Articles by alphabetic order
 Ā Ī Ñ Ś Ū Ö Ō
1 2 3 4 5 6 7 8 9 0

How to Do Things with Candrakirti: A Comparative Study in Anti-skepticism

From Chinese Buddhist Encyclopedia
Jump to navigation Jump to search
MG 5813.JPG
Mount tai rock inscriptions.jpg
Uluchi Arujann.jpg
Shantideva2 (2).jpg
Win ound.jpg
316 ium.jpg
AmazinvArt 2.jpg
Grea n1.jpg
3cvb0 n.jpg
773 vement.jpg
Pure land of.jpg
102vbnn3 n.jpg
10bnhg14 n.jpg

How to Do Things with Candrakirti: A Comparative Study in Anti-skepticism
Arnold, Dan

Philosophy East and West v.51 p.247-279
University Press of Hawaii

Honolulu, HI [US] (


Two strikingly similar critiques of epistemological foundationalism are examined: J.L. Austin's critique of A. J. Ayer in the former's Sense and Sensibilia, and part of Candrakīrti's critique of Dignāga in the first chapter of the Prasannapada. With respect to Austin, it is argued that his writings on epistemology in fact relate quite closely to his better-known philosophy of speech acts, and that the appeal to ordinary language is part of a transcendental argument against the possibility of radical skepticism.

Full Text:

If we had anything at all you could call certainty, then it would have to be based on a reliable warrant, or not based on a reliable warrant. But we don't! How so? If there were the possibility of doubt with respect to this, there could be a certainty opposed to it and dependent upon it. But when we don't have any doubt in the first place, then how could there be a certainty that is or is not opposed to it? Since there's no relation to another term, it would be like [demanding reliable warrants for showing] the length of a donkey's horn. And when, in this way, there's no certainty, then we will imagine reliable warrants for the sake of proving what?[1]

In this way, the Indian Buddhist philosopher Candrakīrti (fl. ca. 650 C.E.) begins a very interesting critique of another Buddhist philosophical project, one that could plausibly be characterized (with due qualification and circumspection) as foundationalist. Specifically, Candrakīrti's statement here initiates his response to an imagined interlocutor whose position seems very much like that of Dignāga (fl. ca. 500 C.E.), the progenitor of an Indian Buddhist philosophical tradition that Western scholars have sometimes simply dubbed that of the "Buddhist Epistemologists." [2] Candrakīrti deploys this imagined interlocutor to challenge him to show how and why he is justified in his claims regarding the ultimately "empty" character of phenomena. That is, the Buddhist Epistemologist, as an important player in the then burgeoning Indian debate on prāmāṇyan ("epistemic reliability"), wants to know what pramāṇas ¡X what valid cognitive instruments, what reliable warrants¡Xprovide the epistemological foundations for Candrakīrti's position. As is suggested by this beginning to Candrakīrti's long and complex response to this challenge, Candrakīrti goes on to argue that if the questions and the phenomena under consideration are properly understood, then the epistemologist's question ought not even to arise. Indeed, Candrakīrti will argue that the epistemologist's imagined requirements are themselves evidence of what is precisely the problem to be overcome. In particular, Candrakīrti will argue that the Buddhist Epistemologist's demand for reliable warrants is a nonstarter, since the doubt that he thinks must be met cannot even arise in the first place if meaningful discourse is to be possible.

I would like to suggest that Candrakīrti's argument here has some striking affinities with some of the characteristic arguments of J. L. Austin. Given Candrakīrti's frequent appeal to what could reasonably be translated as "ordinary language," [3] this is not surprising, and affinities with Austin have, no doubt, been remarked on before. But in the present essay I would like to attend in particular to one section of Candrakīrti's response, and to suggest that here the affinities with Austin are particularly with the arguments developed in Austin's Sense and Sensibilia. The latter, it seems fair to say, is one of the more seldom-studied and less influential of Austin's works. This is, perhaps, not surprising, given that its concern with epistemological questions may make this work seem irrelevant to the kind of "speech act" theory that is thought by many to be Austin's chief contribution.

The neglect of Austin's epistemological writings is, however, regrettable-and not least because, as I hope to show, Sense and Sensibilia in fact relates quite closely to the works that might be thought to be more characteristic of Austin's larger project. Indeed, it seems to me that Austin's Sense and Sensibilia develops a position with respect to epistemology that relates quite naturally to what might be compartmentalized as his "philosophy of speech acts." Moreover, the converse seems to me also to apply; that is, it seems to me that the linguistic problem that Austin diagnoses (and treats) at greater length in the more well known How to Do Things with Words logically relates to an internalist, foundationalist epistemology precisely such as Austin rejects in Sense and Sensibilia (specifically, that of A. J. Ayer). I hope, then, to show that there is a very close connection between what Austin diagnoses as the "descriptive fallacy" and what, in the context of epistemologly, we might characterize as the subjectivist or internalist fallacy. Thus, I will suggest that A. J. Ayer can think a project such as his is required in the first place only because Ayer has allowed himself to be misled by something like Austin's descriptive fallacy, of which Ayer's peculiarly narrow demand for verification is a paradigm case. And Ayer's version of Austin's descriptive fallacy is closely tied to the kind of skepticism that gives rise to such doubts as Ayer thinks must be displaced by warranted certainty. But, as suggested by our quote from Candrakīrti, if there is no doubt which it is thought must be displaced in the first place, then a foundationalist project such as Ayer's is a nonstarter.

For both Austin and Candrakīrti, then, epistemological foundationalism has developed as a set of answers to questions that ought not to have been asked in the first place, and the epistemological project is superfluous once this is seen. Moreover, for both Austin and Candrakīrti, the fact that the questions ought not to have arisen in the first place can be seen from attention to actual use of language (actual use being, on Austin's reading, far more complex than is reflected in the abstract notion of purely "descriptive" statements). I would like to elaborate on these important similarities between Austin and Candrakīrti particularly in the hope of achieving the best "rational reconstruction" of the latter's argument, which is often sufficiently dense and elliptical that its deep affinities with Austin's elegant arguments could easily be overlooked. In the present essay, then, I would like not only to show that Austin's writings on epistemology are in fact quite closely related to his more well known work on speech acts, but also that Candrakīrti can plausibly be read as making an argument that deploys some precisely similar moves. The comparison with Candrakīrti can in turn, I think, help to clarify an important (but largely undeveloped) point in Austin, that is, the extent to which Austin's arguments against radical skepticism might plausibly be characterized as transcendental arguments.

I will elaborate on this point in due course. Briefly, however, the point is that, in the arguments of both Austin and Candrakīrti, there is a claim that the radical skeptic's doubt not only ought not to arise but, in an important sense, cannot arise; for the relative absence of radical doubt is a condition of the possibility of meaningful discourse in the first place. This point, though implicit in Austin's work, seems to me to be more clearly thematized in Candrakīrti. My comparison of Austin and Candrakīrti, then, is designed to elucidate the claim that Candrakīrti (and like him, Austin contra Ayer) makes the following argument: Dignāga's account of pratyak ṣa (perception) can only be the preferred account if most people are wrong when they talk about pratyak ṣa most of the time; but in fact, people can't be wrong about that most of the time, since successful discursive transactions (lokavyavahāra) depend on people being right about conventions; and we know that successful vyavahāra is possible, since that is what we're engaged in! Thus Candrakīrti can reasonably conclude that it is precisely his interlocutor's demand for justification that is unreasonable.

I will conclude, though, by noting that the comparison with Austin cannot be undertaken without at least acknowledging an important difference that, in the context of Candrakīrti's argument, becomes impossible to avoid. Specifically, I must raise and at least briefly explore the question of what difference it makes that Candrakīrti's appeal to ordinary language comes in the context of a characteristically Buddhist "two truths" hermeneutic. That is, I would like to ask how Candrakīrti's understanding and use of ordinary language differs from Austin's specifically to the extent that Candrakīrti deploys this notion as in some sense complementary to "ultimate truth." I will conclude by asking whether or not Austin's philosophy might leave room for something like this move of Candrakīrti's, and whether Candrakīrti's argument might benefit from revision in light of Austin.

Austin versus Ayer.- Ordinary Language and the Problem with Epistemology

To begin with, let us consider how there can be a close connection between the problem that Austin diagnoses as the descriptive fallacy and the internalist, foundationalist epistemology that is Austin's target in Sense and Sensibilia. In fact, I think the relation is sufficiently strong that we can use Austin's notion of the descriptive fallacy to sketch something of the story of how epistemology came, over the course of a few centuries, to be the central problematic in philosophy. Austin's well-known (although, no doubt, much abused) notion of "performative utterances" finds its first expression in his 1946 essay "Other Minds." In that essay, Austin flirts with the point that the expression "I know.. ." is something like a performative utterance, a phrase that has more in common with expressions like "I promise ..." than with expressions like "I believe ...."[4] Thus, his approach to the question of how we "know" other minds is to examine our uses of the expression "I know! in order to see what it is we are doing when we use this expression. His principal target in this examination is something like the Cartesian use of the expression, according to which knowing consists in indubitable cognition, that is, such that "If I know, I can't be wrong."[5] In this account of knowing, the "problem of other minds" arises as peculiarly intractable; for we seem to have privileged access to our own minds, and our position thus seems to be such that we can't be wrong about the contents of our own minds, in a way that we can never realize with respect to other minds.

Austin's response to this conception of the problem is to ask: do we use the word "know" in this way? Can it really be said that what we are doing when we use the word is claiming that, in the cases in which we use it, we can't be wrong? It is here that Austin introduces the notion that "knowing" is more like "promising" than like "believing" (pp. 98-101). That is, just as in the case of promising, "saying 'I know' is taking a new plunge. But it is not saying 'I have performed a specially striking feat of cognition, superior, in the same scale as believing and being sure, even to being merely quite sure': for there is nothing in that scale superior to being quite sure" (p. 99). In other words, "knowing" is not simply the consummate or paradigm case of, say, "believing"; rather, to claim to know is to perform an altogether different act, one such that, for example, "I give others my word: I give others my authority for saying that's is P"' (i bid.).

With this early attempt to assimilate the case of knowing to the class of what he will call "performative utterances," Austin particularly intends to refute the Cartesian understanding of "knowing" specifically insofar as the latter reflects the descriptive fallacy, which notion also makes its appearance in "Other Minds" (p. 103). This notion, along with the contrasting notion of performative utterance, is further developed in Austin's later essay "Performative Utterances," in which much of the ground is laid for How to Do Things with Words. In "Performative Utterances," it becomes clear that the descriptive fallacy consists in taking all utterances as bivalent propositions (or, to use Austin's term, "statements"). Thus, as he says in framing the essay, "We have not got to go very far back in the history of philosophy to find philosophers assuming more or less as a matter of course that the sole business, the sole interesting business, of any utterance¡Xthat is, of anything we say¡Xis to be true or at least false" (p. 233). Austin characterizes this as the descriptive fallacy because, according to this view, making an utterance can be seen always to consist of giving a report or a description of the conditions that must obtain in order for the utterance to be true. It is against this view that Austin will go on to urge that "It he total speech act in the total speech situation is the only actual phenomenon which, in the last resort, we are engaged in elucidating."[6] Among the upshots of this holistic approach is that it "certainly will not lead to a simple distinction of 'true' and 'false'; nor will it lead to a distinction of statements from the rest, for stating is only one among very numerous speech acts of the illocutionary class"[7]

But, specifically with respect to the question of epistemology, it is important to understand how the descriptive fallacy, as characterized by Austin, relates to the internalist epistemology that is definitively characteristic of the foundationalism of, say, A. J. Ayer. Thus, in arguing that we must not be understood, by our use of explicit performatives like "promise," as making descriptive statements, Austin elaborates that "the one thing we must not suppose is that what is needed in addition to the saying of the words in such cases is the performance of some internal spiritual act, of which the words then are to be the report" (p. 236). Here, it becomes particularly clear that the descriptive fallacy, of which Austin's project is a sustained critique, in fact fits very closely with the internalism of foundationalist epistemologies.

Consider, for example, how Austin's paradigmatically performative utterances fare in verificationist accounts of meaning (and it is, of course, a verificationist project that is served by, or expressed in, the foundationalism of Ayer). According to the verificationist's criterion, determining the truth of a statement consists of finding the observable fact that must be the case in order for it to be true. Thus, to make an utterance like "I promise" a candidate for status as a true statement, we first put the expression in truth-functional terms: "'I promise' is true if and only if I promise." Then, we must find the act of "promising," which is what makes the statement true.[8] And one thing we can imagine answering to this description is some "internal spiritual act," of which the statement is then a report or description. As Austin says, in explaining how the verificationist criterion is thus invoked to endow utterances with the only kind of "seriousness" that the verificationists allow (i.e., the kind that goes with being either true or false), "we are apt to have a feeling that their being serious consists in their being uttered as (merely) the outward and visible sign, for convenience or other record or for information, of an inward and spiritual act: from which it is but a short step to go on to believe or to assume without realizing that for many purposes the outward utterance is a description, true or false, of the occurrence of the inward performance."[9] It is, then, precisely the verificationist criterion (i.e., in Austin's terms, the descriptive fallacy) that encourages what we might call the internalist or subjectivist fallacy.

And this is precisely as we should expect, given that it is perhaps definitively characteristic of foundationalism to presuppose an internalist epistemology. What characterizes such an epistemology is the claim that "warrant" or "justification" will always be available (at least in principle, according to weak versions of this) through introspection, that is, that they are "internal" to those who have the beliefs in question.[10] This is, in effect, the "KK thesis," that is, the claim that knowing can only be said to take place when a subject knows that he or she knows; and it is precisely such second-order justification, such warrantable certainty, that the foundationalist desires. Indeed, this is a good way to express the understanding of "knowing," which the tradition stemming from Descartes ultimately ends up explaining (and which Austin convincingly shows is not, in fact, what is ordinarily in play when we use the word "know"). That is, what is thought by many philosophers uniquely to warrant our use of the word "know" is the kind of second-order justification that obtains only after one has asked a second-order question such as "Might I have reason to doubt this?" In Ayer's account, what can guarantee a negative answer to this question is the fact that what one really "knows" is internal sense-data, to which, by definition, knowers have privileged access. Thus, the statement "I know X" is true in such an account, just in case the statement accurately describes some inward sense datum, which, insofar as it is available to introspection, cannot (according to this view) be doubted.[11]

It is, then, a small step from the descriptive fallacy to foundationalist accounts of knowledge and perception more generally. Successfully countering the descriptive fallacy may thus forestall the development of foundationalist epistemologies. In hopes of this, Austin proceeds by attending to our ordinary usage of language, to show something of the range of things that must already be in play there for communicative practices (including that of asking epistemological questions!) even to be possible in the first place. His point, then, is to show that once we appreciate the extent to which our ordinary use of language already accounts for many of the cases that philosophers take to be problematic, the force of their questions often vanishes. Thus, when we turn to Austin's most sustained account of specifically epistemological questions (i.e., that of Sense and Sensibilia), we find that his point is almost exactly like the one made in our quote from Candrakīrti, to wit: if we have no doubt in the first place, then how can we be thought to require the kind of certainty that is the opposite of that?

Austin thus begins by suggesting that, in fact, the doubts that Ayer thinks must be addressed can only arise if we ignore ordinary linguistic usage¡Xand this despite Ayer's contention that it is precisely our ordinary practices that he is explaining.[12] Thus, with respect to a characteristic passage from Ayer, Austin asks:

[I]s it not rather delicately hinted in this passage that the plain man is really a bit naive? It "does not normally occur" to him that his belief in "the existence of material things" needs justifying¡Xbut perhaps it ought to occur to him. He has "no doubt whatsoever" that he really perceives chairs and tables¡Xbut perhaps he ought to have a doubt or two and not be so easily "satisfied".... Though ostensibly the plain man's position is here just being described, a little quiet undermining is already being effected by these turns of phrase.[13]

Bringing to mind our quote from Candrakīrti, Austin continues: "But, perhaps more importantly, it is also implied, even taken for granted, that there is room for doubt and suspicion, whether or not the plain man feels any.... But in fact the plain man would regard doubt in such a case, not as far-fetched or over-refined or somehow unpractical, but as plain nonsense; he would say, quite correctly, 'Well, if that's not seeing a real chair then I don't know what is.'"[14] In this way, Austin suggests that Ayer constructs his whole epistemological apparatus in hopes of warranting the kind of certainty that, in the ordinary usage of such terms as "knowing," no one ever claims to have in the first place; and this kind of certainty seems to be required for Ayer because he thinks that there is a threat posed by the ever-present possibility of unremitting skepticism.

But the kind of skepticism that Ayer thus thinks must be answered is not only not attested in ordinary linguistic usage; more importantly, the successful functioning of linguistic usage makes skepticism of the degree imagined by Ayer positively impossible. Or, more precisely, the successful functioning of language doesn't so much make this impossible as it presupposes its impossibility, for the functioning of linguistic usage already presupposes our assent to certain minimal conditions. This point, it seems to me, is rather underdeveloped by Austin, but there is some warrant for it in his work. It is important that we develop something like this (largely implicit) point, since it seems to me to be an important part of the response to what is surely a typical objection to Austin's appeal to ordinary language here (and, hence, to his whole procedure in general). This is the objection that ordinary language cannot be trusted as a guide to philosophical "truth," since, as we're surely all aware, all of us "ordinary people" get things wrong all the time; surely, this objection goes, what the philosopher is after is some rules for learning how precisely not to be "ordinary," how not to get things wrong in the ways in which all of us so often do.

To this kind of objection, Austin has a compelling rejoinder: it is not Austin's claim that attention to ordinary language necessarily tells us what is true, but that such attention tells us what is and what is not really in need of explanation. Thus, if a philosophical account of, say, knowing or perceiving does not tell us anything with respect to the kinds of situations in which we use such words, then it is not "knowing" or "perceiving" that is being explained. Austin simply claims, that is, that attention to ordinary linguistic usage ought to constrain our focus to such issues as are really in question. With respect to philosophical accounts that are not so constrained, Austin can ask, "How could anything be a question of truth or falsehood, if anyone can always say whatever he likes?" (p. 60). In other words, how could we ever be in a position to say whether Ayer's account of, say, "perceiving" is adequate, if what he is in fact explaining is something other than what we mean by "perceiving"? It seems that the most we could ever grant of such an account is that, say, this is a good account of what Ayer means when he uses the word "perceiving." But surely Ayer would agree that we're after something more than that.

According to Austin's reading, then, Ayer's whole procedure depends precisely upon the failure to notice what, according to our ordinary linguistic usage, is really in question. Moreover, the problem is not simply that Ayer thus ends up explaining something other than what he takes himself to be explaining; more importantly, Ayer is thus setting out in response to a doubt that not only ought not to have arisen, but which, if the issues involved were properly understood, can't have arisen. Thus, for example, Ayer begins by surveying some of the sorts of things we take to be self-evident (e.g., that the "real shape" of a penny remains the same regardless of the perspective from which one sees it). Austin characterizes Ayer's modus operandi thus: "These 'assumptions,' Ayer would presumably grant, look plausible enough; but why, he now says, shouldn't we just try denying them, all the same?" With respect to such a procedure, Austin says, "If we allow ourselves this degree of insouciant latitude, surely we shall be able to deal¡Xin a way, of course¡Xwith absolutely anything. But is there not something wrong with a solution on these lines?" (Sense and Sensibilia, p. 58). What is wrong with this way of proceeding is that it depends on the possibility of our entertaining doubt with respect to the kinds of things upon which successful linguistic practices (including that of "entertaining doubt"!) depend in the first place.

This is, it seems to me, the point that Austin makes (but leaves undeveloped) at the end of the essay "Other Minds"¡Xthat is, the transcendental argument that I have been suggesting Austin can make in response to the objection above to appeal to ordinary language. Concluding "Other Minds," Austin returns from his linguistic forays and addresses, once again, the essay's ostensible question thus: "It seems ... that believing in other persons, in authority and testimony, is an essential part of the act of communicating, an act which we all constantly perform.... But there is no 'justification' for our doing [these things] as such" (p. 115). Both here and in Sense and Sensibilia, then, what Austin is after is a suggestion of the impossibility of the kind of radical skepticism that Ayer takes to raise the doubts that he then thinks are what we must address¡Xfor in fact, our language can't even work unless such radical doubts are impossible, and it has built into its workings myriad ways of accommodating this fact. As Austin reminds us in Sense and Sensibilia, "it is important to remember that talk of deception only makes sense against a background of general non-deception" (p. 11). Here, I think, is where Austin's argument might be characterized as a transcendental argument. That is, Austin starts from the observed fact that we successfully use language all the time; he then suggests that a condition of the possibility of this is, for example, our "believing in other persons," and so forth.[15] Thus, by characterizing Austin's as a transcendental argument, I mean to call attention to the fact that it is an argument from some state of affairs that noncontroversially obtains (namely, there is meaningful discourse) to the more controversial point (namely, radical doubt is not possible) that is adduced as a condition of the possibility of the former.[16]

We will return to this point with our consideration of Candrakīrti. For now, let it suffice to say that Austin persuasively shows Ayer's procedure to depend on his entertaining doubt with respect to the kinds of things that, Austin thinks, our successful use of language will not really allow us to doubt. Given this characterization of Ayer's modus operandi, let us now consider one particularly important example of how Ayer deploys it, in order to see precisely how Ayer's argument in this way addresses a spurious doubt. The example in question pertains to Ayer's appeal to different "senses" of the word perceive (and related words, such as "see"). In a key passage, Austin addresses Ayer's move here:

Ayer says: "If I say that I am seeing a stick which looks crooked [i.e., because it is partially immersed in water), I do not imply that anything really is crooked." Now this is quite true; but what does it show? It is evidently meant to show that there is a sense of "see" in which to say that [when] something is seen [it] does not entail saying "that it exists and that something really has the character that the object appears to have." But the example surely does not show this at all. All that it shows is that the complete utterance "I see a stick which looks crooked" does not entail that anything really is crooked. That this is so in virtue of the sense in which "see" is here used is an additional step, for which no justification is given. And in fact, when one comes to think of it, this step is not only undefended, but pretty certainly wrong. For if one had to pick on some part of the utterance as that in virtue of which it doesn't entail that anything really is crooked! , surely the phrase "which looks crooked" would be the likeliest candidate. (Sense and Sensibilia, p. 88)

This move, as characterized and addressed by Austin, should remind us again of Austin's characterization of the descriptive fallacy; for what Ayer wants here is to be able to "read off," just from the form of the words, some particular "sense" and, hence, some fact of which the statement (here, a statement uttered in the context of some visual cognition) is straightforwardly a description.

In other words, Ayer wants some formal criterion to which he can appeal to explain, simply by virtue of the form of the words, what is happening. Against this, Austin urges (though not, here, in so many words) that it is "the total speech act in the total speech situation" to which we must attend if we are to understand what action is being performed here. And it seems clear that in this and other examples, the "total speech act in the total speech situation" is highly complex, with several variables required to give a sense of what is being claimed by an utterance such as "I see a stick that looks crooked." This is, Austin says, exactly as we should expect, and the word "looks" in this expression is just one of innumerable ways in which we flexibly adjust our utterances to the relevant context. Given such flexibility and complexity, though, Ayer is forced instead to appeal to different "senses" of the verb, if he is to have any chance of telling just from the form of the! words what is going on.

Moreover, what he wants to tell us in this way is the one true account of what is going on, that is, the "real" event that perception, as such, really is. It is, finally, this motive that drives Ayer's argument. According to Austin's reading, then,

part of what has gone wrong is this: observing, perfectly correctly, that the question "What does X perceive?" can be given¡Xnormally at least¡Xmany different answers, and that these different answers may all be correct and therefore compatible, Ayer has jumped to the conclusion that "perceive" must have different "senses"¡Xfor if not, how could different answers to the question all be correct? But the proper explanation of the linguistic facts is not this at all; it is simply that what we "perceive" can be described, identified, classified, characterized, named in many different ways. (Sense and Sensibilia, pp. 97-98)

That is, the proper explanation involves attention to "the total speech act in the total speech situation" (a situation that, in the present example, includes the fact that we can see that the stick is immersed in water, so that this will already be part of our description). But Ayer, according to Austin's exegesis, needs instead to exploit his dubious contention that there are different "senses" of the words in play. What Ayer most needs to justify with this appeal to different "senses" of the word is, finally, the introduction of his own, technical sense of the word. Ayer's new "sense" of the word serves the "wish to produce a species of statement that will be incorrigible; and the real virtue of this invented sense of 'perceive' is that, since what is perceived in this sense [i.e., introspectable "sense-data"] has to exist and has to be as it appears, in saying what I perceive in this sense I can't be wrong."[17]

As we will see quite shortly, this exegesis and characterization of Ayer's argument would serve very well to characterize the epistemological project of Dignāga, the philosopher against whom Candrakīrti directs the eminently Austinian argument to which we will soon be turning. For now, though, let us summarize by noting again how neatly Austin's argument against Ayer lines up with the argument in "Other Minds" and, hence, with the more well known project of How to Do Things with Words . In both "Other Minds" and Sense and Sensibilia, Austin is concerned with forestalling the kind of philosophical problems that can arise only given a failure to appreciate how, in fact, we use words like (respectively) "know" and "perceive." In both cases, the problems in question arise only following the view that we are only entitled to use these words when we are incorrigibly "right" about what we know or perceive; that is, only following the view that "If I know (or perceive), then (by the definitions of these words), I can't be wrong." The desire for such "incorrigibility" is, of course, definitive of "foundationalism." As Austin says, the doctrine that "knowing" only takes place when we can't be wrong amounts to the view that knowledge has foundations:

It is a structure the upper tiers of which are reached by inferences, and the foundations are the data on which these inferences are based.... Thus¡Xso the doctrine runs-the way to identify the upper tiers of the structure of knowledge is to ask whether one might be mistaken, whether there is something that one can doubt; if the answer is Yes, then one is not at the basement. And conversely, it will be characteristic of the data that in their case no doubt is possible, no mistake can be made. So to find the data, the foundations, look for the incorrigible. (Sense and Sensibilia, p. 105)

But in fact, we typically don't use words like "know" or "perceive" in such a way as to suggest that we are claiming to have apprehended something incorrigible. Indeed, we typically use these words precisely in such a way as to allow for the countless things that might go wrong. Moreover, our ordinary linguistic practices could not even get off the ground if what was required of us was that we reserve such words for the kind of incorrigible certainty that, in the nature of the case, could rarely (if ever) obtain, for we would be left with virtually no use for what are, in fact, eminently common words. It is a mistake, then, to suppose that what we are doing when we use these words is always describing or reporting some fact, by virtue of their correspondence with which the utterances are true.

Here again, let us note the connection between the descriptive fallacy and the kind of internalist epistemology that is characteristic of Ayer's project. For in Ayer's account, what ultimately makes our fundamental perceptions incorrigible is the fact that their objects are internal sense-data, that is, mental representations of things. According to this view, then, to say "I see a stick immersed in water" is, ultimately, to describe (or report on) a datum that is incorrigibly available to introspection (i.e., to describe the sense-datum that is before the mind's eye). Thus, it can plausibly be argued that it is precisely by virtue of his succumbing to the descriptive fallacy that Ayer is led to suppose that there must be one true account of what "perceiving" and "knowing" are; that this account must be one such as will reserve the use of these words for cases in which something is incorrigibly apprehended; and that, since we plainly don't use these words in this way most of the time, a technical usage is called for. In other words, this project is wedded to the descriptive fallacy insofar as it is Ayer's verificationist criterion of meaning that leads him to suppose that, as Austin says, "the sole business ... of any utterance ... is to be true or at least false," and that ascertainment with respect to these two (exhaustive) options will always be a function of finding the unique fact that, in any particular case, could alone make an utterance true.

In the case of the question of whether we can "know other minds," then, there will indeed be a real problem if it is thought that all that could count as "knowing" another mind would be knowing it with the kind of incorrigible certainty with which one knows one's own mind (never mind the question of how well one knows one's own mind!). For this is, ipso facto, a case in which one will never have the kind of foundational certainty that is thus claimed for introspection; one can never "describe" or report on another person's mind as one would ostensibly describe one's own conscious states. But this can only be said to be a failure to "know" other minds if we take "knowing" to consist of that kind of introspective event; if we don't, then the question doesn't arise. If, instead, we do not approach knowledge-claims with the expectation that claims to "knowledge" amount to incorrigible claims¡Xthat, rather, what is done is simply the act of "giving others one's authority for saying that 'S is P"'¡Xthen we are in a position to attend to all those features of the "total speech situation" that, as we see all the time, provide all the flexibility we need to discriminate valid or useful actions from others. And since, given this kind of flexibility, what is wanted is never incorrigible certainty, the "problem of other minds" does not arise as a uniquely intractable one.

Indeed, given the kind of flexibility and complexity that are characteristic of "total speech acts," there is a sense in which there really can't be any doubt about, for example, "other minds," for the kinds of communicative practices that allow such a "doubt" to be communicated in the first place already presuppose that such a doubt cannot be real. Thus, there turns out to be a very close connection between, on the one hand, Austin's contention that "the total speech act in the total speech situation is the only actual phenomenon that, in the last resort, we are engaged in elucidating" and, on the other hand, his contention that Ayer's epistemological project derives its impetus from the threat of a doubt that cannot really arise.[18] For, to paraphrase Candrakīrti, "doubt" only makes sense in relation to "certainty"; and when actual linguistic usage does not attest to anyone's claiming certainty with respect to something, then how could we say that anything less than certainty amounts to real doubt?[19]

Candrakīrti versus Dignāga: Conventional Truth and the Problem with Pramāṇas

Having thus recurred to our quote from Candrakīrti, let us now look at Candrakīrti's argument against the Buddhist Epistemologist Dignāga. What is of chief interest to me here is the question of how precisely Candrakīrti's obvious similarities with Austin allow us to map his project onto Austin's, although we will also be compelled to look briefly at Candrakīrti's "two truths" hermeneutic, which adds to ordinary language a complementary term ("ultimate truth") that is clearly without an analogue in Austin's work. The lack of such an analogue will, of course, make it important to see whether the introduction of such a notion results in any important tensions with Austin's aims. This possible difference notwithstanding, the affinities between Austin and Candrakīrti should seem sufficiently close that we will be justified in characterizing them as fundamentally the same kind of arguments; and, as I have been suggesting, an important aspect of their approaches is their deployments of transcendental arguments, a fact that may emerge more clearly when we have considered Candrakīrti. In the present section of this essay, I will proceed by offering my translation of significant sections of Candrakīrti's argument, annotating the argument as we go along, in particular so as to indicate the affinities with Austin's argument. [20] But before we begin, it is important to sketch Dignāga's position, so that we can have a sufficient idea of the position that Candrakīrti is criticizing.[21]

 Given the high degree of similarity that, as I have already suggested in anticipation, obtains between Austin and Candrakīrti, it perhaps ought not to be surprising to see that Dignāga's position in fact has striking similarities with Ayer's (at least as the latter is characterized by Austin). Thus, according to Dignāga, there are only two cognitive instruments that are "reliable warrants" (pramāṇas): inference (anumina) and perception (pratyak ṣa). The former instrument has as its object "universals" (sam,inyalaksanas), while perception has as its object "bare particulars" (svalaksanas). Given the relevant metaphysical presuppositions (chiefly, the Buddhist doctrine of "momentariness"), these bare particulars come to be defined in a peculiarly narrow and technical (and eminently problematic!) sense. For the division into "universals" and "particulars" here comes in the context of an unusually thoroughgoing reductionist project. According to this view, the self is perhaps the paradigm case of a "universal" (i.e., something whose existence is derivative or further reducible; a mere "construct" or "imputation" the reality of which is supervenient upon the particulars into which it can be analyzed). The Buddhist reductionist project with respect to the self was sufficiently thoroughgoing that there was a peculiar reluctance to allow anything even remotely "medium-sized" to remain unanalyzed (since to allow this was to risk allowing that such medium-sized goods could do the conceptual work of a "self," a notion the utility of which Buddhists ardently wished to disallow). Accordingly, the basic or primitive "pieces" into which a derivative existent is ultimately analyzable come to be theorized as vanishingly small. By the time of Dignāga, then, the "bare particulars" (the svalaksanas) that are the only ultimately real objects of "perception" are held to be something like self-identical moments of space-time.

Obviously, the claim that such entities are the objects of perception requires a peculiar conception of "perception." In particular, Dignāga held that perception was defined by its being inherently "free of conceptual activity" (kalpanipodha). Thus, Dignāga could argue that we seem to ourselves to "perceive" medium-sized dry goods ("jars" or "pots" are stock examples in these discussions) only because perception is distorted by the imputations of conceptual activity. That is, the fact that we take ourselves to perceive medium-sized entities is the result of a cognitive mistake. The goal for Buddhist epistemologists like Dignāga, then, becomes that of eliminating the conceptual error that impedes the functioning of our perceptual faculties. The prospects for such elimination are promising, given that theirs can be said to have been an "optimistic epistemology," insofar as the "defilements" that impede this functioning are merely adventitious; if left to its own devices, un! trammeled perception would function passively to register the world as it is in itself, that is, as a world of evanescent, concrete particulars. Even in the context of our distorted apprehension of perceptual events, though, it is our perceptual contact with bare particulars that ultimately warrants our knowledge. Clearly, the spirit of this project is very well captured by Austin's characterization of the foundationalist enterprise of Ayer. Just as in Ayer's case, so, too, Dignāga seems to have been after "a species of statement that will be incorrigible, and the real virtue of this invented sense of 'perceive' is that, since what is perceived in this sense has to exist and has to be as it appears, in saying what I perceive in this sense I can't be wrong."[22]

This, then, is the doctrine of perception against which Candrakīrti is here concerned to argue. He begins his critique thus:

Moreover, because it doesn't include instances of worldly discourse such as "a jar is perceptible," and because of the acceptance of the discourse of ordinary people, (your) definition is too narrow, so it doesn't make sense. [23]

In this way, Candrakīrti makes two points that guide his critique in much the same way as similar points guide Austin's critique of epistemology. First, Candrakīrti thinks that it is the "discourse of ordinary people" that should be accepted, and, with respect to the particular issue in question here, he notes that such discourse typically includes usages like "a jar is perceptible." In other words, the point is that "perception" (or "perceptible"¡Xthe word pratyak ṣa has both senses) is a common word, one that people typically use with respect to medium-sized objects; people do not typically (or even ever!) speak of themselves as "perceiving" bare particulars (svalak ṣa ṇas). Thus, any account of "perception" that disallows usages such as "a jar is perceptible," is, ipso facto, no longer an account of perception!

In regard to this, Candrakīrti ascribes to his interlocutor (Dignāga)[24] a response that should be familiar from Austin's exegesis of Ayer; that is, he has Dignāga appeal to a peculiar sense of the word "perceive." Thus, according to Candrakīrti, Dignāga counters:

Things like color, which are the causal bases of (perceptions of] jars, are perceptible [simply] because of their determinability by the reliable warrant that is perception. And thus, just as, having made figurative reference to the effect with respect to [what is really] the cause, it is taught that "the birth of Buddhas is happiness"-in the same way, having made figurative reference to the cause with respect to [what is really] the effect, a jar, too-which has as its external aspect color and other perceptibles¡Xis indicated as perceptible.[25]

This requires a bit of unpacking. Dignāga here rejoins that he is entitled to use the word "perception" as he does because his is a figurative or secondary sense of the word (upacāra). Moreover, he suggests that such figurative "senses" of the word are already in play in ordinary discourse. He suggests as much by making the observation that people really don't speak of perceiving jars per se; they speak, rather, of perceiving their color, shape, and so forth, that is, the properties that are what (in a strictly causal sense) give rise to our perceptions. And he adds the banal (and tautological) point that these are called "perceptible" simply by virtue of the fact that perception is the faculty that grasps them. The contention, then, is that, in fact, our ordinary usage is already figurative, insofar as we consider "perceptible" what are really the effects or outputs of perception (e.g., "jars"), when what really causes such perceptions is what we should consider to be really "perceptible." To warrant this appeal to a figurative sense, he invokes the example of a well-known Buddhist statement, namely the claim that "the birth of Buddhas is bliss." Here, the point is that, just as a Buddha's birth is called "happiness," even though it is in fact only the cause of happiness, in the same way, jars are called "perceptible" (pratyakṣa) even though they are really only the effects (or outputs) of perception.[26] Of course, the Epistemologist wants this concession since, once it has been granted that it is the specifiable causes of perception that should be called "perceptible," it is a small step to the claim that what ultimately causes our perceptions (and hence what is really "perceptible") is not even colors and shapes, but the Epistemologist's bare particulars (svalakṣaṇāni).[27]

To this move,Candrakīrti replies:

[Appeal to] figurative usage does not make sense with respect to a cognitive object of this kind. For in the world, birth is known as different from happiness. Indeed, since (birth has as its] nature the characteristic of [being] compounded, which fact is the cause of many hundreds of evils, it [i.e., birth) is precisely unhappiness. When, in this way, what is being taught¡Xe.g., "it [i.e., birth) is happiness"¡Xis incoherent, this is the kind of case where figurative usage makes sense. But in the present case¡X"a jar is perceptible"¡Xthere is nothing at all called a jar that is imperceptible, separately apprehended, that, based on figurative usage, could have the quality of perceptibility.[28]

Here, Candrakīrti's response is again much like Austin's.[29] Just as Austin appealed to context to suggest that it is the "total speech act in the total speech situation" that tells us what is intended (and not, e.g., some special "sense" of the words), so, too, Candrakīrti here notes that appeal to figurative usage makes sense only in certain contexts, and that such a context does not obtain in this case. Candrakīrti's remarks here can be related to a rich body of Sanskrit literature on the subject of poetics. According to these Sanskrit conventions, what alerts us to the fact that a figurative usage is in play is the fact that the primary or manifest meaning (mukhyārtha) of an utterance is "blocked" (or contradicted, bādhita) by something else in the utterance that is inconsistent with it.[30]

The point of these Sanskrit theorists is, it seems to me, much like a point Paul Grice has made in the context of his discussions of "conversational implicature." Grice has argued that we must presuppose certain things about the intentions of our interlocutors if we are to stand any chance of recognizing when some implicature (e.g., irony), quite apart from the manifest meaning of the utterance, has been made. Thus, for example, if we presuppose (as Grice thinks we must) that our conversation partners intend to make contributions to the conversation that are "appropriate to the immediate needs at each stage of the transaction," then we are obliged to take any apparent failure to be thus "appropriate" as an indication that what is intended is some implicature.[31] So, too, for Candrakīrti: insofar as it is widely held by Buddhists that birth is the cause only of suffering, Buddhists can be expected to realize that the primary meaning is "blocked" (or, in Grice's terms, that a "conversational maxim" has been violated) when he or she is confronted with the phrase "the birth of Buddhas is happiness." This is the kind of case in which we are entitled to look (indeed, we must look) for some figurative sense. (Here, what is figuratively conveyed is the point that the birth of Buddhas is the cause of happiness for the beings who benefit from their teachings.)

But Candrakīrti denies that the phrase "a jar is perceptible" similarly requires appeal to figurative usage in order for it to make sense. His way of making this point is rather oblique, as he merely alludes to the kind of scenario that would have to obtain in order for it to be the case that the manifest meaning of the phrase "a jar is perceptible" is "blocked" and cannot be allowed. Thus he says that "there is nothing at all called a jar that is imperceptible. ..." The point here is that, in order for it to be the case that the manifest meaning of the phrase is incoherent, it would have to be the case that, for example, there is something that is in fact imperceptible that we call "jars." But in fact, the phrase "a jar is perceptible" presents no problem in the first place; the phrase makes perfectly good sense, and there is no good reason to suppose that the expression requires explanation in terms of figurative usage.

Candrakīrti continues in this vein, considering another sense in which the Epistemologist might try to appeal to a figurative sense of "perceive":

If it is said that perceptibility is figurative because of the nonexistence of a jar apart from [perceptible qualities] like color and so forth, then [appeal to] figurative usage makes even less sense, since there is no basis that is being figuratively described, for the sharpness of a donkey's horn is not spoken of [even] figuratively.[32]

Here, the Epistemologist returns to his best argument for the contention that our ordinary usage is already figurative, that is, insofar as it is really a jar's perceptible properties that we perceive. But Candrakīrti rejoins that this move is even less promising, for, in ordinary language, figurative usage requires that there be some specifiable locus of our figurative ascription, and the Epistemologist's tack here amounts to the claim that jars do not, in fact, exist at all, being (eliminatively) reducible to their parts and/or properties. In that case, though, to speak of a jar's being perceptible is to predicate a property of something altogether nonexistent. It would, to update Candrakīrti's stock Indian example of an uninstantiated entity, be as though we were to make a figurative statement such as "The unicorn is man's best friend," for, in such a case, appeal to figurative usage is no help, since the problem is that unicorns just don't exist

Candrakīrti further suggests that this tack opens up a problematic regress:

Moreover, if it is imagined that a jar, which is included in worldly discourse, has [only] figurative perceptibility since it doesn't exist apart from its color and so forth, then surely, this being the case, because color and so forth, too, do not exist apart from earth and so forth, the [merely] figurative perceptibility of that color and so forth ought also to be imagined.[33]

Here we see a hint of one of the characteristic differences between Candrakīrti's version of the Buddhist reductionist project and that of the Buddhist Epistemologists.

For the latter, entities are reducible to vanishingly small "bare particulars" (svalakṣaṇāni, but these are held to be further irreducible, and thus ultimately to exist in some important way. Indeed, as I suggested in my brief sketch of the Epistemologist's project, it is these that ultimately provide the foundations of our knowledge, insofar as these are the unique objects of the kind of "perception" posited by the Epistemologists. Candrakīrti, in contrast, maintains that the only way to be consistent about the reductionist project is to acknowledge that one can never "reach the bottom"; the whole point just is that there is nothing that is further irreducible. Thus, in further response to the Epistemologist's claim that the phrase "a jar is perceptible" is figurative insofar as jars are reducible, Candrakīrti says, in effect: Fine! But if we thus introduce critical analysis (i.e., the kind of critical analysis that does not obtain at the level of ordinary discourse), then we'll see that there is no irreducible remainder¡Xthat, for example, the perceptible properties of a jar depend, in turn, upon the elements ("earth and so forth") in which they are instantiated. In this way, with regard to the Epistemologist's attempt to argue that perceptions are ultimately caused by irreducible primitives, Candrakīrti can always respond that these will never be irreducible, insofar as they, in turn, must depend on something else. This, then, is Candrakīrti's point when he responds that, if the perceptibility of jars is figurative insofar as jars are reducible to their parts and properties, then the Epistemologist must be prepared to grant that all language is merely figurative in this way, insofar as a jar's parts are, in turn, reducible, and so on, ad infinitum.

This point is amplified when Candrakīrti concludes this section of his argument thus:

Therefore, since your definition doesn't include such worldly usage, [your] definition has insufficient extension. For the perceptibility of jars and so forth and of colors and so forth is not accepted from the point of view of one who knows reality; but in terms of worldly convention, the perceptibility of jars and so forth is precisely to be accepted[34]

This passage amounts to a very succinct statement of just how similar Candrakīrti's approach is to that of Austin and, at the same time, of what the crucial difference is. Thus, Candrakīrti concludes this section of his critique by pointing out that the Epistemologist's definition of "perception" cannot, in fact, account for the ways in which the word is ordinarily used¡Xin which case, insofar as the Epistemologist's definition therefore has "insufficient extension" (i.e., it doesn't extend far enough to cover everything that it purports to explain), it is not really perception that has been explained at all! But then, in what may be Candrakīrti's key departure from Austin's company, he continues that neither medium-sized goods like jars nor the vanishingly small particulars into which such things can be reduced ultimately exist, that is, exist "from the point of view of one who knows reality" (tattvavidapek ṣayā, i.e., of a Buddha). Nevertheless, from the perspective of ordinary worldly usage, it is things like jars that are said to be perceptible, and not, for example, "bare particulars."

We can put this point in terms that will be at once familiar to Austinians and Buddhologists: for Candrakīrti, what is conventionally true is just "conventions." Here, it is important to note that "conventions" does not have the same sense that it would have for Austin or Grice.[35] Rather, "conventional truth" is simply the customary translation for the term (saṃvṛtisatya) that denotes the level of discourse that is complementary to Buddhism's "ultimate truth" (paramārthasatya), and it could just as well be rendered as "provisional" or "mundane truth." But in this case, it is in fact felicitous to retain the customary translation, insofar as it allows us to see the correlation between this "level of truth" and the kinds of epistemic practices (e.g., language usage) that enable us to form beliefs about it (e.g., to talk about it).

Thus, Candrakīrti's point is that the level of conventional truth just consists of the unanalyzed conventions of which our ordinary language usage is one example. Indeed, the notion of "critical analysis" (vicāra) is the pivotal concept here, with conventional truth (and, correspondingly, ordinary language) being defined by its functioning without critical analysis, and ultimate truth being conceived as what remains in the wake of unrelenting critical analysis. But, of course, it is precisely Candrakīrti's point that nothing withstands such analysis. His point about conventions, then, becomes, in a way, the point that the only ultimate truth is that there is no "ultimate truth"; conventions, in the peculiarly Buddhist sense of the term, are all there are. Or, to concentrate on the texts before us and take ordinary language as the paradigm case of such conventions, the ultimate truth is that our ordinary linguistic accounts of things are as "true" as any accounts could be. If, instead, we try to get behind such accounts and explain them in terms of technical categories, we are, ipso facto, no longer remaining at the level of conventional truth. Rather, any attempt at critical analysis of our practices, simply by virtue of its being critical analysis, has, ipso facto, pretensions at arriving at ultimate truth. But since the only ultimate truth is that there is nothing more true than conventions, it won't do to search for something behind them.

This is not an unproblematic view, and it is here (as I will point out shortly) that Candrakīrti most significantly differs from Austin. There is, though, still a parallel with Austin's critique of Ayer: recall that Austin (rightly) notes that "though ostensibly the plain man's position is here [i.e., by Ayer] just being described, a little quiet undermining is already being effected" by Ayer. So, too, in the Buddhist context: since nearly all Buddhists will wish to grant that the ultimate realization of a Buddha vastly exceeds our limited ability to talk about it, Dignāga, being a good Buddhist, will claim that his peculiar doctrine of perception is simply an account of our conventional epistemic practices.[36] But, says Candrakīrti, this can only be true if, in fact, most of what we think of as our "conventional" practices are wrong. [37] It is, then, as a response to Dignāga's claim that his is an account of our conventions that Candrakīrti's remarks here must be understood. When Candrakīrti counters that what is conventionally true is just our conventions, he is saying, in effect, that our conventional practices do not require any explanation or justification (at least not of the sort advanced by Dignāga).

Indeed, the kind of critical scrutiny that seeks such explanation or justification is precisely definitive of the nonconventional. It seems to me, then, that Candrakīrti can plausibly be read as arguing, with Austin, that conventional truths are simply "as much an irreducible part of our experience as, say, giving promises, or playing competitive games, or even sensing coloured patches.... But there is no 'justification' for our doing [these things] as such." Similarly, in an account such as Candrakīrti's, as Jay Garfield has aptly said, "we should seek to explain regularities by reference to their embeddedness in other regularities, and so on. To ask why there are regularities at all, on such a view, would be to ask an incoherent question: the fact of explanatorily useful regularities in nature is what makes explanation and investigation possible in the first place and is not something itself that can be explained."[38] Thus, Candrakīrti's project, as contra Dignāga's, consists of "taking conventions as the foundation of ontology, hence rejecting the very enterprise of a philosophical search for the ontological foundations of convention."[39]

This is, again, the move that I have been suggesting might be characterized as a transcendental argument. Thus, in insisting that "in terms of worldly convention, the perceptibility of jars and so forth is precisely to be accepted," Candrakīrti is saying: what defines the use of phrases like "jars are perceptible" as conventional is simply the fact that such usage is what conventionally takes place. This notion of "worldly convention" (or Austin's "ordinary language"), then, can be taken to refer to the kind of discourse that, as an observed and uncontroversial fact, takes place all the time. And a condition of the possibility of such successful language usage is that such conventions cannot themselves be explained, insofar as we depend on them for any explanation in the first place. Therefore, once critical analysis comes into play, one is no longer entitled to claim that it is such conventional usages that are under explanation; rather, one is now trying to get behind them. And it is just Candrakīrti's point that, if we really try thus to get behind them, we find that there is nothing further that explains them (and certainly not the "bare particulars" of the Epistemologist). For Candrakīrti as for Austin, then, it is our ordinary linguistic usage that should tell us what is and what is not in need of explanation, and, for both thinkers, our ordinary language cannot be in need of explanation (at least, not of the kind of explanation that the Epistemologist demands, which could only show the ordinary usage to be false), since the possibility of any such explanation already presupposes the successful functioning of ordinary language. In both cases, then, the point is basically that if we recognize what must be the case for our ordinary language already to function, then we will see that there is no possibility of the kind of radical skepticism that the Epistemologist thinks must be addressed-no possibility, that is, of any serious doubt about whether, for example, it is really "jars" that we perceive. Again, I suggest that this amounts to a transcendental argument against radical skepticism.

But Candrakīrti perhaps parts company in his additional point that the deployment of critical analysis ipso facto means that one is attending no longer to conventional but to ultimate truth. Candrakīrti's additional move perhaps introduces problems, and it is a complex question what, given his characteristically Buddhist assumption of an "ultimate truth," Candrakīrti hopes to accomplish by such a peculiarly Buddhist appeal to ordinary language. Shortly, we will reflect on whether or not the location of Candrakīrti's ordinary language in the context of Buddhism's two truths represents a fundamental incompatibility with Austin. First, though, let us skip over some of Candrakīrti's further arguments against Dignāga's account of perception, and see how he concludes the whole discussion. Since we left him, Candrakīrti has changed the lines of argument, and has pressed against his interlocutor the kind of etymological argument that figures prominently in Indian philosophy, challenging on various grounds Dignāga's etymological justification for his account. [40] After this, Candrakīrti relates the etymological discussion to the Epistemologist's likely claim that he is simply explaining our conventional usage, concluding the whole argument with a polemical flourish:

This word "perception" is indeed well known in the world; but it is described by us [and not by you] precisely as it is in the world. But if, with contempt for worldly categories as they really are, this etymology [of yours] is being developed, [all that] would be well known is [your] contempt for the expression "well-known"! For what is thus called "perception" would not be such, and there would not be, on the part of one visual awareness, which has as its source one moment of sense faculty, the quality of being a perception, since there would be no object of pervasion; and if there is absence of perception-ness on the part of one, there would be [such absence] on the part of many [instances of awareness).
And because you accept that only that awareness from which conception has been removed has the quality of perception; and because, by virtue of that, nobody's discourse would be meaningful (tena ca lokasya saṃvyavahārābhāvāt, "because of the absence of meaningful discourse on the part of the world"); and because of the desirability of explaining worldly discourse with respect to reliable warrants and cognizables¡X[your] conception of the reliable warrant that is perception becomes quite senseless.[41]

With this conclusion (and particularly the points that I have italicized), we are again on ground familiar to students of Austin (and here, perhaps, in terms not only of the content of the argument but also of the sardonic tone!). Candrakīrti thus reminds us that the word "perception" is one that is taken from ordinary language and the use of which is familiar. Because of this, the Epistemologist's peculiar account of perception (according to which evanescent bare particulars are the "real" objects of perception) can be advanced only and precisely to the extent that such ordinary usage turns out to be wrong. If the Epistemologist's account were true, then ordinary people would always be mistaken in their use of the word, since they typically use it with respect to the apprehension of medium-sized objects. Thus, if the Epistemologist's account were true, what people ordinarily call "perception" would not, in fact, be perception at all. Candrakīrti suggests the implications of this when he adds that all of the many things we typically think of as instances of "perception" (e.g., moments of visual awareness) would not, in fact, qualify as such; despite all of the numerous contexts in which people use the word, and think they know what they mean by it, the Epistemologist's account can only be true if these are wrong. [42]

And, as the second paragraph of my translation suggests, it is not possible that people should thus be wrong all the time. For if they were, then there would be no possibility of the kind of meaningful discourse in which Candrakīrti and his interlocutor are now engaged! In this way, then, Candrakīrti, too, makes a point that, although I think it is largely implicit in Austin, I hope to have shown is nonetheless important to Austin's project. That is, in both the cases of Candrakīrti and Austin, attention to ordinary linguistic usage is meant to suggest the impossibility of the kind of radical skepticism to which epistemology is a response. Thus, in concluding the present argument, Candrakīrti says that the Buddhist Epistemologist's conception (kalpanā) of perception is quite literally senseless (vyarthā). This is because the Epistemologist's account is really an account of a narrowly and peculiarly conceived "sense" of the word "perception," one according to which, for example, "perception" is defined by its being inherently free of conceptual thought. But such an account can only be the preferred account if most people are wrong in their ordinary use of the word; hence, Candrakīrti can say that, by virtue of this peculiar sense of the word, it must be the case that there is a complete "absence of meaningful discourse on the part of the world." That is, acceptance of the Epistemologist's account is tantamount to the conclusion that most of the discourse in the world is not "meaningful."

Candrakīrti adduces this as a manifestly absurd entailment of the Epistemologist's project, giving this fact as the reason for the senselessness of his interlocutor's conception. For in fact, it is easy to see that most of the discourse in the world is meaningful. And how could it be otherwise? If it were otherwise, there would be no possibility of the kind of discourse in which Candrakīrti and his interlocutor (or Austin and Ayer) are engaged. Thus, it is very important for our understanding not only of Candrakīrti, but also of Austin, that Candrakīrti in this way gives the absurd entailment of "absence of meaningful discourse on the part of the world" (lokasya saṃvyavahārābhāvā) as the reason for the senselessness of the epistemological project. For this point can return us once again to a point that I hope to have shown with respect to Austin, that is, that his more well known work on speech acts is, in fact, quite naturally related to his less well known writings on epistemology.[43]

Moreover, the sense in which Austin's work on speech acts relates closely to his work on epistemology is precisely the sense in which Austin's larger argument (like Candrakīrti's) might plausibly be said to deploy a transcendental argument. That is, Austin's larger argument is developed in response to a verificationist criterion of meaning (for Austin, under the description of the "descriptive fallacy"), and such a criterion of meaning, in turn, can be shown to have developed in response to a degree of skepticism that, Austin suggests, cannot seriously be entertained, so that the verificationist's doubt ought not to arise in the first place. Thus (to make more explicit how this might qualify as a "transcendental" argument), Austin (like Candrakīrti) starts from the observed fact that we successfully use language all the time. He then suggests that a condition of the possibility of this is, for example, our "believing in other persons" (so that, e.g., the "problem of other minds" cannot seriously be entertained). As Grice has argued in developing what seems to me to be a similar point, the successful use of language presupposes the kinds of "conversational maxims" that Grice takes to come under the heading of the "Cooperative Principle." For Grice's point is that we must presuppose certain things about the intentions of other persons (e.g., that they will generally wish to make "appropriate" contributions to conversation) if we are to stand any chance of recognizing when, say, a statement has been made ironically. This is, it seems to me, a transcendental argument, insofar as the point is that we necessarily presuppose such things. Thus, any attempt to explain our epistemic practices, insofar as such attempts must make use of the kinds of meaningful discourse that already presuppose such practices, can only get off the ground if the very things it purports to explain do not, in fact, require explanation!

To the extent, then, that the project takes itself as precisely motivated by such doubt, it is simply a nonstarter. It is in this sense that the quote introducing this essay should be taken to suggest that Candrakīrti thinks not only that he has no doubt that requires displacement by the Epistemologist's brand of certainty, but that, in an important sense, there can be no doubt about this; and if there is no room for doubt, then there is no room for the kind of certainty that is in need of a warrant. Recall, then, that Candrakīrti says, "if there were the possibility of doubt with respect to this, there could be a certainty opposed to it and dependent upon it. But when we don't have any doubt in the first place, then how could there be a certainty that is or is not opposed to it?" As the conclusion to his argument makes clear, the point is that there only could be such doubt if it turned out that there was a complete "absence of meaningful discourse on the part of the world." But in fact, we know that there is not an absence of such meaningful discourse, and we know this because we are talking about it! Our successful use of ordinary language, then, shows that we must already presuppose the kinds of epistemic practices that the epistemologist takes to require explanation; and, necessarily presupposing them, we can only incoherently suppose that we must at the same time explain them.

Conclusion: Ultimate Truth and the Buddhist Context of Candrakīrti's Argument

Thus, our comparison of Austin and Candrakīrti has, I think, both elucidated Candrakīrti's often elliptical argument and helped to clarify the extent to which Austin's project is in fact rather closely related to his writings on epistemology¡Xand in particular to a critique of the kind of radical skepticism to which Austin takes traditional epistemology to be a response. This is so insofar as Austin's, like Candrakīrti's, involves something like a transcendental argument. But we must conclude by returning to the question that we bracketed above, namely the question of whether Candrakīrti's appeal to ordinary language is in tension with Austin's to the extent that the former stands as complementary to the category of "ultimate truth." There seems to me no prima facie reason why the latter notion should render Candrakīrti's use of ordinary language fundamentally divergent from Austin's. This is especially so given that there is an important sense in which the notion of "ultimate truth" is something of a cipher; that is, on Candrakīrti's own account, "ultimate truth" is virtually defined by the fact that it can't be given any specific content.[44] And perhaps it is precisely insofar as this concept cannot be given any specifiable content that there is no obvious incompatibility with Austin (although Austin might quite reasonably ask, in this case, what need do we have for this place-holder).[45]

But in fact, in Candrakīrti's account, there is a sense in which "ultimate truth" can be given some content after all. That is, the ultimate truth is that there is no "ultimate truth," at least not of the sort that the Epistemologist demands. The ultimate truth, then, is just the abstract state of affairs of there being no irreducible existents such as could warrant our knowledge-claims. "Ultimate truth," then, does not (as it might for the Buddhist Epistemologist) consist of a set containing specifiable, enumerable entities; rather, it denotes simply the fact of there being no such set. Austin would not, I think, have any problem with this notion. But he might still ask whether there could be any point in calling this state of affairs "ultimate truth"¡Xthat is, whether there could possibly be any practical advantage to extolling this abstract state of affairs as the "ultimate truth." This question is an important one, and it is not clear that Candrakīrti could give a good! answer to it. It is not clear to me, anyway, how it is that Candrakīrti could appeal to his notion of "ultimate truth" to commend, for example, any particular religious practices.

Moreover, even if we bracket this specifically soteriological question, there is still a problem with Candrakīrti's position simply from the worldly perspective. Austin and Candrakīrti agree, as I hope to have shown, that ordinary language should inform our conclusions as to what is and what is not really in need of explanation, but Candrakīrti seems to make the further move of claiming that any critical or analytic explanation is ruled out by this recognition. Austin, I think, would not go this far, and he would be right to protest Candrakīrti's insistence on this additional step. For Austin, the point of attention to ordinary language is not to maintain that there should be no technical languages of the sort that might advance explanations. Rather, his point is simply that, to the extent that the latter purport to explain phenomena that are familiarly attested in ordinary usage, it is necessary to resort to the ordinary usages to make sure that we are explaining what we claim to be; otherwise, like Ayer, we risk explaining something entirely other than, say, "knowledge" or "perception," despite our claims to be accounting precisely for these. Moreover, as I hope to have shown, some of the questions that philosophers have taken to be significant cannot, in fact, arise if the ordinary usages are even to be possible.

However: this does not mean that no questions can arise with respect, for example, to our epistemic practices, and it is not Austin's point to rule out all critical or technical accounts of these; it is only to rule out those accounts that smuggle in something else under the same description. Candrakīrti, on the other hand, comes perilously close to suggesting that he does think that all such critical analysis is ruled out if we are really to remain at the level of conventional truth. Not only, then, does his position make it difficult to imagine how we could commend any specific (e.g., Buddhist!) soteriological program, but in fact, it makes it seem impossible to do so; for don't countless other practices similarly qualify as "conventional," "well-known in the world," "ordinary," and so forth? Candrakīrti's account thus seems to place him on a slippery slope that leads to antinomianism or relativism; for his refusal to countenance any critical analysis of the conventiona. l makes it seem that we should simply always defer to what is conventional. And surely there are, inter alia, many non-Buddhist religious practices that could qualify as "conventional"[46]

All that Austin has shown, in contrast, is that the kind of radical skepticism that gives rise to the verificationist criterion of meaning cannot, in fact, seriously be entertained, insofar as it brings into doubt precisely the kinds of epistemic practices that we necessarily presuppose if we are to have any chance of meaningful discourse in the first place.[47] But he does not thereby rule out critical attention to doubts with respect to everything attested in ordinary language. Candrakīrti and Austin are, I think, both persuasive in the quasi-transcendental argument that shows that the radical skeptic's doubt does not, in fact, need to be addressed, so that the epistemologist's requirement for radical certainty becomes spurious. But I think, further, that Austin is also correct in holding that this conclusion does not rule out all critical analysis of our ordinary linguistic and epistemic practices, for Austin's point, again, is simply that ordinary language can tell us w! hat is and what is not really in need of explanation. Ordinary language does not, in telling us this, tell us that nothing can be in need of explanation. The best rational reconstruction of Candrakīrti's argument, then, would have Candrakīrti similarly moderate his point about the primacy of ordinary language. Unless he does so, he not only stands little chance of commending specifically Buddhist practice but also makes the mistake of thinking that ordinary language alone expresses (conventional) truth. As Austin recognized, though, the point should be not that ordinary language commits us to any particular account of truth, but only that it can best tell us what it is whose truth might really be in doubt.


  1. Candrakīrti, Prasannapadā 56.4-57.1: Yadi kaścin niścayo nāma-asmākam syāt, sa pramā ṇajo vā syād apramā ṇajo vā. Na tv asti. Kiṃ kāra ṇaṃ? Ihaaniścayasaṃbhave sati syāt tatpratipak ṣas tadapek ṣo niścayaḥ. Yadā tv aniścaya eva tāvad asmākaṃ nāsti, tadā kutas tadviruddhāviruddho niścayaḥ syat saṃbandhyantaranirapekṣatvātkharaviṣāṇasya hrasvadīrghatāvat. Yadā ca-evaṃ niścayasya-abhāvaḥ, tadā kasya prasiddhyarthaṃ pramānāni parikalpayiṣyāmaḥ? References, by page and line numbers, are to the Bibliotheca Buddhica edition of Louis de La Valle Poussin: Mūlamadhyamakakārikās (Mādhyamikasūtras) de Nāgārjuna, avec la Prasannapadā Commentaire de Candrakīrti, Bibliotheca Buddhica IV (1903-1913; reprint, Osnabrück: Biblio Verlag, 1970). All translations are my own.
  2. Note, though, that the Tibetan dGe-lugs-pa tradition takes Candrakīrti here to be arguing against Bhāvaviveka. See, for example, M. D. Eckel, "Bhāvaviveka and the early Mādhyamika Theories of Language," Philosophy East and West 28 (3) (1978): 323-337. While it is certainly the case that Candrakīrti dedicates much of the first chapter of his Prasannapadā to a critique of Bhāvaviveka, it seems clear that the argument I will consider occurs within a section (pp. 55.11-75.13) that addresses a position much like that of Dignāga. That is how Siderits (see note 20 below) and Hattori (note 21 below) read this section, and there seem to me to be several smoking guns that lead back to Dignāga.
  3. Candrakīrti makes use of a cluster of terms that convey notions very similar to this, chiefly lokavyavahāra ("worldly transactions," "worldly custom," "business as usual") and the corresponding level of saṃvṛtisatya ("conventional" or "provisional truth").
  4. The expression "I promise ..." will, of course, become Austin's paradigm case of a performative utterance.
  5. J. L. Austin, "Other Minds," reprinted in Philosophical Papers, 3d ed., ed. J. 0. Urmson and G. J. Warnock (Oxford: Oxford University Press, 1979), p. 79; cf. pp. 97ff. The essay "Performative Utterances" is also reprinted in this volume (pp. 233-252). Henceforth, references to both of these essays will be to page numbers in this edition, and will be given parenthetically in the text.
  6. Austin, How to Do Things with Words, 2d ed., ed. J. 0. Urmson and Marina Sbisà (Cambridge: Harvard University Press, 1975), p. 148.
  7. Ibid., p. 147. It is important to note (what is too seldom noted) that Austin here blunts the sharpness of his distinction between "performative" and "constative" utterances, in effect showing this to have been a distinction adopted only for its heuristic value; his point, in the end, was thus not to reify a new category of "performative utterances" but rather to call for an appreciation of the inherent complexity of all "speech acts." Moreover, this aspect of his project was clear even in the earlier essay "Performative Utterances" (see, especially, pp. 249-252), a fact that makes the failure to appreciate the fluidity of Austin's categories all the more regrettable. The tendency to focus exclusively on Austin's notion of "performative utterances" seems to me to be largely influenced by Derrida's reading of Austin; see Jacques Derrida's essay "Signature Event Context," reprinted in Margins of Philosophy, trans. Alan Bass (Chicago: University of Chicago Press, 1982), pp. 307-330. For a useful critique of Derrida's reading (one that particularly emphasizes the importance of understanding Austin's project in its relevant context, that is, as a reaction against logical positivism), see Stanley Cavell, "What Did Derrida Want of Austin?" in Cavell's Philosophical Passages: Wittgenstein, Emerson, Austin, Derrida (Blackwell, 1995), pp. 42-65, and especially p. 50.
  8. Cf. Austin, "Other Minds," p. 109: "It is as silly to ask 'What, really, is the anger itself?' as to attempt to fine down 'the disease' to some one chosen item ('the functional disorder').... There is no call to say that 'that' ('the feeling') is the anger."
  9. Austin, How to Do Things with Words, p. 9.
  10. For a very useful account of "internalism" and "externalism" in epistemology, together with a most interesting application of the distinction, see Paul Griffiths, "The Limits of Criticism," in Jamie Hubbard and Paul L. Swanson, eds., Prunng the Bodhi Tree: The Storm over Critical Buddhism (Honolulu: University of Hawaii Press, 1997), pp. 145-160. For thoughts on the extent to which this distinction can (or cannot) be mapped onto the Indian Buddhist debate in which Candrakīrti was participating, see note 43 below.
  11. In characteristic fashion, Austin uses analysis of ordinary language to make something like this point about the introspection of inward sense-data; see, for example, Austin, "Other Minds," pp. 96-97, on the problems that follow from taking "know" always to have a direct object. Here, Austin urges that, with respect to an utterance such as "I know what I am feeling," "this does not mean that there is something that I am both knowing and feeling."
  12. As we shall see, Candrakīrti allows that it is similarly characteristic of Dignāga's approach to argue that he is simply explaining our ordinary, conventional usage (see notes 36 and 42 below). Given the peculiarly technical character of Dignāga's system, it is perhaps appropriate that Candrakīrti will respond to this contention with ridicule (see, especially, note 41 below).
  13. J. L. Austin, Sense and Sensibilia, ed. G. j. Warnock (Oxford: Oxford University Press, 1962), p. 9. Parenthetical page references in the text are to this edition.
  14. Ibid., p. 10. Cf. also Austin's point that our claim to know, say, that a goldfinch is before us is not properly undermined by the (radically skeptical) challenge that we can't be said to have known if it turns out to do "something outrageous" (such as turn into a bowling ball). In such a case, Austin counters, "we don't say we were wrong to say it was a goldfinch, we don't know what to say. Words literally fail us.... It seems a serious mistake to suppose that language (or most language, language about real things) is 'predictive' in such a way that the future can always prove it wrong. What the future can always do, is to make us revise our ideas about goldfinches or real goldfinches or anything else" (Austin, "Other Minds," pp. 88-89).
  15. Something like the same "condition of the possibility" of successful language use is, it seems to me, what Paul Grice thematizes in his notion of "conversational maxims," which come under the heading of Grice's "Cooperative Principle." This represents Grice's attempt to specify the kinds of things we must presuppose about the intentions of other persons if we are to stand any chance of recognizing when, say, a statement has been made ironically (our general success at which is one of the observed facts about our ordinary use of language) (see Grice's "Logic and Conversation," in Studies in the Way of Words [Cambridge: Harvard University Press, 1989], especially pp. 26-31). But, of course, one of the difficult things about such a transcendental argument is that it would be difficult to argue to a conclusion with much content; for it is difficult to specify, in this way, exactly what our language use necessarily presupposes. We will return to all of this later (see, especially, note 47 below).
  16. The notion of transcendental arguments is, of course, particularly closely associated with Kant, whose modus operandi was typically to ask, of something that is noncontroversially the case: how is this possible? Thus, Kant asked, paradigmatically, how experience was possible. In each case, he argued that the phenomenon in question is possible only given some specific condition of its possibility¡Xwith this condition being the controversial matter that Kant had set out to demonstrate (in the case of experience, Kant's list of categories). Since Kant's time, people have understood "transcendental arguments" in many ways, and have argued about whether or not they even represent a logically distinct type of argument. There are, however, several twentieth-century philosophers who have made arguments similar to the one I here attribute to Austin (and, shortly, to Candrakīrti). Thus, for example, P. F. Strawson says of the radical skeptic: "He pretends to accept a conceptual scheme, but at the same time quietly rejects one of the conditions of its employment. Thus his doubts are unreal, not simply because they are logically irresoluble doubts, but because they amount to the rejection of the whole conceptual scheme within which alone such doubts make sense. So, naturally enough, the alternative to doubt which he offers us is the suggestion that we do not really, or should not really, have the conceptual scheme that we do have; that we do not really, or should not really, mean what we think we mean, what we do mean. But this alternative is absurd. For the whole process of reasoning only starts because the scheme is as it is; and we cannot change it even if we would" (P. F. Strawson, Individuals: An Essay in Descriptive Metaphysics [London: Routledge, 1959], p. 35.) In a somewhat similar vein, Donald Davidson has argued: "What makes interpretation possible ... is the fact that we can dismiss a priori the chance of massive error. A theory of interpretation cannot be correct that makes a man assent to very many false sentences: it must generally be the case that a sentence is true when a speaker holds it to be" (Donald Davidson, Inquiries into Truth and Interpretation [Oxford: Clarendon Press, 1984], pp. 168-169). For more on transcendental arguments, see Richard Rorty, "Verification ism and Transcendental Arguments," Nous 5 (1) (1971): 3-14 (as well as the other articles in this Nous issue); Peter Bieri et al., eds., Transcendental Arguments and Science: Essays in Epistemology (Dordrecht: D. Reidel, 1979); Franklin I. Gamwell, The Divine Good: Modern Moral Theory and the Necessity of God (Dallas: Southern Methodist University Press, 1990), pp. 85-126; and, for a specifically comparative use of the category, David Peter Lawrence, Rediscovering God with Transcendental Argument: A Contemporary Interpretation of Monistic Kashmiri Śaiva Philosophy (Albany: State University of New York Press, 1999).
  17. Austin, Sense and Sensibilia, p. 103. For Ayer's response to Austin's critique, see his article "Has Austin Refuted Sense-Data?" in K. T. Fann, ed., Symposium on J. L. Austin (New York: Humanities Press, 1969), pp. 284-308.
  18. Here, as elsewhere in Austin (as, e.g., when he says of some utterance that it is the sort of thing we "cannot" say), the objection will arise: but of course this doubt can arise! In Ayer's work, it just did arise! But I use the expression "cannot really arise" advisedly, for the point is that if communicative practices are to be successful, then such a doubt cannot, as it were, be lived-cannot, that is, be seriously entertained, but only raised as a straw man. Here again, I am hinting at the reasons for considering Austin's as something like a transcendental argument.
  19. For a further development of some arguments in this vein (together with a comprehensive review of the relevant literature), see Michael Williams, Unnatural Doubts: Epistemological Realism and the Basis of Scepticism (Princeton: Princeton University Press, 1996). See, especially, pp. 139-148. Williams' main criticism of Austin's approach seems to concern Austin's failure to account for the fact that there is a context in which the skeptic's questions make sense; specifically, the skeptic can always claim that the "total speech situation" for her enterprise involves the special context of "doing epistemology." Moreover, Austin's peculiarly narrow examples (which involve, e.g., bird identification) don't really do justice to the peculiarly general questions asked by epistemologists (questions, e.g., regarding "our knowledge of other minds"). But Williams ends up crediting Austin with an insight (which, he says, goes rather undeveloped in Austin's hands) much like what Williams, in the end, is really after: "what has now [i.e., in light of Austin] become problematic is the very possibility of representativeness.... Rejecting epistemological realism leaves us with examples of knowledge but no paradigms.... The question to press is not 'Can a claim be both generic and concrete?' but 'Why suppose that a knowledge claim becomes representative simply by being generic (i.e., by involving only the existence of a generic object)?'.... [What I think Austin means to claim is that there are no universally relevant error possibilities" (pp. 167-168). This amounts, it seems to me, to taking back the earlier criticism; for in fact, if it is granted that this is what Austin "means to claim" (and it seems plausible to take it so), then Austin is entitled to use the narrow examples he does, and to rule out the epistemologist's asking of peculiarly general questions; for the whole point just is that no cases of "knowing" can be taken as peculiarly "exemplary! " of what knowing is.
  20. As far as I am aware, the best study of Candrakīrti's sustained argument against Dignāga is that of Mark Siderits, in his "The Madhyamaka Critique of Epistemology II," Journal of Indian Philosophy 9 (1981): 121-160. Siderits treats the whole argument (which, again, spans pp. 55.11-75.13 in La Valle Poussin's edition); in the analysis that follows, I will be concerned chiefly with the section spanning 69.13-75.5.
  21. The locus classicus for Dignāga's epistemology is his Pramāṇasamuccaya, the relevant chapter of which is available in a good English translation by Masaaki Hattori, Dign5ga, On Perception (Cambridge: Harvard University Press, 1968).
  22. Cf. the apt characterization of Dignāga's project by Paul Griffiths (see note 10 above): "No questioning [of perceptual beliefs] is possible or needed: error is not, as Indian logicians would have it, located at the level of pratyakṣa, or perception; the very occurrence of a perceptual act carries with it irrefragable guarantees that it accurately represents what is perceived" (Griffiths, "The Limits of Criticism," pp. 149-150).
  23. Prasannāpada 69.13-14: Kim ca "ghaṭaḥ pratyakṣa" ityevamidikasya laukikavyavahārasya-asamgrahād, anāryavyavahārābhyupagamāc ca, avyāpitā laksanasyaiti na yuktam etat. It seems clear that Candrakīrti's point here recommends, instead of the text's abhyupagamit, a more normative sense, for example abhyupagameyatvāt, "because it should be accepted." Cf. 71.3-5 (note 34 below) and 74.1-8 (note 41), where the normative sense of Candrakfrti's point is made explicit. For the whole passage under consideration in this essay, see also Candrakīrti's Catuḥśatakavṛtti, chapter 13 of which advances a very similar argument against the Epistemologists' account of perception. See Tom Tillemans, Materials for the Study of Āryadeva, Dharmapāla and Candrakīrti, Wiener Studien zur Tibetologie and Buddhismuskunde, Heft 24, 1-2 (Wien: Arbeitskreis for tibetische and buddhistische Studien, 1990), vol. 1, pp. 176-179 (translation), and vol. 2, pp. 63-68 (Tibetan text).
  24. I should again make clear that, throughout this essay, the attribution of positions to "Dignāga" always has as its antecedent Candrakīrti's attribution of positions to an unnamed interlocutor (whose position happens to look much like Dignāga's). See note 2 above.
  25. 70.1-3: Atha syāt: ghatopādānanīlādayaḥ pratyakṣaāh, pratyakṣapramāṇaparicchedyatvāt. Tataś ca yathaiva kāraṇe kāryopacāraṃ kṛtvā, "buddhānām sukha utpāda" iti vyapadiśyate, evaṃ pratyakṣanīlādinimittako 'pi ghaṭaḥ kārye kāraṇopacāraṃ kṛtvā pratyakṣa iti vyapadiśyate.
  26. This same example is cited in Vasubandhu's Abhidharmakośabhāṣyam (ad Abhidharmakośa 1.10); cf. the edition of P. Pradhan (Patna: K. P. Jayaswal Research Institute, 1975), p. 7. Just as in Candrakirti's text, the example is adduced by Vasubandhu to illustrate "kāraṇe kāryopacāra," "figurative use of the effect in place of the cause." Cf. also Dhammapada 14.16 (stanza 194): sukho buddhānām uppādo sukhā saddhammadesanā / sukhā sahghassa samaggī samaggānaṃ tapo sukho.
  27. It is, however, a reasonable question (as it is throughout this discussion, although I have seldom been concerned to ask it) whether Candrakrti has fairly represented Dign5ga here. In fact, Dignāga seems in at least one place to make a rather different use of the appeal to upacāra. Cf., for example, Hattori, Dign5ga, On Perception (note 21 above), p. 106, n. 1.65, where it is explained that, for Dign5ga, all talk of pratyakṣa is "figurative" insofar as, ultimately, "there is only the one fact of sva-sa ṁvitti.. . ." (Hattori aptly makes reference, in this regard, to Sthiramati's remarks on upacāra from the Triṃśikābhāṣya.) Here, then, Dignāga's usage may particularly require his commitment to vijñaptimātratā as the relevant context, and Candrakīrti seems rarely (if ever) to read him that way.
  28. 70.3-7: Na evaṃ vidhe viṣaya upacāro yuktah; utpādo hi Joke sukhavyatirekenopalabdhah. Sa ca saṃskṛtalakṣaṇasvabhāvatvād anekaduṣkaraśatahetutvād, asukha eva. Sa sukha iti vyapadiśyamāno 'saṃbaddha evety, evaṃviṣaye yukta upacāraḥ. Ghaṭaḥ pratyakṣa ity atra tu, na hi ghato nāma kaścid yo 'pratyakṣaḥ pṛthagupalabdho yasya-upacārāt pratyakṣatvaṃ syāt.
  29. For this whole section on "figurative usage," see, especially, Austin, Sense and Sensibilia, pp. 88-103, where Austin considers Ayer's appeal to different "senses" of the terms he claims to explain.
  30. See, for example, Kāvyaprakāśa 2.
  31. Paul Grice, "Logic and Conversation" (note 15 above), p. 28; cf. pp. 28-31. While Grice is perhaps persuasive in this account of what is likely to have occurred in order for us to know that an "implicature" has been effected, he is probably wrong to think that similar criteria can be given from which we can be certain what the implicature is.
  32. 70.8-9: Nīlādivyatiriktasya ghaṭasya-abhāvād aupacārikam pratyakṣatvam iti cet, evaṃ api sutarām upacāro na yukta, upacaryamāṇasya-āśrayasya-abhāvād; na hi kharaviṣāṇataikṣṇyam upacaryate.
  33. 70.10-12: Api ca, Iokavyavahāraṅgabh ūto ghaṭo yadi nīlādivyatirikto nāstīti kṛtvā tasya-aupacārikaṃ pratyakṣatvaṃ parikalpyate, nanv evaṃ sati pṛthivyidivyatirekeṇa nīlādikam api nāstīti, nīlāder asya-aupacārikam pratyakṣatvaṃ kalpyatāṃ.
  34. 71.3-5: Tasmād evamādikasya Iokavyavahārasya lakṣaṇena-asaṃgrahād, avyāpitaiva lakṣaṇasyeti. Tattvavidapekṣayā hi pratyakṣatvam ghaṭādīnāṃ nīlānāṃ ca na-iṣyate; lokasaṃvṛtyā tv abhyupagantavyam eva pratyakṣatvaṃ ghaṭadīnāṃ. Note that with abhyupagantavyam, Candrakīrti has used a gerundive such as we would have expected in an earlier passage (see note 23 above). I have omitted from the text a scriptural quotation by Candrakīrti that simply repeats his point. The passage (71.1-2) goes: Yathoktam: "Rūpādivyatirekeṇa yathā kumbho na vidyate, vāyvādivyatirekeṇa tathā rūpaṃ na vidyate" iti ("As it is said [in Āryadeva's Catuḥśataka, chap. 14, v. 151, just as a pot does not exist as separate from its form and so forth, so, too, form does not exist as separate from wind and so forth [i.e., as separate from the elements]"').
  35. This point has been well put by Matthew Kapstein, who notes that, in the Buddhist account, "even the necessary truths of logic and mathematics, which may be known apriori, are here termed 'conventional' (Skt. sāṅketika, Tib. tha snyad pa). But 'conventional' in its Buddhist uses should not be taken to imply 'freely chosen from a given set of alternatives/ and much less 'arbitrary.' It refers, rather, to all language and propositional knowledge, and to the principles to which they conform and to their objects; for none of these is or directly points to that absolute reality whose realization is spiritual liberation. That absolute, of which not even the categories of the one and the many or of being and non-being can be affirmed, wholly transcends the familiar conventions of logic, experience, language and thought" (Matthew Kapstein, "Self and Personal Identity in Indian Buddhist Scholasticism" [Ph.D. diss., Brown University, 1987], pp. 397-398).
  36. Candrakīrti attributes this claim to Dignāga in a few places, as when he has his interlocutor say (68.5-6): Kimanayā sukṣmekṣikayā? naiva hi vayaṃ sarvaprama ṇaprameyavyavahāraṃ satyam ity ācakṣmahe; kiṃ tu lokaprasiddhir eṣāmunā nyāyena vyavasthāpyata iti ("What's the use of this hair-splitting? For we do not say that all discourse involving warrants and cognizables is true; rather, this [which is] familiar in the world is [all that is] established by this argument"). Cf. also 58.14-59.1, where Candrakīrti attributes to his interlocutor the following explanation of why the epistemological project is called for: athasyād esa eva pramā ṇaprameyavyavahāro laukiko 'smābhi ḥ śāstre ṇa-anuvar ṇita iti / tad anuvar ṇanasya tarhi phalaṃ vācyaṃ / kutārkikai ḥ sa nāśito viparītalakṣa ṇābhidhānena, tasya-asmābhih samyaglakṣa ṇam uktam iti cet ... ("Perhaps it will be said: this is just the worldly convention regarding warrants and cognizables, which is explained by us with this text. In that case [Candrakīrti replies,] the payoff of this explanation should be explained. [Dignāga answers:] This [i.e., our epistemic practice] has been destroyed by faulty logicians [i.e., Naiyāyikas], through their predication of false characteristics; what we have explained is its real characteristics..."). See note 42 below for Candrakīrti's conclusion to this exchange.
  37. See note 42 below.
  38. Jay Garfield, Fundamental Wisdom of the Middle Way.- Nāgārjuna's Mulamadhyamakakārikā(Oxford: Oxford University Press, 1995), p. 116 n.
  39. Ibid., p. 122.
  40. I have, then, omitted the section of the text from p. 71.5 through p. 73, which chiefly concerns etymological arguments for and against Dignāga's doctrine of perception. These etymological arguments essentially concern the question of what kind of causal story can properly be taken as sanctioned by the word "perception" (pratyakṣa).
  41. 74.1-8: asty ayaṃ pratyakṣaśabdo loke prasiddhaḥ. Sa tu yathā loke, tathāsmābhir ucyata eva. Yathāsthitalaukikapadarthatiraskāreṇa tu tadvyutpāde kriyamāne, prasiddhaśabdatiraskāraḥ prasiddhaḥ syāt; tataś ca pratyakṣam ity evaṃ [na] syāt, ekasya ca caksurvijnānasya-ekendriyakṣa ṇāśrayasya pratyakṣatvam na syād vipsārthābhāvāt; ekaikasya ca pratyakṣatvābhive, bahunām api na syāt Kalpanāpoḍhasyaiva ca jñānasya pratyakṣatvābhyupagamāt, tena ca lokasya saṃvyvahārābhāvāt, laukikasya ca pramā ṇaprameyavyavahārasya vyākhyātum iṣṭatvāt, vyarthaiva pratyakṣapramā ṇakalpanā saṃjiyate. Note that in the underlined section of this passage, Candrakīrti again makes clear his normative claim that it is ordinary usage that should be accepted; cf. notes 23 and 34
  42. Cf. Prasannapadā 58.14-59.3 (see notes 36-37 above). After he has Dignāga thus explain that his account is offered by way of correcting that of the Naiyāyikas, Candrakīrti rejoins (59.1-3): Etad apy ayuktaṃ. Yadi hi kutarkikair viparītalaksanapranayanaṃ krtaṃ laksyavaiparītyaṃ lokasya syat. Tadarthaṃ prayatnasāphalyam syāt. Na ca etad evaṃ iti vyartha evāyaṃ prayatnaiti ("This doesn't make sense either; for if, based on the expression of a false definition by bad logicians, everyone made a mistake regarding what's under definition, then the point of this [i.e., of your proposed alternative to Nyāya epistemology] would be one whose effort was fruitful. But it's not so, and this effort is pointless"). As I am suggesting is the case for both Candrakīrti and Austin, it is not possible that people should thus always be mistaken; as Austin says, "it is important to remember that talk of deception only makes sense against a background of general non-deception. (You can't fool all of the people all of the time.)"
  43. In regard to the project of mapping Candrakīrti's argument onto that of Austin, I have not here gone into the question of whether Candrakīrti's opponent (i.e., Dignāga), like Austin's (i.e., Ayer), might be said to subscribe specifically to the "descriptive fallacy," nor whether Dignāga's might be said, like Ayer's, to be an internalist epistemology. In fact, Mark Siderits has argued that the contemporary internalist/externalist dichotomy does not map well onto Indian discussions. See, for example, Siderits' essay "Madhyamaka on Naturalized Epistemology" (unpublished ms.), where Siderits argues that the best hope for finding an analogue to contemporary "internalism" in Indian debates is in the Mimamsaka doctrine of svata ḥ prāmā ṇya (a suggestion that had been put forward by J. N. Mohanty), and that, in fact, svata ḥ prāmā ṇya is not an instance of an internalist epistemology. This seems to me to be precisely correct. Indeed, it seems to me difficult to understand svata ḥ prāmā ṇya as anything other than a critique of precisely the kinds of intuitions that guide internalist epistemologies. (For an account of svatah prāmānya vis-à-vis contemporary philosophy, see my article "Of Intrinsic Validity: A Study on the Relevance of P ūrva Mīmāṃsā," Philosophy East and West 51 [11 [2001].) It seems to me more promising to consider the possibility of finding an analogue to internalism in Dignāga's doctrine of svasaṃvitti¡Xor, at least, svasaṃvitti as (mis)represented by Candrakīrti, for Candrakīrti seems fairly clearly to read Dignāga's account of this doctrine as a specifically epistemological move. See, for example, 61.7-9: Atha tad api karmasādhanaṃ, tadā tasya-anyena kāraṇena bhavitavyaṃ, jñānāntarasya kara ṇabhāvaparikalpanāyām anavasthādo ṣaś ca-āpadyate ("And if this (additional awareness), too, has as its denotation an object, then it must (also) have a separate instrument; and given this conception of relation to an instrument on the part of another awareness, the fault of infinite regress ensues"). Candrakīrti has his interlocutor follow this by appealing immediately to svasaṃvitti (61.1 Off.). Clearly, then, Candrakīrti considers Dignāga's doctrine of svasaṃvitti, or "apperception," as one that is invoked precisely to stop the vicious regress that he takes to be opened up here. Thus, he sees svasaṃvitti as being invoked at precisely this point in the development of Dignāga's position in order to explain how it is that awareness might in fact be both an instrument and an object. It is an interesting question whether Candrakīrti's view of svasaṃvitti as an epistemically (as opposed to metaphysically) motivated category could contribute to our understanding of the doctrine. It may well be that Candrakīrti misrepresents Dignāga, or at least reads him out of context, for Dignāga's doctrine of svasaṃvitti seems to me crucially to presuppose vijñaptimātratā (indeed, to be simply another way of stating vijñaptimātrata). See, for example, Hattori, Dignāga, On Perception, nn. 1.61, 1.64-65 (pp. 102-106), and see notes 10 and 27 above. Still, it remains an interesting question whether Candrakīrti's critique might have some purchase against Dignāga. I intend to take a longer look at this issue in the near future.
  44. As Candrakīrti says early on in his argument, for those who have fully realized it, ultimate truth is "a matter of silence" (paramārtho by āryā ṇaṃ t ū ṣ ṇīṃbhāva ḥ) (57.7-8).
  45. While Austin would likely see this as a nonsensical notion, it is important to remember that Candrakīrti's argument is developed in the context of an essentially religious project. Thus, in comparing Candrakīrti with Austin, it must not be forgotten that CandrakTrti is first and foremost a Buddhist philosopher. For an especially compelling expression of this context, see the first chapter of Candrakīrti's Madhyamakāvatāra (bhā ṣya). Thus, whatever philosophical problems there might be with respect to Candrakīrti's category of "ultimate truth," it must be remembered that his fundamental commitment is to precisely this category and its soteriological implications.
  46. It would be reasonable, I think, to characterize the debate between Candrakīrti and Bhāvaviveka (that is, between the so-called "Prāsaṅgika" and "Svātantrika" Mādhyamikas, respectively) as concerning precisely this point. That is, the "Svātantrika" critics of Candrakīrti can be seen as concerned precisely with allowing for the ways in which we can still make distinctions, strictly at the conventional level, between "true" and "false" conventions. Indeed, later Svātantrikas such as Jñānagarbha and Śāntarak ṣita introduced precisely such a distinction with their notion of tathya- and mithyā-saṃvṛti. On this notion, see, for example, M. D. Eckel,jñānagarbha's Commentary on the Distinction between the Two Truths (Albany: State University of New York Press, 1987), pp. 54-55, 75, 111-112, 123, and Masamichi Ichigō, "Sāntarak ṣita's Madhyamakālamkāra," in Luis Gomez and Jonathan Silk, eds., Studies in the Literature of the Great Vehicle (Ann Arbor: Collegiate Institute for the Study of Buddhist Literature, University of Michigan, 1989), p. 160. See also Paul Williams, The Reflexive Nature of Awareness: A Tibetan Madhyamaka Defence (London: Curzon, 1998), which attends to attempts by Tibetan commentators (chiefly, Mi-pham) to allow for similar distinctions. For Candrakirti's attempt to address this issue, see Tillemans, Materials for the Study of Āryadeva, Dharmapāla and Candrakīrti, pp. 44-50, which considers at some length what is precisely Candrakirti's alternative to something like Jñānagarbha's tathya- and mithyā-saṃvṛti (with Tillemans' observations being largely based on Madhyamakāvatāra 6.23-25).
  47. Of course, there is still the question to what extent we could ever be in a position to specify which epistemic practices are thus necessarily presupposed, and which are optional for successful language use. in other words, how could we know which epistemic practices we can doubt? I have not gone into this question here, but it would obviously be an important one to consider in pressing a critique of the arguments of Austin and Candrakīrti (to the extent that they are similar on this point).