Xuanzangs Proof Of Idealism

Author(s): Eli Franco
Publisher: Eli Franco
Catalog link: https://jainqq.org/explore/269514/1

Page #1
Xuanzang's proof of idealism (vijnaptimatrata)* Eli Franco Imagine a country that cultivated strong cultural and religious ties with ancient Greece from the beginning of the Presocratic period. The intellectuals of this country become interested in Greek philosophy and begin to translate certain Presocratic works into their own language as well as some Platonic dialogues that no longer survive in the original Greek. Imagine further that, because of some mysterious event, the philosophical link is severed and the philosophers of our imaginary country then never hear of Aristotle, Hellenistic philosophy and all the other developments that took place in Europe in the subsequent centuries. The last Greek philosopher they know of is Plato, and an indigenous tradition of logic is then developed in the form of direct or indirect comments on his early dialogues. Surely, this tradition would be a fascinating object of study. Of course, it is clear why I made up this fable. The philosophical link between the Chinese and the Indian tradition, at least as far as the pramana tradition is concerned, was severed exactly in the crucial period between Dignaga and Dharmakirti. Thus, the Chinese Buddhist canon has preserved some very early dialectical-philosophical tracts that no longer exist in the original Sanskrit such as the *Upayahrdaya (or *Prayogasara), *Tarkasastra, and the Nyayamukha of Dignaga. However, no work of Dharmakirti is known to have been translated into Chinese, and the whole tradition of Buddhist logic in China, known as the yin ming, or the science of logical reasoning (hetuvidya), was developed on the basis of Dignaga's hetucakra as preserved in the Nyayapravesa. Nevertheless, the fascinating tradition of yin ming has hitherto remained largely ignored. In fact, since the pioneering studies of Tucci in the end of the 1920s very little has been done, at least in European languages. It was, therefore, with great anticipation that we were looking forward to Christoph Harbsmeier's study of Buddhist logic in China which was in the making for many years. I had the privilege of meeting Harbsmeier in Berlin in 1990, and at that time he already informed me that his voluminous work on the subject * I am indebted to my wife, Karin Preisendanz, for having read several drafts of this pa per, and to Professor Shoryu Katsura for very helpful comments. I would also like to thank my student, Shinya Moriyama, for translating the relevant passages from the Chi nese. 199

Page #2
Horin 11/2004 was practically finished. When the book finally appeared in 1998 as Volume 7 in the prestigious series Science and Civilisation in China,' it was, at least for me, a disappointment. The announced voluminous work on Buddhist logic had shrunk into a chapter of some fifty pages (pp. 358-408), scant in historical information, and unfortunately displayed an astonishing lack of insight into its subject matter. Harbsmeier's woeful error was to attempt to interpret Chinese Buddhist logic without an adequate study of the Indian logic from which it was derived. For some reason which is not clear to me, he actually believed that reading Buddhist logic in Chinese translation is easier and clearer than read-1 ing it in the original Sanskrit." The results, as could be expected, are not impressive. One should add, perhaps, that Harbsmeier did consult some Sanskritists. However, in what seems a puzzling procedure, he consulted only scholars who were not familiar with the subject matter. Thus, he writes: "I have been reading the Nyayapravesa together with three Sanskritists who between them have spent over sixty years reading Buddhist Sanskrit texts. The Sanskrit Nyayapravesa presented considerable concrete problems of grammatical interpretation to all of us. By comparison my task of expounding the literal meaning of the Chinese transla tion was easier." Well, take fifty scholars of Buddhism who know nothing about Buddhist logic and you could get a thousand years of Sanskrit reading among them, but you will still not be able to understand much of the Nyayapravesa. Let me illustrate the above criticism with the example of Harbsmeier's interpretation of Xuanzang's argument for Consciousness Only, which in Harbsmeier's own words is the most famous historical example of the art of Chinese Buddhist argumentation. However, it should be added perhaps that Joseph Needham, Science and Civilisation in China. Volume 7. Part I: Language and Logic. By Christoph Harbsmeier. Edited by Kenneth Robinson. Cambridge University Press 1998. 2 The Ph.D. dissertation of Uwe Frankenhauser is much more informative in this respect; cf. Uwe Frankenhauser, Die Einfuhrung der buddhistischen Logik in China. Wiesbaden 1996. Unfortunately, the philosophical and logical aspects of Buddhist logic in China are not Frankenhauser's main concern and their treatment leaves much to be desired. Xuanzang's proof of vijnaptimatrata is discussed in some detail on pp. 73-97 (cf. also pp. 116-117), but without providing much clarity. One can only regret that Frankenhauser did not consult any specialist on Buddhist logic for the preparation of his useful dissertation. Cf. for instance, Harbsmeier, op. cit., pp. 399-400. Cf. also p. 402: "Xuanzang's Chinese translation (of the Nyayapravesa] is not only often an improvement on the Sanskrit original, it turned out to my great surprise - to be generally easier to read as well." * Harbsmeier, op. cit., p. 402. Harbsmeier, op. cit., p. 392. 200

Page #3
Eli Franco: Xuanzang's proof of idealism (vijnaptimatrata) there is really nothing Chinese about it except that it survives in a Chinese translation. Xuanzang thinks like an Indian logician. His proof was delivered in a public debate in Kanyakubja (modern Kanauj), most probably in Sanskrit, and was intended not only for Buddhists, but also for a Hindu and Jain audience. In his presentation of the argument Harbsmeier uses Arthur Waley's translation: "Philosophically speaking, such forms of matter as are accepted by us and by you are inseparable from the eye-consciousness (Proposition). Because according to the eighteenfold classification that we too accept, these forms of matter are not included under the heading 'Organ of Sight' (Proof). Like eye-consciousness itself (Analogy)."' Waley himself could not make head or tail of this syllogism, and said that "[i]t is certain at any rate that the famous syllogism comes perilously near to being nonsense." However, in what seems like an afterthought, he added the following footnote: "I mean, of course, nonsense from the standpoint of his own contemporaries; whether it does or does not conform to our conventions of thought is irrelevant." The fact that Xuanzang's inference could not be refuted was merely due, according to Waley, to King Harsha's intimidating authority: "Neither during the five days of the assembly nor during a further period of eighteen days' grace did any opponent dare raise his voice. It was obvious indeed that King Harsha was not in a mood to tolerate criticism of the Mahayana or of Tripitaka (Xuan zang's monastic name) as its exponent." Nevertheless, Waley did try to make sense of this nonsense in his endnotes (pp. 271-272): "Matter and Eye-consciousness are alike, in that they are neither of them listed under the heading 'Organ of Sight' in the enumeration of the Eighteen Groups (dhatu). EyeConsciousness is 'inseparable from itself, i.e. from eye-consciousness. Therefore, matter [(which is 'like' eye-consciousness) must also be 'inseparable from eyeconsciousness."10 The debate presumably took place in the presence of King Harsha, accompanied by eighteen vassal kings, three thousand Buddhist monks and two thousand Hindus and Jains (cf. Harbsmeier, op. cit., p. 392). ? Arthur Waley, The Real Tripitaka and other pieces. London 1952, p. 64. 8 Op. cit., p. 63. Op. cit., pp. 64-65. 10 Waley then continues to explain the three qualifying terms "philosophically speaking," "as accepted by us and by you" and "that we too accept," but we need not concern ourselves with these here. 201

Page #4
Horin 11/2004 Thus, according to Waley's explanation, Xuanzang's argument is based on an analogy and is patently false, for it has the following structure: a and b share the property X, b has the property Y, therefore, a has the property Y. In Waley's defence we must add, however, that very little was known about Dignaga's logic at the time he was writing and there was no way for him to make sense of an inference which operated according to rules that were not yet worked out by modern (Western and Japanese) research on Buddhist logic. Further, Waley could not rely on the indigenous Chinese scholarly works on yin ming. Chinese scholarly interest in Xuanzang's inference declined after the Song dynasty," and consequently, when interest arose again towards the end of the Ming dynasty, some elements of the tradition, especially the immediate knowledge transmitted from teacher to disciple, seem not to have survived. Unlike in China where interest in it waxed and waned, in Japan Xuanzang's inference seems to have been the object of uninterrupted interest and study continuing up to the present time, especially in the Hossb-school. According to Frankenhauser, 12 the scope of the qualification zhengu (*paramarthatah) alone was an object of intense debate for three hundred years. Although numerous treatises in the form of shiki ("private illustrations") and sho ("compendia") were composed, the core of Xuanzang's inference seems to have remained unintelligible in Japan as well. Thus, the polymath Hajime Nakamura, who was certainly well acquainted with traditional Japanese scholarship on the subject, came to the conclusion that "Xuanzang himself does not understand Indian Logic (yin ming) in its normal manner." After translating the inference, Nakamura adds: "However, this formula [i.e., *paramarthatah, from the point of view of absolute truth] is not convincing to the people who do not accept the ultimate truth beyond our senses. In other words, it is meaningless as an inference for others (pararthanumana). It is also meaningless to provide a reason which is acceptable only for one's own position, but not for the opponent, because the reason must be acceptable to both, i.e., opponent and proponent. This means that Xuanzang does not understand the difference between inference for oneself and inference for others." 1 Cf. Frankenhauser, op. cit., p. 75. 12 Ibid. Of particular importance was the ardent debate that took place between the Nanji line, represented by Myozen (789-868), and Hokuji, represented by Zenju (723-797). 19 Cf. Hajime Nakamura, Introduction to the Nyayapravesa (Kokuyaku issaikyo, Wakan senjutu 43, Ronsobu 23, Daito Shuppan, Tokyo 1958). I owe this reference to the kindness of Mr. Shinya Moriyama. 14 Ibid. 202

Page #5
Eli Franco: Xuanzang's proof of idealism (vijnaptimatrata) By contrast, Gregor Paul who has spent a lifetime working on Chinese and Japanese philosophy, and to whom we owe a thorough survey of philosophy in Japan from its beginnings to the Heian period, is more respectful to Xuanzang, and is well aware that he could not solve the crux of the matter; thus, he suggests only tentatively that a possible meaning of the syllogism could be that since the organ of vision does not see, the visual consciousness would have to see. 15 However, let us go back to Harbsmeier's recent study. Xuanzang's argument does not make more sense to Harbsmeier than it does to Waley, Nakamura or Paul, and Harbsmeier too feels he has to make sense of this nonsense. He begins by declaring (p. 393) that Xuanzang's argument should be expressed in the form "esse est aestimari esse 'to be is to be deemed to be' and that a detailed comparison with Arthur Schopenhauer on this issue is most instructive. Objective existence is doomed to be exactly what it is deemed to be. We have no access to it other than deeming." This, however, happens to be the exact opposite of what Xuanzang said, namely, objective existence is not what it is deemed to be. Things are deemed to have an objective existence, but are in fact only images in our consciousness. Let one be reminded that the purpose of the syllogism is to prove Consciousness Only, i.e., that external objects do not exist. However, let me skip the various stages in Harbsmeier's argument and go straight to his conclusion, that is, his rephrasing of the argument: "Thesis: Speaking philosophically, and assuming the meaning of the terms ('visible things' and 'be not separable from eye-consciousness') accepted by the disputing parties: VISIBLE THINGS ARE DEFINITELY NOT SEPARABLE FROM EYE-CONSCIOUSNESS. Reason: Because, using the term ("visible things') in the opponent's sense of the term: though being in the visual sphere they are not perceived by the visual organ. Example: (Everything which, though in the visual sphere, is not perceived by the visual organ is not separated from eye-consciousness.) As for example eye-consciousness." Is Cf. Gregor Paul, Philosophie in Japan. Von den Anfangen bis zur Heian-Zeit. Munchen 1993, p. 191: Ein moglicher Sinn des Syllogismus durfte darin liegen, dass, da nicht das Sehorgan sehe, das Sehbewusstsein 'sehen' musse." Paul also refers in this context to critical studies of Shoho Takemura, Immyo-gaku. Tokyo 1986, but does not provide any details as to Takemura's interpretation of the inference. It would be interesting to pursue the discussion of Xuanzang's inference in Chinese and Japanese publications. A number of references are provided by Frankenhauser and Harbsmeier, but this is an area of study which lies beyond my competence. 10 Harbsmeier does not actually draw the comparison; thus it remains unclear why or how it is instructive. I am also not sure that esse est aestimari esse is correctly rendered by "to be is to be deemed to be," but I will not enter this matter here. 203

Page #6
Horin 11/2004 This is, of course, a brand new inference which is radically different from the original inference of Xuanzang. Xuanzang's reason is: "they are not included under the heading 'Organ of Sight""; that is, they are not the same as, or they are different from the "organ" of sight; not "they are not perceived [my emphasis] by the organ of sight." Furthermore, Harbsmeier's reason is clearly false: the property "not being perceived by the organ of sight" also applies to the "organ" of sight because it cannot perceive itself," and thus the reason is not absent in the vipaksa which is "that which is separate from the eyeconsciousness," i.e., includes the visual "organ." As for the example, Harbsmeier explains that the addition in parentheses makes explicit what, according to the Chinese logicians, the example serves to exemplify. But this is clearly wrong, at least from both the Indian and Xuanzang's perspective. The example is, by definition, an instance of the property to be proved, in our case, "being inseparable from eye-consciousness." And it is obvious, indeed tautological, that eye-consciousness is inseparable from, i.e., identical with, eye-consciousness. As to why this, and no other example can be used here will become clear from the discussion below. Harbsmeier probably suspects that his version of the argument also comes perilously near to being nonsense and therefore adds apologetically and condescendingly: "Let us keep in mind that this is an argument belonging not to the 20th but to the +7th century" (p. 394), implying that for us, clever as we are, the inference carries no conviction, but to those guys in the seventh century this poppycock must have been good enough. It is not my intent here it to criticize and rectify every mistake Harbsmeier makes in his explanation of the argument, which runs for two more pages. Let me show instead what, I submit, is really going on here. To begin with, a literal translation of the inference is in fact simpler and clearer than Harbsmeier's: 18 17 This is so not only because the sense (rather than the "organ") of sight is incapable of reflexive action, but also because it is invisible, like all the other sense faculties. Throughout the discussion Harbsmeier confuses sense organs (like the eyeball) with sense faculties (like the ability to see). Xuanzang's inference refers only to the senses, not to the organs. From the Buddhist point of view, organs, such as the eye and the nose, are classified as visible things (i.e., dhatu 2, cf. the table below), not as senses. Cf. also La Siddhi, p. 42, quoted below (n. 31): "Comme l'indique leur nom d'indriya (Kosa, ii, p. 103) ils sont seulement des 'puissances' (sakti)...." 18 I am indebted to Shinya Moriyama for this translation. [In his so-called "Great Commentary" E on Xuanzang's translation of an introduction to Dignaga's hetuvidya (Chin. yin ming ), Kui Ji (632-682) renders Xuanzang's famous syllogism as follows: Zhen Gu Ji Cheng Se Bu Li Yu Yan Shi Zong . Zi Xu Chu San She Yan Suo Bu She Gu Yin . (Taisho, vol. 44, No. 1840, p. 115). Gregor Paul.] 204

Page #7
Eli Franco: Xuanzang's proof of idealism (vijnaptimatrata) Thesis: From the point of view of absolute reality (*paramarthatah), colour and form (*rupa, i.e., the visual or visible objects), which are well known among the people (*lokaprasiddha), are not separate from the visual consciousness." Reason: Because while being included in the first three [dhatus) that (we too] accept, they are not included in the sense of vision. Example: Like the visual consciousness. The core of the argument is this: The "visual objects" (i.e., the objects we can see with our eyes) are in fact nothing but images in our visual consciousness ("are not separate from visual consciousness," i.e., do not exist outside our consciousness), because they are different from the sense of vision." Other parts of the argument, such as the statement that forms of matter are accepted by both the proponent and the opponent, as well as the statement of the example, fulfill some formal conditions required in a public debate, but they are immaterial to the logic of the argument and need not concern us here.21 Before I explain the argument itself, however, we need to understand the historical context in which it was employed. We do not know exactly what Dignaga's original contribution to logic was because we know so little of his predecessors. The theory of trairupya, which specifies three conditions that every valid reason must fulfill, existed prior to his time, for it appears clearly in early works such as the Tarkasastra.22 However, it was Dignaga's Hetucakradamaru and other works that made the theory of trairupya extremely persuasive and contributed to its dissemination among all contemporary schools of thought. Its wide acceptance must have also facilitated its usage in public debate. Indeed, Dignaga made this theory so persuasive that it almost looks like a specimen of deductive logic. Its deductive appearance is strengthened by the fact that Dignaga does not use elements of Buddhist metaphysics for his illustrations, but relies on the ontology of the Vaisesika. 19 Following Waley and Harbsmeier I use the word "consciousness" as equivalent to vi jnana. Note, however, that this term is more commonly translated as "cognition" or "awareness." 20 So also Waley's translation quoted above. Harbsmeier, however, has: "using the term ('visible things') in the opponent's sense of the term." 21 The reason for these qualifications is that Xuanzang's inference refers to forms of mat ter, which ultimately, that is, from the Yogacara point of view, do not exist as such. This may seem to violate the requirement of ubhayasiddha; but Xuanzang thwarts a possible objection by pointing out that the Yogacaras too accept the visible etc., but unlike the realists they do not accept them as distinct entities, but as aspects in consciousness. Cf. Tarkasastra p. 13 (in Guiseppe Tucci, Pre-Dinnaga Buddhist Texts on Logic from the Chinese Sources. Repr. Madras 1981): asmabhis trilaksano hetuh sthapitah. tad yatha paksadharmah sapaksasattvam vipaksavyavrttis ca. 205

Page #8
Horin 11/2004 A clear example is provided by his use of atoms. In earlier works such as the Upayahrdaya" atoms are used, in accordance with the Buddhist doctrine, as an example of impermanent entities. Dignaga, on the other hand, uses them as an example of permanent entities in agreement with the Nyaya-Vaisesika doctrine 24 As a result of this state of affairs, there is a brief period of some hundred years in the history of Indian logic in which "everyone" is trying to construe inferences that fulfil the three conditions of a valid reason. Perhaps at this point it may be briefly reminded what the three conditions are. They are 1) the presence of the reason in the subject of inference, 2) its presence in the sapaksa, that is, the group of things that possess the property to be proved, and 3) its absence from the vipaksa, that is, the group of things that do not possess the property to be proved. For instance, when one says "this mountain possesses fire because it possesses smoke," the inference is valid if and only if the reason "possessing smoke" is present in the subject, that is the mountain, and in the sapaksa, that is the group of things possessing fire, for instance a kitchen, and is absent in the vipasa, that is the group of things that do not possess fire, for instance a lake. Let it be recalled that no particular relation between the properties is required by the theory of trairupya; the above inference does not presuppose that smoke is caused by fire or that there is any particular relationship between them, except the fact that they happen to occur in the same place, or put more technically, the properties "possessing smoke" and "possessing fire" happen to belong to the same property-possessor. Thus, an inference was considered valid if its reason fulfilled the three conditions, no matter how bizarre or irrelevant that reason may have sounded. This, of course, encouraged the use of sophisms, and indeed a number a famous sophisms, such as the sadvitiyaprayoga attributed to the Carvakas, originated at that time. This exciting period in Indian logic ends, fortunately or unfortunately, with Dharmakirti. According to him only two types of relations, namely causation and own nature, may be used in a valid inference. Xuanzang, however, sojourned in India precisely in the period between Dignaga and Dharmakirti. Dharmakirti's dating may be a moot issue, but 23 Cf. Tucci, op. cit., p. 21. * I have to admit, though, that I previously considered the Hetucakradamaru to display a system of deductive logic in which the validity of an inference is derived from the rela tionship between the terms and not from their putative existence. 2 On the sadvitiyaprayoga cf. Tom Tillemans, "Dharmakirti on some Sophisms," in Ernst Steinkellner (ed.), Studies in the Buddhist Epistemological Tradition (Proceedings of the Second International Dharmakirti Conference, Vienna, June 11-16, 1989). Wien 1991, pp. 403-415. 206

Page #9
Eli Franco: Xuanzang's proof of idealism (vijnaptimatrata) there is no evidence that Xuanzang knew of him or that he was familiar with his new theories of inference. Thus, Xuanzang's inference must be appreciated against the background of Dignaga's logic. Perhaps the most crucial point to be considered here is that Dignaga's inferences are not deductive. This point concerns the relationship between the paksa, the subject of inference, the sapaksa, the group of all things that possess the property to be proved, and the vipaksa, the group of all things that do not possess the property to be proved. First, is the paksa included in the sapaksa, as was usually the case in the later period after Dharmakirti?26 Claus Oetke has argued strongly and convincingly that if one takes Dignaga's formulations literally, the paksa was excluded from the sapaksa and the vipaksa.27 It is especially the latter exclusion that is most important in the present context. This double exclusion is not entirely surprising, since at the time of drawing an inference we do not yet know whether the subject possesses the property to be proved or not. If the subject of inference, e.g., the mountain that possesses smoke, is already included in the groups of all things that possess fire, then the inference seems unnecessary. If, on the other hand, it is included in the group of things that do not possess fire, the inference is false. It is precisely this feature that makes the inference non-deductive. If we say: "Wherever there is smoke there is fire, and this mountain possesses smoke; therefore this mountain possesses fire," this is a deductive inference. We may change the terms without affecting its validity and say: "Wherever there is X there is Y, and Z has X, therefore Z has Y." The inference is valid no matter what X, Y and Z are. But if we say: "In all cases except this moun- . tain, wherever there is smoke there is fire, and this mountain possesses smoke; therefore this mountain too possesses fire," this is no longer a deductive inference, even if "all cases" would mean thousands or millions of cases. Thus, every Dignagan inference divides the entire world into three mutually exclusive parts. The sapaksa which is the group of all things that possess the property to be proved; the vipaksa which is the group of all things which do not possess the property to be proved; and the subject of inference, about which there is a doubt because we do not yet know whether it possesses this property or not. In the case of the above inference, the property to be proved is "possessing fire," and it divides the entire world into three groups: things that possess fire (exemplified by the kitchen), things that do not possess fire (exemplified by the lake), and the mountain, of which we are attempting to prove that it possesses fire. 26 On sapaksa cf. Tillemans, Scripture, Logic, Language. Boston 1999, pp. 89-116. 21 Cf. Claus Oetke, Studies in the Doctrine of Trairupya. Vienna 1994, pp. 17-73. 207

Page #10
Horin 11/2004 The same operation should be followed in the interpretation of Xuanzang's inference. The terminology of the argument clearly refers to the 18 dhatus. These are: 1. caksus (sense of vision) 2. rupa (the visible) 4. srotra (sense of hearing) |5. sabda (sound) |7. ghrana (sense of smell) 8. gandha (odour) 3. caksurvijnana (visual consciousness) srotravijnana (auditory consciousness) 9. ghranavijnana (olfac tory consciousness) 12. jihvavijnana (gustatory consciousness) 15. kayavijnana (tactile consciousness) 18. manovijnana (non sensuous consciousness) 10. jihva (sense of taste) 11. rasa (flavour) 13. kaya (sense of touch) 16. manas (sense of mental objects) 14. sprastavya (the tangi- ble) 17. dharmas (non-sensuous objects) Now, in Xuanzang's inference, the property to be proved is "not being separate from visual consciousness." This can be simplified as meaning "being visual consciousness." Accordingly, our inference divides the world as follows: subject of inference (visual objects): dhatu no. 2 sapaksa (visual consciousness): dhatu no. 328 vipaksa (all the rest): dhatus no. 1,4-18 Thus, our inference could also be formulated as follows: dhatu no. 2 (rupa) is not separate from (i.e., is identical with) dhatu no. 3 (caksurvijnana), because it is not dhatu no. 1 (caksurindriya). Now what we have to do next is see whether the reason, namely, "not being the sense of sight," i.e., not being dhatu no. 1, i.e., being dhatus nos. 2-18, fulfills the three conditions. The first condition is fulfilled because the property of not being dhatu no. 1, which occurs in (or which extends over) dhatus nos. 2-18, is present in dhatu no. 2. The second condition is also fulfilled because the property which occurs in dhatus nos. 2-18 is present in dhatu no. 3. However, the third condition is obviously not fulfilled. The property which occurs in dhatus nos. 2-18 is obviously not absent in dhatus nos. 4-18. So the inference in this form is clearly false, and indeed does seem nonsensical. However, it is clear that the reason has a qualification (visesana). It is a conspicuous feature of the Indian syllogism that the reason is usually ex 28 We can understand now why the example of the inference ("like eye-consciousness it self") was chosen: the sapaksa contains a single entity (or better: a single type of entity). Thus, nothing else could be used as an example. 208

Page #11
Eli Franco: Xuanzang's proof of idealism (vijnaptimatrata) pressed in the form of a single property. If two properties are needed, the second property is expressed in the form of a qualification of the first property. The qualification appears in the discussion and Harbsmeier's translation, but he fails to realize its function and implications: "THOUGH BEING IN THE VISUAL SPHERE THEY ARE NOT PERCEIVED BY THE VISUAL ORGAN." The qualification is clearly "though being in the visual sphere." What Harbsmeier calls the visual sphere2 comprises dhatus nos. 1-3, that is, the sense of vision, the visible object and the visual consciousness. Consequently, we can reformulate the inference as follows: dhatu no. 2 (rupa) is not separate from dhatu no. 3 (caksurvijnana), because it is not dhatu no. 1 (caksurindriya) though it is included in dhatus nos. 1-3. If we now combine the two statements (not being dhatu no. 1 and being included in dhatus nos. 1-3), it is clear that the reason consists of a property which occurs in dhatus no. 2 and 3. Therefore, it is present in the sapaksa and it is absent in the vipaksa. For the distribution of the sapaksa and the vipaksa was as follows: sapaksa (being visual consciousness): dhatu no. 3 vipaksa (all the rest): dhatus nos. 1, 4-18 Thus, the reason is valid. This implies, of course, that one could transpose any of the three dhatus and say, for instance: the visual sense is not separate from the visual consciousness because, while it belongs to the visual sphere, it is not the visual object. In this case the subject, dhatu no. 1, is not separate from dhatu no. 3 because it is not dhatu no. 2, although it is included in dhatus nos. 1-3. The property "not being dhatu no. 2" is present in dhatus nos. 1-3 and cannot be found in dhatu no. 2, which means that it is found in dhatus nos. 1 and 3 only. Thus the distribution into paksa, etc., would be as follows: Subject: no. 1 sapaksa: no. 3 vipaksa: nos. 2, 4-18 29 I am not aware of an equivalent term in Sanskrit that would comprise the three visual dhatus. The Chinese version of Xuanzang's inference does not use such a term. 30 Should one say that the property must be no. 3 because it cannot be no. 1, since no. 1 is the paksa? Probably not, because the requirement is that the paksa is excluded from the domain of the sapaksa, not from the domain of the hetu. 209

Page #12
Horin 11/2004 The first condition holds because the property "not being dhatu no. 2" which occurs in dhatus nos. 1 and 3 is present in dhatu no. 1; the second condition holds because the same property is present in dhatu no. 3, and the third condition holds because, owing to the qualification of the reason, being included in dhatus nos. 1 and 3 is absent in dhatus nos. 2,4-18. Could the opponent contest Xuanzang's position by formulating a competing inference? He could not, because any re-arrangement of the three visual elements is acceptable to Xuanzang: If the visible object is not separate from visual awareness (because it is not the visual sense), then visual awareness is (tautologically) also not separate from the visual object (for the same reason). Even if one takes the visual sense as the subject of inference, inferring that the sense of vision is not separate from visual awareness because it is not the visual object, or that it is not separate from the visual object because it is not visual awareness, the inferences will merely prove the Yogacara position that the senses are nothing but seeds in the awareness." Moreover, if the opponent were to suggest such an inference, he would be guilty of contradicting his own position (siddhantavirodha), for the opponent, being an adept of Conservative Buddhism, does not admit that any dhatu is inseparable from or identical with any other dhatu. Still, Xuanzang's inference may seem to be merely a clever trick. Reduce the world of discourse into three entities - in practice this is what the qualification does by excluding the reason from being present in dhatus nos. 4-18 - and you could say: Entity no. 1 is (not separate from) entity no. 2 because it is not entity no. 3 (and if one needs an example, one may add: like entity no. 2). Since our world has only three entities, the reason does not need a qualification in order to be valid. The distribution into paksa, etc., is: Subject: no. 1 sapaksa: no. 2 31 Cf. L. de La Vallee Poussin, Vijnaptimatratasiddhi. La Siddhi de Hiuan-Tsang. Paris 1928, p. 42: "Comme l'indique leur nom d'indriya (Kosa, ii, p. 103) ils sont seulement des 'puissances' (sakti), non pas des choses exterieures constituees par la matiere derivee des quatre grands elements (upadayarupa, bhautikarupa, Kosa I, p. 21). Un Rupa sapratigha exterieure a la pensee est rationnellement inadmissible: donc les cinq organes, comme leurs cing objets, en un mot les dix ayatanas, ne sont que le developpement-manifestation du Vijnana." Cf. also ibid., pp. 230-231. According to Frankenhauser, op. cit., pp. 76 and 78, Xuanzang's inference was the central part of a larger treatise and was directed against a certain Prajnagupta, a Sammitiya from South India, who wrote a treatise refuting the Mahayana in seven hundred slokas. Others, however, are of the opinion that Xuanzang's opponent was a Sarvastivadin. In fact, Xuanzang's inference could be used against any representative of Conservative Buddhism. 210

Page #13
Eli Franco: Xuanzang's proof of idealism (vijnaptimatrata) vipaksa: not no. 2, i.e., nos. 1 and 3, but it cannot be no. 1, because no. 1 is the paksa; thus it must be no. 3. The reason, "not being no. 3," is present in the subject (no. 1) and in the sapaksa (no. 2), but absent in the vipaksa (no. 3)." What could be the response of the realist Buddhist opponent to Xuanzang's inference? As Harbsmeier points out "nobody dared raise any objections against the formidable Xuanzang on that occasion in India in the year +642." However, the Korean patriarch Wonhyo (618-686) is supposed to have challenged Xuanzang's inference with the following one: "Thesis: Philosophically speaking, and taking the term visible thing in the commonly accepted meaning: Visible things are separate from eye-consciousness. Reason: Because, using the term 'visible thing' in the proponent's sense, though in the visual sphere, they are not perceived by the eye. Example: (Everything which, though in the visual sphere is not perceived by the eye, is separate from eye-consciousness.) Like the eyeball." Something seems to have gone seriously wrong in this translation: Visible things are by definition those things that are perceived (or more precisely, are capable of being perceived) by the "eye." How could visible things not be perceived by the "eye"? Again, the literal translation kindly prepared by Mr. Moriyama is simpler and clearer; it is also radically different from Harbsmeier's: "From the ultimate point of view, colour and form, which are well known among the people, are separate from visual consciousness. 33 To make the argument less abstract, imagine a world with only three classes of things in it: apples, strawberries and bananas. Following the above mentioned rules of inference (trairupya plus the exclusion of the paksa from sapaksa and vipaksa) one should be able to make the following inference: Thesis: An apple is not separate from a strawberry. Reason: Because it is not a banana. Example: Like a strawberry. The distribution into the domains of paksa, etc., would be as follows: paksa: apple sapaksa: a strawberry vipaksa: not a strawberry, i.e., an apple and a banana, but also not an apple because apple is the subject of the inference, therefore, only a banana. The first condition applies because the property "not being a banana" is present in the apple. The second condition applies because the same property is present in a strawberry. The third condition applies because the property "not being a banana" is absent in a banana. * Op. cit., p. 396. The point, however, is not that nobody dared to contradict Xuanzang, but that his inference was perfectly correct. 35 Harbsmeier's translation, op. cit., p. 396. 211

Page #14
Horin 11/2004 Because while included in the first three [dhatus] that (we) accept, they are not in cluded in the visual consciousness, like the sense of vision." What Wonhyo is trying to do here is clear: He does not so much attempt to prove his own thesis, but rather to make Xuanzang's inference invalid by using a reason called viruddhavyabhicarin, i.e., construing a contradictory inference whose reason also fulfils the three conditions. For according to the rules of debate, if two contradictory inferences fulfill the three conditions (trairupya), both inferences are invalid. So does Wonhyo's inference fulfill the three conditions? As far as I can see, we can interpret the inference in two ways. Either one takes "being separate" in the thesis and "not being included" in the reason as synonymous, in which case the inference fulfills the three conditions, but is tautological (that is, the sadhya and the hetu are identical; the inference amounts to stating that colour and form are separate from visual consciousness, because ... they are separate from visual consciousness). Or, what seems much more probable to me, "being separate" and "not being included" are not synonymous; for instance, one may understand "not included" as "not included [by the Buddha in his enumeration of the dhatus]". In this case, the reason "not being included by the Buddha in visual consciousness" is present in the subject (colour and form), is also present in the sapaksa (sense of vision), and it is also absent in the vipaksa (visual consciousness)." Thus, at least as far as the three characteristics are concerned, it seems that Wonhyo succeeded, after all, to annul Xuanzang's brilliant inference." 38 * The Nyayapravesa gives the following example of two contradictory inferences that fulfil the three conditions: 1) Sound is impermanent because it is produced, like a pot. 2) Sound is permanent because it is audible, like soundness (or: like the universal "being a sound"). Cf. The Nyayapravesa. Part I: Sanskrit Text with Commentaries. Ed. A.B. Dhruva. Baroda 1968, p. 4.21f. 37 One formal point for which Wonyo's inference may be criticized is that the qualification in the reason is superfluous. Wonyo has certainly included it as an allusion to (perhaps even parody of) Xuanzang's reason. Of course, one may raise at this point further questions concerning the nature of inclusion and whether a dhatu can be said to be included in itself. However, as far as I can see such questions were not raised in the present context. In the discussion that followed the reading of this paper, Professor Shoryu Katsura raised the question whether Dignaga would have accepted Xuanzang's inference and argued that Dignaga's statements on paksabhasa preclude such sophisms. Unfortunately, I am unable at present to undertake a detailed investigation of this question, but I hope to return to this issue on a future occasion. It is interesting to note that according to the biography of Wonhyo, the expert logicians did not consider the inference to be contrary to Dignaga's doctrine; we read (as translated by Frankenhauser, op. cit., p. 74) that "Als (Xuanzang] danach die allgemein akzeptierte Inferenz im Sinne hochster Wahrheit aufstellte und das Hinayana widerlegte, machten die Sastrameister der Westgebiete keine Erklarungen zu der Inferenz. Sie alle sagten, 'nur Dignaga konnte diese Inferenz erklaren'." 212