, only something that is itself “mental,” could be at once cause and content of any moment of consciousness. In order, though, to make good on his claim that something intrinsic to cognition makes sense as causing perceptual content, Dignāga ends up having to grant a couple of different senses of “being a cause.” The first way of “being a cause” is the one that makes sense if we adopt an objective, third-personal perspective on the occurrence of mental events. From this perspective—which need not involve reference to what the subject of a cognition herself takes its content to be—what Dignāga has in mind as “intrinsic to cognition” is something temporally prior to the mental event it causes; in particular, what causes occurrent cognitions is things like mental “seeds” (bīja) or “latent disposition” (vāsanā), which denote our beginninglessly habituated capacities and tendencies to experience the world in certain ways. It makes sense to say that such things are identified from a third-person perspective because even though there is a sense in which these are indeed internal to cognition—specifically, they are carried forward “within” a mental continuum, passed down until they “ripen” within the same series—they are nevertheless phenomenologically inaccessible. That is, the envisioned process of long-term mental development is not transparent to the subjects thereof, and the kinds of things Dignāga has in mind—long-transmitted dispositions to act this way or that—are not themselves available to introspection. From the subject’s perspective on any moment of experience, then, it is not seen that his or her experience represents the fruition of unconscious mental processes.

But that is of course to say that moments of experience are typically not of things like “habituated dispositions”; does that not, however, mean that Dignāga’s appeal to these various psychological artifacts cannot, after all, satisfy the other condition on being a percept (viz., that these be what the cognition is of)? Dignāga can say that the artifacts of past mental events that he has in mind are not, so long as their capacities are dormant, part of experiential content; they become contentful only when they “ripen,” at which point they show up as part of this or that experience. The Buddhist image of seeds is supposed to make sense of this. Just as a seed’s capacity remains dormant until it is no longer a seed but a sprout, so, too, subconscious processes can continuously transmit latent dispositions until some moment when they somehow burst into consciousness. Suppose, though, that “seeds” and “habituated dispositions,” even though not themselves experienced, thus meet the content condition insofar as they do become items of content upon ripening; even so, there is, Dignāga recognized, a further problem: the mental item that figures as the content of experience—a latent disposition or “seed,” at just the moment that it ripens into an occurrent sense datum—has to be understood, it seems, as part of (as “in”) that experience. But if the phenomenal content of any cognition must be understood as a part thereof, it becomes hard to see how it could still make sense as a cause of the cognition; for it does not make sense that a presently integral part of anything be at the same time a cause of the whole that comprises it. How could any part precede, as cause, the effect of which it is presently part?

It seems, then, that when mental content is identified from a first-person perspective—when we attend, that is, not to an experience’s long-term psycho-genesis, but to what is most salient for the subject thereof (namely, that it seems to be an experience of something)—it is hard to retain the same notion of causation that figures in an account involving “seeds” and “habituated dispositions” and the like. Thus, in order to salvage his claim that something intrinsic to awareness can make sense as both content and cause thereof, Dignāga has to allow that when considered from a first-person perspective, the “percepts” that show up in experience meet the causal condition—they count, that is, as also causes of the cognition whose content they are—only on a different understanding of what “being a cause” consists in. He appeals, in particular, to an alternative understanding of “cause” according to which anything co-occurrent with some event—as, for example, anything’s defining characteristic is, ipso facto, occurrent along with the thing itself—can be called one of that event’s “causes.”26 So, the phenomenologically accessible content of any cognition—what the cognition seems to its subject to be of—can be reckoned as one of the cognition’s causes only in the limited sense that whenever a cognition occurs, its content is present, too. Clearly, though, that is a very different sense of “cause” than when we entertain the idea that moments of experience are “caused” by a long and complex psychological past.27

It is not, then, an altogether straightforward matter even for an idealist to make sense at once of the causes and content of cognitions. The complexity of the issues here in play is evident in Dharmakīrti’s elaboration of a line of argument clearly prefigured by Dignāga’s concise “Critical Investigation of Percepts”—the line of argument, in particular, developed by Dharmakīrti around Pramāṇavārttika 3.208–218. In general, Dharmakīrti’s commitment to causal efficacy as the criterion of the real had him rather more sanguine than Dignāga about the viability of a causal account of conventionally understood perception; for example, he entertained the thought that while Dignāga rightly argued that individual atoms do not cause the kind of perceptual content we typically have in mind, it is nevertheless possible for aggregated atoms to gain a kind of collective causal efficacy that makes it possible for them, after all, to meet both of the conditions on being a “percept” condition. Nevertheless, for Dharmakīrti, too, such a picture was just provisionally acceptable, and when he argues that this is finally unsustainable, he advances an argument much like the foregoing one from Dignāga. There is, Dharmakīrti thus argued, a fundamental contradiction between the essential unity of momentary cognitions, and the manifold complexity that typically characterizes the content thereof—as, for example, when one is aware, in a single moment, of the variegated wings of a butterfly. Like Dignāga, then, Dharmakīrti finally argued that any attempt to reconcile momentary cognitions with a world of complex physical objects will break down. Indeed, he concludes at Pramāṇavārttika 3.209cd (here as elaborated by the commentator Manorathanandin) that even idealist alternatives are finally unsustainable so long as one remains committed to dualist conceptions of a unitary “subject” who enjoys any kinds of “objects” (whether external or internal) as content: “However things (colors, for example, whether they consist in something external or in cognition itself) are considered—whether as being unitary or as being complex—they break down, which is to say there is no way at all they can be established.”28

Despite the elusiveness of Dharmakīrti’s elaboration of the train of thought initiated by Dignāga’s “Critical Investigation of Percepts,” it is clear at least that both of these highly original and influential thinkers meant to advance radically revisionary accounts of the mental and of our epistemic situation; consistent with the Buddhist tradition’s radical contention that we are not at all as we habitually think we are, they thus aimed to show that the analysis particularly of perception shows that we are really warranted only in conclusions that are fundamentally at odds with common-sense understandings of mind and world. It is clear, as well, that the foregoing line of argument differs significantly from the familiarly empiricist kinds of arguments these thinkers more typically advance; Dignāga’s a priori analysis of the concept of a “percept,” and Dharmakīrti’s related interrogation of unitary cognitions having variegated content, represent lines of argument that are logically distinct from arguments to the effect that (e.g.,) we are immediately aware only of mental representations. Particularly as in the foregoing passage from Dharmakīrti, these basically metaphysical arguments show, as well, that the kind of disagreement concerning “external” and “internal” that is typically thought to divide realists from idealists may finally be something of a red herring. The real issue, rather, has to do with whether the unity taken to characterize momentary cognitions can be reconciled with the complexity that characterizes the content thereof (this is a distinctively Buddhist variation, surely, on the perennial philosophical problem of the relation between “the one” and “the many”)—and that is, finally, just as difficult a problem for idealists as for realists. In many significant Buddhist debates subsequent to those thus initiated by Dignāga and Dharmakīrti—in, for example, the “neither-one-nor-many” argument that was the signature move of Śāntarakṣita and Kamalaśīla, and in debates among late Indian proponents of Yogācāra over whether cognition itself must finally be essentially distinct from its content, or whether instead content is integral to cognition—questions about the reality of the physical often became, accordingly, secondary to a more basic question: the question of how or whether anything at all that is properly irreducible can relate to the manifest complexity of ordinary experience.29

Other Contributions and Influence

There is much else to be found in the works of Dignāga and Dharmakīrti; their collective corpus is quite large, and we are still very far from having even well-established critical editions of all of their works, let alone reliable translations of these into modern languages. Recent decades have, though, seen real growth in philosophically sophisticated treatments of their work, and there are exciting developments afoot in the recovery of Sanskrit texts of their writings (many of which still remain extant only in Tibetan translation).30 While the modern, critical study of the often elusive texts of Dignāga and Dharmakīrti thus remains, though, in a relatively fledgling state, still-flourishing Tibetan monastic curricula preserve a long tradition of engagement particularly with Dharmakīrti, whose lasting influence in India and Tibet greatly surpasses that of Dignāga.31 (In contrast, Chinese translations of Dignāga made some of his works available in East Asia, whereas none of Dharmakīrti’s works was ever translated into Chinese.) Traditionally trained Tibetan scholars well know, for example, that Dharmakīrti advanced an extensive proof, virtually unique in the tradition, of the reality of rebirth—and this in a section of his Pramāṇavārttika that contains, as well, an influential critique of theism, among other things.32 And the contributions of both thinkers to the theorization of logic and dialectics clearly figures importantly in the ritualized debate practices of Tibetan scholars, which abundantly exemplify the forms of argument commonly commended by Dignāga and Dharmakīrti.33 In the later history of Indian philosophy, too, their contributions to logic were decisively influential, and particularly the works of Dharmakīrti were much engaged by subsequent Indian thinkers, Buddhist and non-Buddhist alike. Indeed, anyone interested in classical Indian philosophies from any period subsequent to these great Buddhist thinkers will find it necessary to have some acquaintance with the thought and argument especially of Dharmakīrti, whose works posed challenges that were recognized as such by thinkers from all the various Brahmanical traditions of thought. To these other Indian philosophers, Dharmakīrti was surely most well known for the difficult but profound apoha theory of meaning, and for the sophisticated elaboration of a representationalist epistemology that could not go unanswered by those who argued, against him, for direct realism. And while the doctrine of svasaṃvitti was, in this regard, resisted especially by proponents of the Mīmāṃsā school of thought, who would have no truck with any doctrine that gave comfort to idealism, Dharmakīrti’s position was taken up by other Brahmanical thinkers, albeit for ends that were antithetical to the core Buddhist doctrine of no-self (anātmavāda).34 A number of great traditions of philosophical thought, then, would not be as they are but for the signal influence of Dignāga and Dharmakīrti.

Review of the Literature

Standard overviews of Dignāga and Dharmakīrti’s thought include Richard Hayes, Dignāga on the Interpretation of Signs, and John Dunne, Foundations of Dharmakīrti’s Philosophy; both volumes include significant translated excerpts from major works by the respective thinkers. Vincent Eltschinger gives an uncommonly good concise introduction to Dharmakīrti’s thought, while Georges Dreyfus’s Recognizing Reality gives a thorough, philosophically sophisticated account particularly of the Tibetan reception of Dharmakīrti. Useful overviews (by Richard Hayes and Shoryu Katsura, inter alia) of Dignāga’s thought can also be found in volume 9 of Karl H. Potter, ed., Encyclopedia of Indian Philosophies, which is entitled Buddhist Philosophy from 350 to 600 A.D.; a projected volume on Buddhist Philosophy from 600 to 750 A.D. (being edited by Eli Franco) will similarly make Dharmakīrti’s works available. Potter’s continuously updated Bibliography online to the Encyclopedia is a well-nigh exhaustive resource.35

Complete critical editions (to say nothing of modern translations) of the works of both thinkers remain desiderata; Birgit Kellner has given a thorough account of the state of progress with regard to Dharmakīrti’s works, while Eli Franco’s review of the inaugural volume in the series Sanskrit Texts from the Tibetan Autonomous Region concerns the impact of exciting new manuscript finds for the study particularly of Dignāga; the reviewed volume is a critical edition of Jinendrabuddhi’s 8th-century commentary on chapter 1 of Dignāga’s Pramāṇasamuccaya. (The availability of Jinendrabuddhi’s text enabled a reconstruction, by series editor Ernst Steinkellner, of Dignāga’s text itself, which is available online.) The second volume of Sanskrit Texts from the Tibetan Autonomous Region is a critical edition of chapters 1 and 2 of Dharmakīrti’s Pramāṇaviniścaya. Raniero Gnoli’s critical edition of chapter 1 of Dharmakīrti’s Pramāṇavārttika is standard, and is widely taken to have established the correct order of verses in that text; subsequent scholarly consensus is that the first chapter concerns svārthānumāna (“inference for one’s own sake”)—despite which, many widely available editions of the same text (e.g., Shastri’s) instead give the alternative order of chapters recommended by Manorathanandin’s commentary, which makes it important to note which chapter order any particular scholarly study follows.36

Despite the foregoing advances, Masaaki Hattori’s 1968 translation of the first chapter (on pratyakṣa) of Dignāga’s Pramāṇasamuccaya remains the principal point of access to Dignāga’s corpus for many scholars; Ole Holten Pind has made available the same work’s fifth chapter, concerning apoha. The 2016 collaborative volume Dignāga’s Investigation of the Percept gives a detailed, interdisciplinary overview of Indian and Tibetan perspectives on Dignāga’s concise Ālambanaparīkṣā. Significant translations from Dharmakīrti’s much larger corpus include Tillman Vetter’s German translation of the Pramāṇaviniścaya’s first chapter, and Tom Tillemans’s annotated translation of nearly 150 verses from the fourth chapter of the Pramāṇavārttika; John Dunne’s Foundations of Dharmakīrti’s Philosophy also includes extensive excerpts translated from Dharmakīrti and his commentators. Important thematic studies engaging Dignāga and Dharmakīrti include Mark Siderits et al., eds., Apoha: Buddhist Nominalism and Human Cognition; and volume 38/3 of the Journal of Indian Philosophy (2010), which comprises seven articles on various aspects of svasaṃvitti. For annotated references to further editions, translations, and thematic studies, see Dan Arnold, “The Philosophical Works and Influence of Dignāga and Dharmakīrti,” Oxford Bibliographies Online.37

Further Reading

Chu, Junjie. “On Dignāga’s Theory of the Object of Cognition as Presented in PS(V) 1.” Journal of the International Association of Buddhist Studies 29.2 (2006): 211–253.Find this resource:

Dravid, Raja Ram. The Problem of Universals in Indian Philosophy. Delhi: Motilal Banarsidass, 1972.Find this resource:

Dunne, John. “Realizing the Unreal: Dharmakīrti’s Theory of Yogic Perception.” Journal of Indian Philosophy 34.6 (2006): 497–519.Find this resource:

Eltschinger, Vincent. Penser l’autorité des Écritures: La polémique de Dharmakīrti contre la notion brahmanique orthodoxe d’un Veda sans auter. Vienna: Verlag der Österreichischen Akademie der Wissenschaften, 2007.Find this resource:

Kajiyama Yuichi. An Introduction to Buddhist Philosophy: An Annotated Translation of the Tarkabhāṣā of Mokṣākaragupta. Vienna: Arbeitskreis für tibetische und buddhistische Student Universität Vienna, 1998.Find this resource:

Kellner, Birgit. “Proving Idealism in Indian Buddhist Philosophy: Vasubandhu and Dharmakīrti.” In The Oxford Handbook of Indian Philosophy. Edited by Jonardon Ganeri. New York: Oxford University Press, 2015.Find this resource:

MacKenzie, Matthew. “The Illumination of Consciousness: Approaches to Self-Awareness in the Indian and Western Traditions.” Philosophy East and West 57.1 (2007): 40–62.Find this resource:

Matilal, Bimal Krishna. Perception: An Essay on Classical Indian Theories of Knowledge. Oxford: Clarendon, 1966.Find this resource:

Matilal, Bimal Krishna. The Character of Logic in India. Edited by Jonardon Ganeri and Heeraman Tiwari. Albany: SUNY Press, 1998.Find this resource:

Mookerjee, Satkari. The Buddhist Philosophy of Universal Flux: An Exposition of the Philosophy of Critical Realism as Expounded by Dignāga. Delhi: Motilal Banarsidass, 1975.Find this resource:

Patil, Parimal G.Against a Hindu God: Buddhist Philosophy of Religion in India. New York: Columbia University Press, 2009.Find this resource:


(1.) In recurrently characterizing his own position as sautrāntika (“following the sūtras”), Vasubandhu may not have meant to refer to any “school” of thought, but only to suggest that disputes among Ābhidharmikas ought finally to be settled with reference to the sūtra literature that alone was taken to express the Buddha’s own teachings; later, however, the term ossified in ways that arguably make it anachronistic to read the associated school of thought back into Vasubandhu’s work. On the term as a doxographical category pertaining to Vasubandhu, see Robert Kritzer, “Sautrāntika in the Abhidharmakośabhāṣya,” Journal of Indian Philosophy 26.2 (2003): 331–384. On the different subtypes of “Sautrāntika,” see Katsumi Mimaki, “Le Grub mtha’ rnam bźag rin chen phreṅ ba de dKon mchog ’jigs med dbaṅ po,” Zinbun: Memoirs of the Research Institute for Humanistic Studies, Kyoto University 14 (1977): 84. On the philosophically constructive nature of Tibetan doxographical texts, see José Cabezón, “The Canonization of Philosophy and the Rhetoric of Siddhānta in Indo-Tibetan Buddhism,” in Buddha Nature: A Festschrift in Honor of Minoru Kiyota, eds. Paul Griffiths and John Keenan (Reno, Nevada: Buddhist Books International, 1990), 7–26. On the Abhidharma literature more generally, see Steven Collins, “What Are Buddhists Doing When They Deny the Self?” in Religion and Practical Reason: New Essays in the Comparative Philosophy of Religions, eds. Frank Reynolds and David Tracy (SUNY Press, 1994), 59–86; Collett Cox, Disputed Dharmas: Early Buddhist Theories on Existence (Tokyo: The International Institute for Buddhist Studies, 1995); Rupert Gethin, Foundations of Buddhism (Oxford University Press, 1998), 202–223; Alexis Sanderson, “The Sarvāstivāda and its Critics: Anātmavāda and the Theory of Karma,” in Buddhism into the Year 2000 (Los Angeles: Dhammakaya Foundation, 1994), 33–48; and Paul Williams, “On the Abhidharma Ontology,” Journal of Indian Philosophy 9 (1981): 227–257.

(2.) For “Buddhist Logic,” see Th. Stcherbatsky’s dated but still useful Buddhist Logic (New York: Dover, 1962), the second volume of which comprises a complete translation of Dharmakīrti’s Nyāyabindu. For “Buddhist Epistemology,” see, inter alia, Mark Siderits, Buddhism as Philosophy (Indianapolis, IN: Hackett, 2007), 208–230. Useful overviews of these thinkers include Richard Hayes, Dignāga on the Interpretation of Signs (Dordrecht: Kluwer, 1988); Vincent Eltschinger, “Dharmakīrti,” Revue Internationale de Philosophie 64.3 (2010): 397–440; and John Dunne, Foundations of Dharmakīrti’s Philosophy (Boston: Wisdom, 2004). With regard to the difficulty of determining the considered position of Dharmakīrti, the latter volume introduces the idea of a “sliding scale” of analysis, according to which Dharmakīrti’s claims differ depending on whether taken from an empiricist perspective or an idealist one. For an argument that these supposedly divergent perspectives actually go more closely together than this suggests, see Dan Arnold, “Buddhist Idealism, Epistemic and Otherwise: Thoughts on the Alternating Perspectives of Dharmakīrti,” Sophia 47.1 (2008): 3–28. For a more extensive, annotated bibliography on both thinkers, see Dan Arnold, “The Philosophical Works and Influence of Dignāga and Dharmakīrti,” Oxford Bibliographies Online, 2012.

(3.) On the interpretation of Indian logic in general, see Jonardon Ganeri, “Introduction: Indian Logic and the Colonization of Reason,” in Indian Logic: A Reader, ed. Ganeri (London: Curzon, 2001), 1–25; Mark Siderits, “Deductive, Inductive, Both or Neither?” Journal of Indian Philosophy 31 (2003): 303–321; and Shoryu Katsura and Ernst Steinkellner, eds., The Role of the Example (Dr̥ṣṭānta) in Classical Indian Logic (Vienna: Arbeitskreis für Tibetische und Buddhistische Studien Universität Wien, 2004). See also Claus Oetke, Studies on the Doctrine of Trairūpya (Vienna: Arbeitskreis für Tibetische und Buddhistische Studien, Universität Wien, 1994); Brendan Gillon, “Dharmakīrti and the Problem of Induction,” in Studies in the Buddhist Epistemological Tradition: Proceedings of the Second International Dharmakīrti Conference, ed. Ernst Steinkellner (Vienna: Österreichische Akademie der Wissenschaften, 1991), 53–58; and Ernst Steinkellner, “On the Interpretation of the Svabhāvahetuḥ,” Wiener Zeitschrift für die Kunde Südasiens und Archiv für Indische Philosophie 18 (1974): 117–129. On Dignāga’s Hetucakraḍamaru, see Huanhuan He and Leonard W. J. van der Kuijp, “Once Again on the *Hetucakraḍamaru: Rotating the Wheels,” Journal of Indian Philosophy 44 (2016): 267–302. For Dharmakīrti’s principal writings on logic, see Ernst Steinkellner, Dharmakīrti’s Early Logic: An Annotated German Translation of the Logical Parts in Pramāṇavārttika 1 and Vṛtti (Tokyo: International Institute for Buddhist Studies, 2013); and Ernst Steinkellner, ed., Dharmakīrti’s Hetubindu: Critically Edited by Ernst Steinkellner on the basis of preparatory work by Helmut Krasser with a transliteration of the Gilgit fragment by Klaus Wille, Sanskrit Texts from the Tibetan Autonomous Region 19 (Vienna: Austrian Academy of Sciences Press, 2016).

(4.) For a translation of Dignāga’s principal work on perception, see Masaaki Hattori, Dignāga, On Perception, Being the Pratyakṣapariccheda of Dignāga’s Pramāṇasamuccaya from the Sanskrit Fragments and the Tibetan Versions (Cambridge, MA: Harvard University Press, 1968). Hattori’s pioneering translation is from the Tibetan translations of the Pramāṇasamuccaya that were all that was known to be available at the time; based on the edition of a Sanskrit commentary that has since been recovered (that of Jinendrabuddhi), Ernst Steinkellner has edited a reliable reconstruction that surely gets us closer to Dignāga’s text: Ernst Steinkellner, Dignāga’s Pramāṇasamuccaya, Chapter 1: A Hypothetical Reconstruction of the Sanskrit Text with the Help of the Two Tibetan Translations on the Basis of the Hitherto Known Sanskrit Fragments and the Linguistic Materials Gained from Jinendrabuddhi’s Ṭīkā (Vienna: Österreichischen Akademie der Wissenschaften, 2005). For a critical edition of the commentary from which Dignāga’s text was reconstructed, see Ernst Steinkellner, Helmut Krasser, and Horst Lasic, eds., Jinendrabuddhi’s Viśālāmalavatī Pramāṇasamuccayaṭīkā: Chapter 1, Part I: Critical Edition, Sanskrit Texts from the Tibetan Autonomous Region 1 (Vienna: Österreichischen Akademie der Wissenschaften, 2005). For a translation and studies of Dignāga’s Ālambanaparīkṣā, see Douglas Duckworth et al., eds., Dignāga’s Investigation of the Percept (Ālambana-Parīkṣā) and its Philosophical Legacy in India and Tibet (Oxford: Oxford University Press, 2016).

5. Quoting Pramāṇavārttika 3.3, which is often cited as 2.3; on the order of chapters in this work, though, scholarly consensus favors the arguments of Raniero Gnoli, made in the introduction to Gnoli, ed., The Pramāṇavārttikam of Dharmakīrti: The First Chapter with the Autocommentary (Rome: Istituto Italiano Per Il Medio ed Estremo Oriente, 1960).

(6.) There is presently no complete translation of either of Dharmakīrti’s most extensive works, the Pramāṇavārttika or the Pramāṇaviniścaya; a critical edition of the newly available Sanskrit text of the first two chapters of the latter, however, is among the important volumes in the recently inaugurated series Sanskrit Texts from the Tibetan Autonomous Region, and collaborative work that is ongoing in Vienna stands to contribute greatly to efforts at making the whole Pramāṇavārttika available in translation. For the former, see Ernst Steinkellner, ed., Dharmakīrti’s Pramāṇaviniścaya, Chapters 1 and 2, Sanskrit Texts from the Tibetan Autonomous Region 2 (Vienna: Austrian Academy of Sciences Press, 2007); for a German translation of (the Tibetan translation of) chapter 1 of the same text, see Tillman Vetter, Dharmakīrti’s Pramāṇaviniścaya, I. Kapitel: Pratyakṣam: Einleitung, Text der tibetischen Übersetzung, Sanskritfragmente, deutsche Übersetzung, Österreichische Akademie der Wissenschaften Philosophisch-Historische Klasse, Sitzungsberichte 250 (Vienna: Hermann Böhlaus Nachf, 1966). For a dated but still useful translation of the Nyāyabindu, see Stcherbatsky, Buddhist Logic, vol. 2; Stcherbatsky’s edition of the same text has been superseded by Paṇḍita Dalsukhbhai Malvania, ed., Paṇḍita Durveka Miśra’s Dharmottarapradīpa, 2d ed. (Patna, India: Kashiprasad Jayaswal Research Institute, 1971); this includes the important and innovative commentary of Dharmottara, as well as the subcommentary of the Brahmin scholar Durvekamiśra. A readily available edition of the Pramāṇavārttika is Swami Dwarikadas Shastri, ed., Pramāṇavārttika, Ācāryamanorathanandivṛttiyuktam (Varanasi, India: Bauddha Bharati, 1968); for a more complete account of the state of scholarship on this extensive and difficult text, however, see Birgit Kellner, “Towards a Critical Edition of Dharmakīrti’s Pramāṇavārttika,” Text Genealogy, Textual Criticism and Editorial Technique: Wiener Zeitschrift für die Kunde Südasiens 52–53 (2009–2010): 161–211.

(7.) For characterizations of Dharmakīrti as typically arguing from broadly empiricist premises, see Dan Arnold, Brains, Buddhas, and Believing: The Problem of Intentionality in Classical Buddhist and Cognitive-Scientific Philosophy of Mind (New York: Columbia University Press, 2012), chs. 1 and 5. Dharmakīrti’s strongest and most explicit argument for idealism is developed at Pramāṇavārttika 3.208–218, which develops a line of thought with close affinities to the argument Dignāga advanced in the Ālambanaparīkṣā; a helpfully concise characterization of these arguments can be found in Vincent Eltschinger, “Dharmakīrti,” Revue Internationale de Philosophie 64.3 (2010): 429–430. More on these arguments below.

(8.) See Richard Hayes and Brendan Gillon, “Dharmakīrti on the Role of Causation in Inference as Presented in Pramāṇavārttika Svopajñavṛtti 11–38,” Journal of Indian Philosophy 36 (2008): 335–404.

(9.) On Dignāga’s account of apoha, see Jonardon Ganeri, Philosophy in Classical India: The Proper Work of Reason (London: Routledge, 2001), 106–111; Hayes, Dignāga on the Interpretation of Signs; Shoryu Katsura, “The Apoha Theory of Dignāga,” Indogaku Bukkyogaku Kenkyu 28.1 (1979), 16–20; and, for an interpretation and translation of Dignāga’s complete elaboration of the doctrine, Ole Holten Pind, Dignāga’s Philosophy of Language: Pramāṇasamuccayavṛtti V on anyāpoha (Vienna: Verlag der Österreichischen Akademie der Wissenschaften, 2015), part 2. On circularity objections, see Pascale Hugon, “Breaking the Circle: Dharmakīrti’s Response to the Charge of Circularity against the Apoha Theory and Its Tibetan Adaptation,” Journal of Indian Philosophy 37.6 (2009): 533–557.

(10.) On Dharmakīrti’s complementary elaboration of the apoha doctrine, see Dunne, Foundations of Dharmakīrti’s Philosophy; Shoryu Katsura, “Dignāga and Dharmakīrti on Apoha,” in Studies in the Buddhist Epistemological Tradition, ed. Ernst Steinkellner (Vienna: Österreichische Akademie der Wissenschaften, 1991), 129–146; and Arnold, Brains, Buddhas, and Believing, ch. 4.

(11.) See Pascale Hugon, “Breaking the Circle.”

(12.) The text in which Dharmakīrti theorizes the two kinds of relations is the Sambandhaparīkṣā (“Critical Investigation of Relations”), for the text of which see Erich Frauwallner, “Dharmakīrtis Sambandhaparīkṣā: Text und Übersetzung,” Wiener Zeitschrift für die Kunde Morgenlandes 41 (1934): 261–300. On Dharmakīrti’s finally causal account of the sense in which inferential relations are “identity” relations, see Dunne, Foundations of Dharmakīrti’s Philosophy, 203–218; see, as well, Steinkellner, “On the Interpretation of the Svabhāvahetuḥ.”

(13.) See Mark Siderits, Tom Tillemans, and Arindam Chakrabarti, eds., Apoha: Buddhist Nominalism and Human Cognition (New York: Columbia University Press, 2011).

(14.) On later elaborations of apoha, see especially Lawrence McCrea and Parimal Patil, Buddhist Philosophy of Language in India: Jñānaśrīmitra on Exclusion (New York: Columbia University Press, 2010); and Parimal Patil, Against a Hindu God: Buddhist Philosophy of Religion in India (New York: Columbia University Press, 2009), 195–247. A sense of the range and the many implications of the doctrine can also be gained from Siderits et al., eds., Apoha (note 15, above).

(15.) See Georges Dreyfus, “Is Perception Intentional? A Preliminary Exploration of Intentionality in Dharmakīrti,” in Birgit Kellner et al., eds., Pramāṇakīrtiḥ: Papers Dedicated to Ernst Steinkellner on the Occasion of His 70th Birthday (Vienna: Arbeitskreis für tibetische und buddhistische Studien, Universität Wien, 2007), 95–113; Shoryu Katsura, “Dharmakīrti’s Theory of Truth,” Journal of Indian Philosophy 12.3 (1984): 215–235; and Dan Arnold, “Dharmakīrti and Dharmottara on the Intentionality of Perception: Selections from Nyāyabindu (An Epitome of Philosophy),” in Buddhist Philosophy: Essential Readings, eds. William Edelglass and Jay Garfield (Oxford University Press, 2009), 186–196.

(16.) Richard Hayes’s “Principled Atheism in the Buddhist Scholastic Tradition” (Journal of Indian Philosophy 16 [1988]: 5–28) nicely brings out something of the ways in which many Buddhist arguments are variations on this.

(17.) See Masaaki Hattori, Dignāga, On Perception.

(18.) For this passage from Dharmottara’s commentary on Nyāyabindu 1.5, see Dan Arnold, “On (Non-semantically) Remembering Conventions: Dharmakīrti and Dharmottara on Saṃketakāla,” in Logic and Belief in Indian Philosophy, ed. Piotr Balcerowicz (Delhi: Motilal Banarsidass, 2009), 551; see also Stcherbatsky, Buddhist Logic, vol. 2, 19–20.

(19.) This account presupposes the reality of rebirth, which is axiomatic for these Buddhist thinkers; given that idea, a newborn is never nursing for the first time, and its purposeful behavior can always be explained as informed by recognition from previous lifetimes. For Dharmakīrti’s proof of rebirth, see Eli Franco, Dharmakīrti on Compassion and Rebirth (Vienna: Arbeitskreis für tibetische und buddhistische Studien, Universität Wien, 1997).

(20.) For Dignāga’s account of svasaṃvitti (or svasaṃvedana), see Birgit Kellner, “Self-Awareness (svasaṃvedana) in Dignāga’s Pramāṇasamuccaya and -vṛtti – A Close Reading,” Journal of Indian Philosophy 38.3 (2010): 203–231. On the vexed question of whether svasaṃvitti is one of three kinds of perception admitted by Dignāga or one of four, see Eli Franco, “Did Dignāga Accept Four Types of Perception?” Journal of Indian Philosophy 21 (1993): 295–299.

(21.) See Jonardon Ganeri, “Self-Intimation, Memory and Personal Identity,” Journal of Indian Philosophy 27 (1999): 469–483.

(22.) F. H. Bradley, as quoted in Ralph Barton Perry, Present Philosophical Tendencies (New York: Longmans, Green, 1921), 134.

(23.) The first quotation is from the commentary on Pramāṇaviniścaya 1.54cd, and the second is the half verse itself; for the text of this, see Steinkellner, Dharmakīrti’s Pramāṇaviniścaya, pp. 40–41. For more extensive discussion of this and proximate passages, see Arnold, Brains, Buddhas, and Believing, 175–183. For an alternative translation, see Georges Dreyfus and Christian Lindtner, “The Yogācāra Philosophy of Dignāga and Dharmakīrti,” Studies in Central and East Asian Religions (Journal of the Seminar for Buddhist Studies, Copenhagen and Aarhus) 2 (1989): 27–52. For an in-depth study of Dharmakīrti’s sahopalambhaniyama argument, see Takashi Iwata, Sahopalambhaniyama: Struktur und Entwicklung des Schlusses von der Tatsache, daß Erkenntnis und Gegenstand ausschließlich zusammen wahrgenommen werden, auf deren Nichtverschiedenheit (Stuttgart: Franz Steiner Verlag, 1991). See, as well, volume 38.2 of the Journal of Indian Philosophy (2010), a special issue on the topic of svasaṃvitti.

(24.) Translated from Steinkellner’s reconstruction (2005), 3.

(25.) Vasubandhu’s argument to this effect is at verses 11–15 of the Viṃśatikā, on which see Matthew Kapstein, “Mereological Considerations in Vasubandhu’s ‘Proof of Idealism’,” in Kapstein’s Reason’s Traces: Identity and Interpretation in Indian & Tibetan Buddhist Thought (Boston: Wisdom Publications, 2001), 181–204.

(26.) Dignāga can here appeal to the Ābhidharmika category of sahabhūhetu, which denotes the kind of “cause” (hetu), which simply goes “with [something’s] being” (sahabhū); see Abhidharmakośa 2.50c–d, where Vasubandhu gives the relation between “characteristic” and “characterized” as an example of this.

(27.) On Dignāga’s Ālambanaparīkṣā, see Douglas Duckworth et al., eds., Dignāga’s Investigation of the Percept (Ālambana-Parīkṣā) and Its Philosophical Legacy in India and Tibet (Oxford: Oxford University Press, 2016). For a reading of that text as finally bringing into play problems regarding the relation between third-personal and first-personal perspectives on cognition, see Dan Arnold, “Philosophy of Mind’s ‘Hard Problem’ in Light of Buddhist Idealism,” in Philosophy’s Perennial Questions: Comparing Buddhist and Western Approaches, ed. Steven Emmanuel (New York: Columbia University Press, forthcoming).

(28.) For a concise characterization of Dharmakīrti’s arguments at Pramāṇavārttika 3.208–218, see Vincent Eltschinger, “Dharmakīrti,” Revue Internationale de Philosophie 64.3 (2010): 429–430. See, as well, Dunne, Foundations of Dharmakīrti’s Philosophy, 98–113, and also (for translations from Pramāṇavārttika 3.194–224 and two commentaries thereon) 396–411.

(29.) On the logically distinct character of the metaphysical argument epitomized by Vasubandhu’s Viṃśatikā (and followed by Dignāga and Dharmakīrti in the foregoing discussion), particularly vis-à-vis the epistemic arguments more typical of Dignāga and Dharmakīrti, see Dan Arnold, “Buddhist Idealism, Epistemic and Otherwise: Thoughts on the Alternating Perspectives of Dharmakīrti,” Sophia 47.1 (2008): 3–28; for a different perspective on the same issue, see Birgit Kellner, “Proving Idealism in Buddhist Philosophy: Vasubandhu and Dharmakīrti,” in The Oxford Handbook of Indian Philosophy, ed. Jonardon Ganeri (Oxford Handbooks Online, 2015). See, as well, Isabelle Ratié, “On the Distinction between Epistemic and Metaphysical Buddhist Idealisms: A Śaiva Perspective,” Journal of Indian Philosophy 42 (2014): 353–375. On the “neither one nor many” argument of Śāntarakṣita and Kamalaśīla, see Tom Tillemans, “The ‘Neither One nor Many’ Argument for śūnyatā and Its Tibetan Interpretations: Background Information and Source Materials,” Études de Lettres (University of Lausanne) 3 (1982): 103–128. On late Yogācāra debates regarding the status of mental content, see Yuichi Kajiyama, “Controversy between the sākāra- and nirākāra-vādins of the yogācāra School—Some Materials,” in Kajiyama’sStudies in Buddhist Philosophy (Selected Papers) (Kyoto: Rinsen, 1989), 389–400. See, as well, Shinya Moriyama, “Ratnākaraśānti’s Theory of Cognition with False Mental Images (*aḷikākāravāda) and the Neither-One-Nor-Many Argument,” Journal of Indian Philosophy 42 (2014): 339–351; and Sara McClintock, “Kamalaśīla on the Nature of Phenomenal Content (ākāra) in Cognition: A Close Reading of TSP ad TS 3626 and Related Passages,” Journal of Indian Philosophy 42 (2014): 327–337.

(30.) See Eli Franco, “A New Era in the Study of Buddhist Philosophy,” Journal of Indian Philosophy 34.3 (2006): 221–227. This is a short review article on a crucial new resource for the study of Dignāga (published as the first volume in the new series Sanskrit Texts from the Tibetan Autonomous Region): Ernst Steinkellner et al., eds., Jinendrabuddhi’s Viśālāmalavatī Pramāṇasamuccayaṭīkā: Chapter 1 (Vienna: Österreichischen Akademie der Wissenschaften, 2005). See, too, Birgit Kellner, “Towards a Critical Edition of Dharmakīrti’s Pramāṇavārttika,” in Text Genealogy, Textual Criticism and Editorial Technique: Wiener Zeitschrift für die Kunde Südasiens 52–53 (2009–2010): 161–211.

(31.) See Georges Dreyfus, Recognizing Reality: Dharmakīrti’s Philosophy and Its Tibetan Interpreters (Albany: SUNY Press, 1997).

(32.) See Franco, Dharmakīrti on Compassion and Rebirth; Richard Hayes, “Principled Atheism in the Buddhist Scholastic Tradition”; Roger Jackson, Is Enlightenment Possible? Dharmakīrti and rGyal tshab rje on Knowledge, Rebirth, No-Self and Liberation (Ithaca, NY: Snow Lion, 1993); and Dan Arnold, “Dharmakīrti’s Dualism: Critical Reflections on a Buddhist Proof of Rebirth,” Philosophy Compass 3.5 (2008): 1079–1096.

(33.) See Daniel Perdue, Debate in Tibetan Buddhism (Ithaca, NY: Snow Lion, 1992).

(34.) See Alex Watson, “Bhaṭṭa Rāmakaṇṭha’s Elaboration of Self-Awareness (Svasaṃvedana), and How It Differs from Dharmakīrti’s Exposition of the Concept,” Journal of Indian Philosophy 38 (2010): 297–321; Isabelle Ratié, “The Dreamer and the Yogin: On the Relationship between Buddhist and Śaiva Idealisms,” Bulletin of the School of Oriental and African Studies 73.3 (2010): 437–478; and Lawrence McCrea, “Abhinavagupta as Intellectual Historian of Buddhism,” in Around Abhinavagupta: Aspects of the Intellectual History of Kashmir from the Ninth to the Eleventh Century, eds. Eli Franco and Isabelle Ratié (Berlin: LIT Verlag, 2016), 263–286.

(35.) Richard Hayes, Dignāga on the Interpretation of Signs (Dordrecht: Kluwer, 1988); John D. Dunne, Foundations of Dharmakīrti’s Philosophy (Boston: Wisdom Publications, 2004); Vincent Eltschinger, “Dharmakīrti,” Revue Internationale de Philosophie 64.3 (2010): 397–440; Georges Dreyfus, Recognizing Reality: Dharmakīrti’s Philosophy and its Tibetan Interpreters (Albany: SUNY Press, 1997); Karl H. Potter, ed., Encyclopedia of Indian Philosophies, vol. 9: Buddhist Philosophy from 350 to 600 A.D. (Delhi: Motilal Banarsidass, 2003).

(36.) Birgit Kellner, “Towards a Critical Edition of Dharmakīrti’s Pramāṇavārttika,” in Text Genealogy, Textual Criticism and Editorial Technique: Wiener Zeitschrift für die Kunde Südasiens 52–53 (2009–2010): 161–211; Eli Franco, “A New Era in the Study of Buddhist Philosophy,” Journal of Indian Philosophy 34.3 (2006): 221–227; Ernst Steinkellner et al., eds., Jinendrabuddhi’s Viśālāmalavatī Pramāṇasamuccayaṭīkā, Sanskrit Texts from the Tibetan Autonomous Region 1 (Vienna: Österreichischen Akademie der Wissenschaften, 2005); Ernst Steinkellner, ed., Dharmakīrti’s Pramāṇaviniścaya, Chapters 1 and 2, Sanskrit Texts from the Tibetan Autonomous Region 2 (Vienna: Österreichischen Akademie der Wissenschaften, 2007); Raniero Gnoli, ed., The Pramāṇavārttikam of Dharmakīrti: The First Chapter with the Autocommentary; Text and Critical Notes (Rome: Istituto Italiano Per Il Medio ed Estremo Oriente, 1960); Swami Dwarikadas Shastri, ed., Pramāṇavārttika, Ācāryamanorathanandivṛttiyuktam (Varanasi, India: Bauddha Bharati, 1968).

(37.) Masaaki Hattori, Dignāga, On Perception, Being the Pratyakṣapariccheda of Dignāga’s Pramāṇasamuccaya from the Sanskrit Fragments and the Tibetan Versions (Cambridge, MA: Harvard University Press, 1968); Ole Holten Pind, Dignāga’s Philosophy of Language: Pramāṇasamuccayavṛtti V on anyāpoha, part 2 (Vienna: Verlag der Österreichischen Akademie der Wissenschaften, 2015); Douglas Duckworth et al., eds., Dignāga’s Investigation of the Percept (Ālambana-Parīkṣā) and Its Philosophical Legacy in India and Tibet (Oxford University Press, 2016); Tillman Vetter, Dharmakīrti’s Pramāṇaviniścaya, I. Kapitel: Pratyakṣam: Einleitung, Text der tibetischen Übersetzung, Sanskritfragmente, deutsche Übersetzung, Österreichische Akademie der Wissenschaften Philosophisch-Historische Klasse, Sitzungsberichte 250 (Vienna: Hermann Böhlaus Nachf, 1966); Tom J. F. Tillemans, Dharmakīrti’s Pramāṇavārttika: An Annotated Translation of the Fourth Chapter (parārthānumāna), volume 1 (k.1–148), Österreichische Akademie der Wissenschaften Philosophisch-Historische Klasse, Sitzungsberichte 675 (Vienna: Verlag der Österreichischen Akademie der Wissenschaften, 2000). John Dunne, Foundations of Dharmakīrti’s Philosophy, 331–415; Mark Siderits et al., eds., Apoha: Buddhist Nominalism and Human Cognition (New York: Columbia University Press, 2011). From the special issue of the Journal of Indian Philosophy concerning svasaṃvitti (38.3, 2010), see, e.g., Birgit Kellner, “Self-Awareness (svasaṃvedana) in Dignāga’s Pramāṇasamuccaya and –vṛtti: A Close Reading,” 203–231.