A Pre-Dharmakīrti Indian Discussion of Dignāga Preserved in Chinese Translation: The Buddhabhūmy-upadeśa
That Dignāga 陳那 revolutionized Indian philosophy is well known. Both Buddhist and non-Buddhist sources attest to that, praising him or contending with him for many centuries. That Dharmakīrti 法稱 (ca. 600666), a century and half or so after Dignāga, took Buddhist logic and epistemology to a new level — or as Yijing 義淨 puts it, "Dharmakīrti revitalized hetu-vidyā”1 — is also well known. What remains far more obscure and controversial are such basic questions as: What was actually radical about Dignga's contribution?2 What was happening in Buddhist hetu-vidyā practice in the interim between Dignāga and Dharmakīrti? In what ways has the success of Dharmakīrti’s “revitalizations” led subsequent Buddhists and scholars to read back Dharmakīrtian ideas into their understanding of Dignāga, perhaps thereby distorting our perception of Dignāga’s philosophy or overlooking what Dignāga himself says about his own ideas?
One reason for the uncertainties concerning these questions is that, aside from some fragments, the Sanskrit originals of Dignāga’s works have not been available.3 Another is that scarcely any Buddhist hetu-vidyā or related texts from the period between Dignāga and Dharmakīrti are currently available in Sanskrit either.4 It is not even clear how many such works might have circulated in that period, much less who their authors might have been. To compensate for the lack of such materials, modern scholars — relying on precedents within the Tibetan tradition — interpret Dignga in the light of the interpretation of Dharmakīrti and subsequent Buddhist prama-vāda discussions, reading these back into Dignāga. While scholars readily concede that there are signifi cant “differences” between Dignāga and Dharmakīrti, determining with precision what these are exactly often remains unclear. Without counterbalancing alternatives, such as pre-Dharmakīrti interpretations of Dignāga, reading Dharmakīrti back into Dignāga has seemed not only prudent but necessary. Dignāga in China
Since the Chinese tradition preserves materials no longer available in Sanskrit, it is a valuable resource when investigating such questions. Tibetan Buddhism, which best preserves the Indian Buddhist pramāavāda tradition — adding many of its own embellishments — was founded post-Dharmakīrti, but its repository of Indian sources from earlier periods, such as the debate traditions preserved in abhidharma literature, is severely limited. In short, the transmission of Buddhism to China was in full swing from the fi rst through seventh centuries CE, after which virtually no new Indian developments entered East Asian soil. Dharmakīrti’s thought was virtually unknown in East Asia, and neither his texts nor any of the subsequent hetu-vidyā or pramāa-vāda texts that assumed prominence in India and Tibet were translated into Chinese. While China preserves a fuller record of Indian and Central Asian Buddhist developments through the seventh century,
Tibet best preserves the important Buddhist innovations from the eighth century onwards, while lacking many of the earlier materials primarily extant only in Chinese. It is as if Buddhism fl owed into China until roughly the end of the seventh century, and then someone turned the spigot, redirecting the fl ow to Tibet into which Buddhism continued to fl ow until its virtual extinction in India around the thirteenth century or so. Thus it is understandable that most Western scholars working on hetu-vidyā have relied primarily on Tibetan and — when extant or rediscovered — Sanskrit materials for forming their understanding of this important dimension of Buddhist philosophy,5 while treating Chinese materials (aside from the most obvious — such as the few well known Chinese translations of Dignāga’s work) with relative neglect.
While the Chinese tradition can offer help with some of the historical questions, it can also exacerbate our frustration by raising other questions. For instance, the standard tabulation of the 657 texts that Xuanzang 玄奘 brought back to China in 645 states that among them were thirty-six hetuvidyā texts. He himself only translated two of these, the Nyāyapraveśa and Nyāyamukha.6 He translated one verse of the Pramāasamuccaya (see below) and paraphrased a few others. That raises the question: What were the other thirty-three texts? If we assume they might have included non-Buddhist hetu-vidyā works, like the Nyāya sūtra, we would still be hard pressed to name with confi dence thirty-six treatises on logic from that period. Did these include the three or fi ve logic (or proto-logic) texts sometimes ascribed to Vasubandhu (which, aside from a portion of the Vāda-vidhi preserved in Tibetan, are no longer extant)?7 Commentaries on Dignāga, such as Dharmapāla’s 護法 commentary on Ālambana-
parīk•ā later partially translated by Yijing8 (if this text can be counted as a hetu-vidyā text at all)? Even allowing such generous speculation, we can only conjure up perhaps a dozen or so possible titles of texts we are aware of — what were the others?9 Vaiśe•ika texts?10 Unknown Nyāya texts? Mīmāsa texts? Vātsyāyana? Praśastapāda? Śabara? Buddhist texts about which we know nothing today? Clearly, we are missing a lot of context as to what transpired in the hetu-vidyā efforts between the time of Asaga and Dharmakīrti, such as Vasubandhu's "missing" hetu-vidyā texts — and shouldn’t assume that these thirty-six texts, whatever their contents, constituted the complete set of this genre at that time in India. It could have been two, three, or a hundred times larger. We don’t know.
Most of the Indian pre-Dignāga works on hetu-vidyā are no longer extant. While we can easily surmise that Dignāga was strongly infl uenced by Vaiśe•ika thinkers,11 the lack of a single surviving Vaiśe•ika text between the founding Vaiśe•ika sūtra (date uncertain) and Praśastapāda (ca. sixth c.), leaves the precise details of Vaiśe•ika infl uence on Dignāga unclear. What is clear is that Praśastapāda, who was either Dignāga's contemporary or came shortly after, radically altered Vaiśe•ika thought,12 and it is his version of the Vaiśe•ika system that subsequently became the prevailing, orthodox Vaiśe•ika view. Dignāga shows no sign of being aware of Praśastapāda; thus what we know as Vaiśe•ika is not what Dignāga knew and responded to.
1. Trikāla-parīk•ā 14
3. Ālambana-parīk•ā 15
Yijing could not have intended this list as exhaustive, since he himself translated an additional Dignāga text, Hastavālaprakaraa (掌中論 Zhangzhonglun, T.31.1621, tr. in 702), which already had been translated into Chinese in the mid-sixth century by Paramārtha (解捲論 Jiejuan lun, T31.1620).16 Generally, of Dignāga’s writings, these eight are considered his hetu-vidyā texts. Of the eight texts Yijing lists, he himself translated four,17 two of which had been translated before, one (Ālambana-parīk•ā) twice before.18 Yijing also tells us that Dignāga had a reputation as a poet in India.19
So, of the eight “main” texts listed by Yijing, four exist in Chinese translation (Nyāyamukha, Upādāya-prajñapti-prakaraa and Sāmānyalak•aa-park•a survive only in Chinese translation), and two (Trikālaparīk•ā and Pramāasamuccaya) exist in complete form only in Tibetan translations. That leaves two texts — *Hetu-mukha and Hetvābhāsa śāstra — of which we know nothing except their titles.20
In addition to preserving two translations of the Nyāyamukha, the Chinese tradition preserves other precious information. Both Xuanzang and Yijing clearly identify Dignāga as a Yogācāra. Since, unlike Xuanzang, Yijing did not identify himself as a Yogācāra, this should be given credence, especially since Indian texts of this period, such as Bhāvaviveka’s Madhyamaka-h•daya-kārikā ch. 5 and Candrakīrti’s Madhyamakāvatāra ch. 6, target Dignāga as a prominent part of their critiques of Yogācāra. We can date Dharmakīrti by the fact that Xuanzang, who returned to China in 645, never mentions him, while Yijing, who arrived in East India in 673, informs us that Dharmakīrti has become an important component of the curriculum at Nālandā. Dharmakīrti’s texts must have emerged into prominence during the interim.
In addition to his translations of Dignāga, Xuanzang’s translations of other texts, notably the Cheng weishilun 成唯識論 and Buddhabhūmyupadeśa 佛地經論 (Fodijinglun, hereafter BBh-U), include discussions that illustrate how, prior to Dharmakīrti, some Indian Buddhists, i.e., Yogācāras, understood and deployed Dignāga. These discussions will be the focus of this essay. The Buddhabhūmy-upadeśa (BBh-U)
What does the world look like through enlightened eyes? How, if at all, does perception for enlightened beings differ from the way nonenlightened beings perceive? These would seem to be natural questions, especially considering the great emphasis Buddhists place on such themes as mental purifi cation, correcting cognitive errors, theories of perception, “seeing things as they are” (yathā-bhūtam), etc. Thus it is surprising that detailed and specifi c discussions of how enlightened beings perceive almost never appear in Buddhist literature, aside from attributing vague, honorifi c adjectives to such cognitions, such as labeling them “transmundane” (lokuttara 出世間), pure, unobstructed, etc. The section of the Buddhabhūmy-upadeśa that is translated and discussed below is a major exception.
The BBh-U is a composite of several commentaries (upadeśa) on a sūtra called the Buddha-bhumi sūtra.21 Exactly how many commentaries contributed to the BBh-U compilation and who the various authors were is uncertain and a matter of dispute, as will be explained shortly.
Xuanzang translated the Buddhabhūmi sūtra in 645, the fi rst year of his translation activity once he returned to China from India, suggesting he assigned it high priority. He translated the BBh-U four years later, in 649. It is attributed to “Bandhuprabha, etc.” 親光菩薩等 (or 親光等菩薩). It presents confl icting opinions that were debated between different Indian Yogācāra thinkers on a variety of topics during the sixth to
early seventh century using the root text, the Buddhabhūmi sūtra, as its springboard. When alternate theories for understanding a term or a passage of the Buddhabhūmi sūtra are presented, often three distinct and usually confl icting theories are offered.22 The section of BBh-U translated and discussed below is a clear example of this. This has led some to think of BBh-U as consisting of, at most, two or three commentaries. That, however, is doubtful.
A Tibetan translation23 of a commentary on the Buddhabhūmi sūtra, attributed in the Tibetan version to Śīlabhadra (ngang tshul bzang po; 戒賢), frequently corresponds to what looks like the core commentary contained in the BBh-U. Śīlabhadra was the head monk at Nālandā — the leading Buddhist university in the ancient world — when Xuanzang arrived there (ca. 637) during his pilgrimage to India. Roughly half of the Chinese BBh-U does not correspond to the so-called Śīlabhadra commentary, and of that non-corresponding half, several large passages
reappear, almost verbatim, in the Cheng weishi lun. Kuiji 窺基, Xuanzang’s disciple and successor, in his commentaries on the Cheng weishilun, attributes some of those shared passages to Dharmapāla 護法, usually without drawing attention to their parallels in BBh-U (he may not have been fully aware of them). Because of the parallels between these texts, some modern scholars have argued that the core commentary is by Śīlabhadra, while the rest, or most of it, is by Dharmapāla, an important sixth century Yogācāra.24 Bandhuprabha, usually understood in the East Asian tradition to be a disciple of Śīlabhadra, may have only been a compiler of others’ commentaries, or he may have been one of the contributing commentators. There is no easy way to decide this.
As to the number of commentaries drawn on for the compilation, that also is uncertain, as mentioned above. The passage translated below clearly consists of three distinct theories that not only offer alternate interpretations of the ideas under discussion, but even propose alternate readings of individual words found in the sūtra. They do not directly engage the other theories in arguments; they simply offer alternate interpretations, giving the impression that these were independent commentaries, not subcommentaries on a primary commentary. Believing that Śīlabhadra is the
root commentator and Dharmapāla is the capping commentator, one might assume that the fi rst and third theories, respectively, presented in the passage to be discussed below were authored by them. However, when we look for a parallel section in the Tibetan version attributed to Śīlabhadra, we fi nd none. Of the entire extended debate on Dignāga and the four-bhāga theory offered in the Chinese text, there is not a hint in the Tibetan.25 So the supposed “core commentary” attributed in the Tibetan to Śīlabhadra is not the source for any of the three theories laid out in the Chinese. In the numerous other places in BBh-U where alternate theories are offered, typically they also come in threes (though not always), one of which is sometimes an opinion corresponding to the Śīlabhadra text. That leads me to tentatively conclude that BBh-U incorporates an indefi nite number of Buddhabhūmi commentaries, of which there are a minimum of four distinct commentaries, one of which is preserved independently in Tibetan.
Whether or not Śīlabhadra and Dharmapāla are the actual authors of some of these commentaries, it is reasonable to assume that the positions discussed represent Yogācāra debates of the sixth to early seventh century, a time when Dharmapāla, etc., were prominent. The sometimes striking difference of opinion expressed by the different commentators demonstrates the great diversity between Yogācāras of that period, quite unlike the near homogenous
impression usually promoted by doxographers. Those who think they know what Yogācāra thought consisted of during this period will perhaps fi nd the second theory shocking. Though surprising, it is clear that the second theory was indeed a Yogācāra theory, since correlates can be found throughout the Cheng weishilun, though space precludes me from documenting that here in detail.
As for the title, bhūmi 地 can mean either "land," or "stage," hence the sūtra's title suggests both "Sūtra on the Buddha Land" and “Sūtra on the Stage of Buddhahood”; the contents of the sūtra fi t both readings, and the preamble of BBh-U plays on both meanings, as if treating the title as a double entendre. BBh-U and Āśraya-parāv•tti
The central concern of the Buddhabhūmi sūtra is the ‘overturning of the basis’ (āśraya-parāv•tti 轉依) of the eight consciousnesses (vijñāna 識), so that they are transformed into the four cognitions (jñānas 智), also described as a purifi cation of the consciousness stream and the manner of cognition, from contaminated or polluted (āsrava 有漏) to uncontaminated (anāsrava 無漏).
In Yogācāra thought the eight consciousnesses are (1-5) the fi ve sensory consciousnesses (seeing, hearing, smelling, tasting, and touching; pañca-vijna-kya 五識身); (6) mental-consciousness (manovijna 意識) which cognizes thoughts as well as takes cognizance of what the previous fi ve consciousnesses sense; (7) manas (末那 or 意), the sense of self-hood; and (8) the [[warehouse
consciousness]] (ālaya-vijñāna 阿賴耶識), also called "all-seeds consciousness" (sarva-bījāka-vijñāna 一切種識), "karmic maturation consciousness" (vipka-vijñāna 異熟識) and "foundational consciousness" (mūla-vijñāna 本識), because it holds the contaminated and uncontaminated seeds, bringing them to karmic maturity and fruition.
When transformed through āśraya-parāv•tti, starting with the eighth, the warehouse consciousness becomes (1) Great Mirror Cognition (mahādarśajñāna 大圓鏡智); manas becomes (2) Equalization Cognition (samatājñāna 平等性智); mental-consciousness becomes (3) Attentive Cognition (pratyavek•anā-jñāna 妙觀察智); and the fi ve sensory consciousnesses become (4) Accomplishing Activity Cognition (k•tyānu•hāna-jñāna 成所作智). Their cognitive fi eld is the Dharma-dhātu (法界). While, prior to āśraya-parāv•tti, the warehouse consciousness 藏識 superimposes habitual tendencies (vāsanā 熏習) into perception, according to the Buddhabhūmi sūtra, the Great Mirror Cognition contains the images of all things, equally, without attachment. While manas views the world in terms of “me” and “others,” valuing “myself” above “them,” Equalization Cognition sees all as the same. Mental-consciousness is easily distracted, but Attentive Cognition remains effortlessly focused. The Accomplishing Activity Cognitions perceive things just as they are thereby enabling one to accomplish all tasks. When all contaminations and obstructions have been removed from the consciousnesses and the uncontaminated seeds reach fruition, the Four Cognitions replace the consciousnesses; that is enlightened perception.
Importance of the BBh-U Passage
1. It explicitly discusses two important texts by the Buddhist epistemologist and logician Dignāga 陳那 (late 5th-mid 6th c.) — namely his Pramāasamuccaya 集量論 and Ālambana-parīk•ā 觀所緣緣論26 — providing us with a rare glimpse of how some Buddhists were utilizing and interpreting those texts prior to Dharmakīrti (ca. 600-665).
2. It attempts to explain how cognition works after overturning the basis.
3. It demonstrates that the Yogācāras of that time all presupposed some sort of correspondence theory — though they differed on the details. For each issue that is raised in this section, three distinct theories are offered, illustrating a greater diversity among Yogācāras of this period than is usually assumed.
4. The passage responds directly to arguments given by Nāgārjuna in his Vigraha-vyāvartanī and Mūla-madhyamaka-kārikā.27 In both texts Nāgārjuna argues that light neither illuminates itself nor others; in the former text he does so specifi cally to criticize the claim that means of knowledge (pramāa 量, 能量) can provide a validly known object (prameya 所量). In direct response to Nāgārjuna, BBh-U argues that consciousness can make both itself and other objects known, i.e., consciousness can be a cognitive object for itself.
i. the content or image part (nimitta-bhāga 相分); ii. a seeing part (darśana-bhāga 見分);
iii. a self-refl ective or ‘being aware of itself’ part (svasavittibhāga 自證分); and
While the theory of the fourth component largely disappears from Buddhist consideration in India once Dharmakīrti provides a more sophisticated version of the fi rst three components,29 it became important in East Asian Buddhism, primarily due to its appearance here and in an expanded discussion in the Cheng weishilun 成唯識論 (Treatise Establishing Consciousness Only), a foundational text of the Weishi Chinese Yogācāra school 唯識瑜伽行派.30 A short excerpt from the Cheng weishilun will be discussed after the BBh-U passage to illustrate some of the additions it offers.
6. It is a prime example of how commentarial style can utilize the declarative statements of a text such as the Buddhabhūmi sūtra to fashion a philosophical discussion. Where opposing theories are presented, each builds its case by interpreting the same key terms (e.g., nirvikalpa-jñāna, *nirākāra, etc.) in its own way, illustrating that these terms were never univocal but always available for a variety of meanings.
7. Since BBh-U is an early, relatively clear formulation of a set of concerns that continued to be of interest to later Buddhist philosophers, who increasingly complicate their own formulations with ever greater sophistication and intricacy, the relative simplicity and clarity of this earlier version may shed some light on the background models behind the formulations of those later thinkers.
8. It is one of the earliest texts to address the issue of whether enlightened cognition is imageless (nirākāra 無相) or involves images (sākāra 有相). The BBh-U argues for the latter position. The nirākāra-vāda vs. sākāra-vāda controversy became more prominent later on in India, and continued to be debated for centuries in Tibet.
After arguing that consciousness can know itself, BBh-U turns to the four components theory, attempting to show how consciousness can know itself without incurring an infi nite regress. Finally it turns to the question of the status of the images (nimitta, ākāra; ābhāsa, nirbhāsa, pratibhāsa, sād•śya 相；似) that appear in consciousness.
Three different theories are offered. Underlying all three is a theory of perception widely found in India and throughout the ancient and medieval world, called prakāśa 照, "illumination," in which a light is believed to go out from the eye and shine on an object (bimba 質) illuminating it, the refl ection (pratibimba 影像) bouncing back to the mind (although the second theory claims the prakāśa theory, since it involves a mediation process, only applies to unenlightened cognition; enlightened cognition is immediate). For contaminated or unenlightened cognition, prakāśa also entails obstructions (āvaraa 障, 障礙), attachments (abhiniveśa 執著), imaginative distortions and overlays (vikalpa, parikalpa 分別, 遍計所執, 執計), and active pursuit or effort (逐). In ordinary perception a grasper (grāhaka 能取) grasps or apprehends (grāhaa, upalabdhi 執著, 取, 所得) a ‘grasped’ (grāhya 所取), i.e., ordinary perception is an act of appropriation, grasping.
For uncontamined cognition, the fi rst theory states that the mind becomes a replica (sād•śya 似) of whatever is in front of it (親照前境), without imaginative construction (nirvikalpa 無分別), like a mirror effortlessly refl ecting what is in front of it. This theory holds that uncontaminated cognition is similar to contaminated perception, except it is devoid of attachment and grasping. The second theory emphasizes that things are seen just as they are; it is not like a mirror that only receives refl ections, nor like a light going out in search of an object; the object itself is immediately known, without grasping or pursuing, such that cognition directly perceives sensory forms (rūpa 色) without obstruction. The third theory has the replica arise from the mind’s uncontaminated seeds (無漏種起).
Indian Buddhists used a rich, nuanced vocabulary for aspects of cognition and types of cognitive objects, with fi ne distinctions that are often lost in translations that render a host of different terms reductively as either "subject" or "object." For instance, while different texts and systems may describe the following terms in different ways, most Buddhists would agree with the following generic descriptions, which are frequently found in Yogācāra texts. An ālambana 所緣 (which I leave untranslated below) is a cognitive object from which mental impressions are derived. An ākāra 相 is a mental image or mental impression drawn from the ālambana, or sometimes the term is taken to indicate the characteristics of an object, such as shape, size, etc.31 A vi•aya 境，境界 is a sense-object (a color, sound, etc.) or a sense-domain (e.g., visible sphere, auditory sphere, etc.). Nimitta 相 is a cognitive object whose characteristics cause a perception resembling it to arise.32 A tattva 真實 is something irreducible and real. A vastu 事 is an actual thing that may underlie a cognition, though whether it is perceived as it is or obstructed by imaginative constructions depends on the extent to which one’s cognitive abilities are purifi ed of contaminants.
The quality of BBh-U’s arguments are crude compared to later developments, but, as a comparison with Williams (1998) and the sources he discusses would demonstrate, the later tradition basically reworked and reiterated the arguments already found here, dropping the fourth component (the svasavitti-savitti) while refi ning and fi ne tuning the rest.
The Cheng weishilun provides a similar description of the four components of perception while adding a few additional wrinkles, one of which is to point out that all four components by and large reduce to the second. The passage from Cheng weishilun explaining that has been included below. In what follows I have resisted as much as possible any attempt to repackage these texts into current philosophical idioms, or to embed them in extended paraphrase. While some of the assumptions and technical terminology these texts employ may be unfamiliar to readers today, staying as close as possible to their way of expressing themselves should help us think with them rather than for them. I will provide whatever background and exposition is necessary to recognize and follow the main arguments, but will refrain from converting their style of argument into our own contemporary style. The Buddhabhūmy-upadeśa (Fodijing lun) Passage
(Translation and Discussion)33
[[[Dignāga’s]]] Pramāasamuccaya says that all citta and caittas are aware34 of themselves; (this self-awareness) is called ‘perception’ (pratyak•a). If that were not the case, there would be no memory, [so that to perceive something would be] just as if [the thing] had never been seen. Hence each and every mental component associated with the Four Cognitions also illuminates (i.e., perceives35) and knows itself. 集量論說。諸心心法皆證自體。名為現量。若不爾者。如不曾見不應憶念。是故四智相應心品。一一亦能照知自體。
The opening lines of this passage pose an intriguing problem, even before we begin to examine what they mean. The passage appears to be offering a direct quotation from Dignāga’s Pramāasamuccaya (集量論說 “Pramāasamuccaya says…”), but the seemingly quoted line (“all cittas and caittas are aware of themselves”) is not found in any extant version of Pramāasamuccaya (hereafter PS).36 Nonetheless, the underlying Sanskrit line translated here is known. However, it does not appear in the Pramāasamuccaya or in any other known work of Dignāga; it appears, instead, in Dharmakīrti’s Nyāyabindu:37 sarva-citta-caittānām ātmasavedanam (Nyayabindu 1:10). Note that in Dharmakīrti’s Sanskrit version, instead of the term sva-savitti or sva-savedana, the sva- (“self, one’s own, itself”) prefi x is replaced by ātma- (“self,” “essentially”).
The sva- in sva-savitti is ambiguous, susceptible to a number of differing interpretations. The standard Dharmakīrtian meanings, which include such senses as a “momentary, discrete, unique particular,” or a refl exive sense of “one’s own,” and so on, as entailed in Dharmakīrti’s treatment of svalak•aa, is typically read back into Dignāga’s system. But Dignāga was on the cusp between earlier, specifi cally abhidharmic usages and these later, Dharmakīrtian pramāa-vāda usages. For instance, in typical abhidharma usage, the difference between sva-lak•aa and sāmānyalak•aa is defi ned differently than the later pramāa-vāda meanings. As explained by Dhammajoti:38
… the relativity as regards the notions of svalak•aa and sāmānyalak•aa is to be noted. Thus, among various rūpa-s different colours, different shapes there is the common nature of being resistant and subject to deterioration. Accordingly, this svalak•aa of a rūpa is distinct from a vedanā, etc. But, at the same time, it is also the sāmānya-lak•aa of these various types of rūpa-s. Similarly, the Great Element, Earth (p•thivī), is both svalak•aa and sāmānya-lak•aa it is said to be a svalak•aa in contrast to the other three Great Elements; and a sāmānya-lak•aa, since all Earth Elements have the characteristic of fi rmness. In this way, the [Mahāvibhā•ā] declares, “there are infi nite distinctions [that can be made] between svalak•aa and sāmānya-lak•aa.” The Sarvāstivāda Ābhidharmikas distinguish two kinds of svalak•aa: The fi rst, dravya-svalak•aa, is the intrinsic characteristic of the dharma as a unique entity in itself, for instance: that of a particular color, say, blue. The second, āyatanasvalak•aa, refers to the intrinsic characteristic of the dharma as a member of a unique class an āyatana of which it is a member, for instance: the particular blue color as a unique class of dharma-s know as “visibles” (rūpa), i.e., the rūpa-āyatana. We can see from this example that, in this context, the āyatana-svalak•aa is, in a sense, a common characteristic in relation to the dravyasvalak•aa. It is for this reason that the [Mahāvibhā•ā] states:
From the point of view of dravya-svalak•aa, the fi ve sensory consciousnesses (pañca-vijñāna-kāya) also take sāmānya-lak•aa as their object (ālambana). But from the point of view of āyatanasvalak•aa, the fi ve sensory consciousnesses take svalak•aa alone as their object.
The categories of svalak•aa and sāmānya-lak•aa continued to undergo revisions over succeeding centuries, sometimes by opposing schools who developed differing defi nitions and usages. Such developments and modifi cations were not undertaken merely for formal exercise, but because important doctrinal implications were entailed, buttressed by an environment of heated and critical debate. Even before Dignāga, Asaga had already defi ned pratyak•a as the exclusion of kalpanā (conceptual constructions). Dignāga’s svalak•aa may have more in common with such developing abhidharma senses — even if primarily in reaction to them — than with the later Dharmakīrtian senses. He never, for instance, states that a svalak•aa is a unique, momentary particular, as does Dharmakīrti. Likewise, Dignāga’s sense of sva-savitti may involve a comparable sense of sva-. Sva-savitti might mean that perception, in order to be perception, entails that it always is someone’s perception, that it has a self-awareness of being aware, of taking note of what it perceives.39 For instance, when Husserl defi nes intentionality as meaning that consciousness is always “consciousness of,” i.e., that it necessarily has content, this would be similar to the view Dignāga criticizes (at PS I.13seq) that is attributed to Vasubandhu, namely that the ‘object’ determines the perception, whereas for Dignāga, the only necessary condition (though not necessarily always its only condition) is consciousness itself, i.e., of the perceiver. This is what gives perception the sense of being “my” perception, localized and particular to me, even if the “objects” perceived within it are public and shared. I remember my perceptions, not yours. I have my memories of our conversation, not yours. Perception is a type of witnessing (sak•in), and to the extent that witnessing takes cognizance of its witnessing, it is svasavitti. Seeing blue is to take cognizance of blue. This taking cognizance of blue is svasavitti. It is an apodictic sensibility, with possibly an implicit veridical dimension.40 I know the water is cold because it feels cold; the coldness of the water is “perceived.” As far as I know, only one actual verse of PS was ever translated into Chinese in the classical period, and that is PS 1:10, which is the verse in which Dignāga equates prameya, pramāa, pramāa-phala.41 This appears in the Cheng weishilun (T.31.1585.10b):
Steinkellner’s Sanskrit for this verse reads:42
yadābhāsa prameya tat pramāaphalate puna | grāhakākārasavittyos traya nāta p•thak k•tam || 10 ||
Hattori’s translation (from the Tibetan):43
k.10. whatever the form in which it [viz., a cognition] appears, that [[[form]]] is [[[recognized]] as] the object of cognition (prameya). The means of cognition (pramāa) and [the cognition which is] its result (phala) are respectively the form of subject [in the cognition] and the cognition cognizing itself. Therefore, these three [factors of cognition] are not separate from one another.
This verse is often cited in later Chinese Buddhist literature, or, more accurately, the part most often cited is the last pada, 此三體無別, which stresses the non-separation of the three parts of consciousness.
In the Nyāyabindu the line sarva-citta-caittānām ātma-savedanam occurs as a description of the third of the four types of perception enumerated by Dignāga and Dharmakīrti,44 the one explicitly concerned with svasavitti. Dignāga’s explanation of the third type of perception suggests that svasavitti here indicates a form of introspection — specifi cally observing kleśas such as lustful, aversive, etc., adverse mental conditions — as a possibly detached witness.45 While the kleśas being observed are kalpanā, and hence not perception properly speaking according to Dignāga’s defi nition, the witnessing of these kleśas, e.g., becoming aware of anger, is perception. This has implications for Buddhist practice (sm•ti, mindfulness), and is not simply an abstract formulation designed to meet some conceptual requirement.
But how did a line we fi nd in Dharmakīrti get attributed to Dignāga’s PS by the BBh-U? There seem to be three possible explanations: (1) BBh-U is mistakenly quoting Dharmakīrti for Dignāga; this seems unlikely. (2) The line in question appeared in a version of PS that is no longer extant; this, too, seems unlikely. (3) This line is a paraphrase of PS ch. 1 that became a stock characterization of Dignāga’s argument, perhaps introduced by the Buddhabhūmi commentary from which BBh-U acquired it, which Dharmakīrti subsequently adopted (and reinterpreted) for his Nyāyabindu. Since Xuanzang shows no other signs of being aware of Dharmakīrti or his writings (though they were almost exact contemporaries and may have been at Nālandā at the same time),46 I suspect that this may be an otherwise unknown “missing link” between Dignāga and Dharmakīrti, i.e., a characterization of Dignāga’s theories that was suffi ciently popular in the early seventh century so as to have been adopted by Dharmakīrti.
If this line is indeed a paraphrase of PS ch. 1, that raises the question of how well does it portray Dignāga’s project. The line is continued (and this is not in the Nyāyabindu) with “If that were not the case, there would be no memory, [so that to perceive something would be] just as if [the thing] had never been seen.” What is being argued is that in order for there to be memories, citta and caittas must be aware of (or “experience,” savedana) themselves (ātma-). Svasavitti or ātma-savedana are necessary conditions for memory.47 This initially seems to allude to PS 1:11.48 However, the BBh-U is here actually paraphrasing a thematic argument that weaves throughout the entire fi rst chapter of PS which repeatedly returns to the problem of memory. Dignāga's intent in PS I is still a matter of some discussion. The BBh-U seems to understand self-awareness (svasavitti 自證49) at this point as something integral to perception, since, if one is not aware of perceiving something, there can be no memory of it in the form of “I remember X.”
Put in different words, a perception has to “register” in consciousness, to be noticed in such a way that one is aware that one is noticing it; otherwise it would only be a momentary sensation, disjointed and disconnected from other sensations, and thus, subsequently for the one who perceived it, “as if” it had never been perceived at all, i.e., it could not be remembered. One of the things Dignāga is trying to accomplish is to provide a defi nition of perception that fi rmly differentiates it from conceptualization (kalpanāpoha 無分別50), an agenda already pursued by Asaga in his treatment of hetu-vidyā in the Yogācārabhumi.51 Thus, svasavitti (self-awareness) has the role of providing perception with a mechanism that registers sensations and yet avoids including any sort of conceptualization (kalpanā). It is a type of bare witnessing. Another sense of svasavitti at play is as a correlate to and development from the idea of “experiencing something for oneself” already found as a necessary component of attaining or realizing (sacchikaroti, sacchikiriyā, adhigama, etc.) the higher attainments along the Path in Early Buddhism.52 “Registering” is a basic condition for memory, i.e., remembering that X had been perceived. The remembered image that is conjured up in a memory is derived from the prior svasavitti that registered it, that impressed it into one’s consciousness stream. Svasavitti is also involved with inner perception, i.e., introspective observation.
Additionally, svasavitti seems to imply a partitioning of cognition into components which go beyond the dyadic grasper and grasped relation (grāhaka-grāhya), or as Dignāga puts it, a grasper of an image (grāhakaākāra) and another cognitive factor which takes cognizance of that (savittyos = svasavitti). There is the perceptual act and the taking cognizance of it, being aware of it. This partitioning of cognition into (1) grasper of (2) an image of which (3) one is aware is what BBh-U will call the fi rst three bhāgas, viz. nimitta, darśana, and svasavitti.
The assertion that cognition (vijñāna) both knows itself as well as other “objects” is often supported by an analogy to light which is said to “illuminate itself as well as other things.” Many Indian schools, Buddhist and non-Buddhist, accepted that analogy and characterization of cognition as axiomatic, though it was challenged by some others, in the Buddhist domain most notably by the Madhyamakas. BBh-U continues by acknowledging and responding to this challenge. [Objection:] Doesn't this contradict how the world works? A knife doesn't cut itself and a fi ngertip cannot touch that [same] fi ngertip. [Reply:] Don't you see that lamps, etc., are able to illumine themselves?
[Objection:] How do you know that lamps, etc. illuminate themselves?
[Reply:] When in perception one sees the absence of darkness, the light, being separate [from darkness], is clearly perceived. If [the lamps, etc.] didn't illuminate themselves they would be obstructed by darkness, and so not seen in perception. Due to this, therefore, know that lamps, etc. illuminate themselves.53
[Objection:] Lamps, etc. are not dark. Is it necessary [that, in addition, they would have to further] illuminate [themselves]? [Reply:] This is just like jars, cloth, etc. Although in themselves they are not darkness, in the absence of a lamp, etc. to illuminate them, they are encompassed by the obstruction of darkness, so one cannot see them in perception. When lamps, etc. illuminate them, [the light] clears away that encompassing darkness, making [those things] visible to perception. We call that “illumination.” Lamps, etc. are the same case. When their self nature [to illuminate] arises, the encompassing obstruction of darkness is cleared away, making them visible to perception; therefore this is called “selfillumination.”
The implicit argument that the Yogācāras seem to be making is that consciousness does indeed "illuminate" (prakāśa) itself as well as other things, but in order to do so, it must partition itself into a seer (darśana) and something seen (ālambana). We will return to this.
This partitioning may be part of the reason that Xuanzang frequently contrasts nengyuan 能緣 and suoyuan 所緣 when translating the single term ālambana, using one or the other Chinese term as he sees appropriate to convey the nuance at work. We will return to this later.
Explanations of consciousness that break it into parts that watch each other — starting with breaking consciousness itself into seeing and seen partitions — raise the spectre of an infi nite regress: If a part of consciousness is viewing another part of consciousness that is viewing something else, is there still another part that is watching the part that is watching the viewing consciousness? Whatever it is that requires that there be a watcher who watches the watcher, this could, and perhaps should also require that there be another watcher watching the watcher of the watcher, and so on, ad infi nitum. The BBh-U and Cheng weishilun are aware of this potential problem, and address it, but not to the satisfaction of the later Buddhist philosophers, who proposed additional arguments to solve this. We will see the BBh-U’s “response” shortly.
Citta and caittas, regardless whether dominant or weak, are all able externally to [[[Wikipedia:cognize|cognize]]] cognitive-conditions and internally to be aware of themselves (svasavitti). This is analogous to light actually illuminating others as well as illuminating itself. It is unlike (i.e., not analogous to) such things as knives, etc., which are of a different sort.
I will leave unremarked the use of "externally" here (or more exactly, the clear contrast being drawn between cognizing externals and inner selfawareness), though, given what follows, this is signifi cant.
It may be useful to observe the possibility that for Dignāga svasavitti was not a separate “partition” (bhāga 分) of consciousness, but rather, as he states in PS I, a prajñaptic (假名) or metaphoric distinction one draws by conceptual abstraction. That is, the svasavitti is the cognitive moment itself, it is consciousness. To my knowledge the term bhāga does not occur in this context in the Sanskrit version of Dignāga’s text that we are recovering from Jinendrabuddhi. If it is the case that bhāga is an addition only found in the Chinese, the fen 分 would be either Xuanzang's own addition or gloss to clarify the meaning, or else, it may refl ect an interpretive usage he encountered in India that he adds to his translations, one which arose with the idea of an "additional" bhāga to supplement the svasavitti. Such additions are frequent occurrences in Xuanzang’s texts, though he never announces them as such. To roughly characterize citta and caittas, each is said to have two parts–an image part (nimitta-bhāga) and a seeing part (darśanabhāga). In Pramāasamuccaya [[[Dignāga]]] explains that citta and caittas all have three parts: 1. a part that is grasped (grāhya),
3. a part that is self-aware (svasavitti).
(2) to argue that despite the fact that the word pramāa grammatically implies an instrument (pra √mā + ana), that usage is only a fi gurative characterization for what is actually the consequence or result (phala) of the process of knowing, viz. coming to know the intended object (artha), so that “knowing” is actually pramāaphala, i.e., what, retrospectively, can be considered the effect of the pramāa process.
For Dignāga, the “instrument” or means of knowledge is a secondary, conceptual abstraction; pramāa, therefore, properly speaking, refers to the act of knowing, not the means. As Dignāga puts it, the pramāa is simply the effect (phala) in which the object appears, in which one takes its “measure” (vi•ayābhāsataivasya pramāa tena miyate). The “appearance of the known” (ābhāsam prameyam) is another way of saying “the result of the activity of coming to know” (pramāa-phala). (i) The grasper (grāhaka), (ii) the mental image (ākāra) and (iii) being aware that one is knowing that image (svasavitti), are three ways of talking about the act of knowing. Prameya and pramāa, thus concludes Dignāga, are linguistic-conceptual distinctions one might retroactively apply, by analysis, to a cognitive act, but, properly speaking, only the pramāa-phala is the actual knowledge-event.
The BBh-U, however, treats pramāa-phala as only equivalent to svasavitti, which indeed is an aspect of Dignāga’s treatment of svasavitti, but when isolated from the full discussion, this may be a distortion. For Dignāga, the known part and the knower, as well as the svasavitti, are all included in pramāa-phala. Xuanzang's translation of pramāa here as 能量 neng liang (lit. “able to measure”) does not fully refl ect the issue that Dignāga is raising, which is that the term pramāa is a grammatical construct indicating an instrument (pramā + ana = instrument of pramā, ‘cognitive knowledge’). Neng liang does refl ect the etymologies of pramāa and mīyate as “measure” (量 liang). Neng 能, especially when contrasted with 所 suo (as in 所量 suo liang, “what is measured, known,” prameya) suggests an active or subjective sense (neng) as opposed to a passive or objective sense (suo), i.e., neng liang implies pramāt• (the knower) rather than pramāa (means of knowledge). Dignāga is trying to argue that we should understand the term pramāa counter to its grammatical implication as an instrument, and instead treat it as a conceptual abstraction derived from retrospective analysis of the effect. The effect is, strictly speaking, the actual cognition of an artha 義 in consciousness.
smārtābhilā•ika ceti pratyak•ābha sataimiram | savyāpārapratītatvāt pramāa phalam sat || 8
svasavitti phala vātra tad-rūpo hy artha-niścaya | vi•ayābhāsataivāsya pramāa tena mīyate || 9
yad-ābhāsa prameya tat pramāa-phalate puna | grāhakākārasavittyos traya nāta p•thak k•tam || 10
The Four Bhāgas
The fi rst two are external [in terms of their cognitive object], the latter two are internal [in that their cognitive objects are other parts of consciousness]. The fi rst is only a 'known' (jñeya), the rest include two types [i.e., known and knower]. That is, the second part only knows the fi rst. Sometimes this is a valid cognition (pramāa), sometimes an invalid cognition (apramāa); sometimes a perception (pratyak•a), and sometimes an inference (anumāna). The third is aware of itself being aware of the second and it is aware of the fourth. The fourth is aware of itself being aware of the third. The third and fourth are classifi ed as valid perception (pratyak•a-pramāa).
The nimitta is the external object. Darśana looks outward toward nimittas. A darśana cognition may be either pramāa (valid, apodictic) or apramāa (invalid, erroneous), either a perception or an inference. The last two bhāgas, on the other hand — svasavitti and *svasavittisavitti — are invariably valid. This may initially appear to contradict what I previously suggested, which was that for Dignāga the notion of a pramāa did not yet carry the weight of providing veridical validity, since pramāa is acquiring a sense of justifi cation or validation, distinguishing cognitions that are “valid” as opposed to “invalid.” According to BBh-U, while a darśana may sometimes be apramāa — i.e., one might misperceive an object (e.g., mistake a rope for a snake), or accept an idea derived from faulty reasoning — the two internal bhāgas — svasavitti and svasavittisvasavitti — cannot be apramāa. Presumably this is because the typical obstructions or hindrances that might produce an erroneous cognition of an external object — such as lack of light when trying to see something visible, or a barrier between the sense-organ and senseobject blocking sense-contact, etc. (issues already thoroughly debated in abhidharma texts — cf. Dhammajoti, 2007, Abhidharma Doctrines and Controversies on Perception, University of Hong Kong — and discussed at length by Asaga in his Yogācārabhūmi hetu-vidyā exposition) — would not occur when one is internally experiencing an inner feeling. Whether one understands the reasons why, or has any concern that one's temper has fl ared up, one feels angry, and that sense of anger is immediate, preconceptual. That is svasavitti, and that feeling itself, since direct and immediate, is apodictic. One doesn’t have to “prove” to oneself — much less to others — that one is angry. That is self-evident (svasavitti). Introspection of this sort — which does not mean inner analysis or conceptual investigation of inner states — is invariably self-evident, and hence pramāa: indubitable, immediate, and given with the cognition itself. “Justifi cation” or validity would be kalpanā (conceptual) for Dignāga, not perception. Svasavitti provides this sort of self-evidential immediacy, not justifi catory validation.
BBh-U, nonetheless, is looking to it for validation (as do many contemporary scholars). The emergence of this desideratum in BBh-U is something new, something that BBh-U will deal with crudely, but which subsequently will receive increasingly sophisticated treatment by Dharmakīrti and his successors. BBh-U offers us a glimpse of the sorts of problems being posed just prior to Dharmakīrti’s advent, problems that Dharmakīrti would soon be heralded for solving or for moving toward solutions. Dignāga has provided them with their basic categories, with a basic framework, but they fi nd his formulation lacking, so much so that they have to add an additional bhāga, one which they think will do the job of justifi cation they feel is now necessary but which Dignāga failed to adequately provide. If, as they wish, svasavitti should take on a justifi catory role, it seems equally clear to them that svasavitti as formulated by Dignāga does not satisfy that role.
We are also told in this passage that three of the four bhāgas are witnessing — and hence grounding — the other bhāgas, such that none remains ungrounded by something other than itself.
If, as I suggested above, Dignāga’s svasavitti is better understood as an integral aspect of consciousness, rather than as a distinct partition, then, one might ask, doesn’t svasavitti cognize itself? If so, why would it be necessary to insist that it cannot do that, that it can only cognize the second and fourth, while now a fourth is necessary to cognize it? And, additionally, one might ask, if there is a part of consciousness that can cognize itself, then why can’t the second part, the darśana-bhāga, do that on its own, without requiring an additional svasavitti part? In fact, I think, Dignāga would concede all of this, because all three aspects are ways of talking about a single act of consciousness. So why do these commentators feel the need to add a fourth “part”? As I suggested, for at least two reasons:
(1) For them, in order for consciousness to properly cognize an object, it and its object must be distinguishable. Therefore, consciousness, to see itself, must turn itself, or part of itself, into an object, an object that another part of consciousness can cognize. Holding such a theory resulted in part from the attempt to neutralize the Madhyamakan criticism of self-illumination. This argument will be made explicit in the text itself shortly.
(2) A new element had entered the picture, which was not Dignāga’s own concern, but which quickly became bound up with his system since it was his system that instigated it. That new element is the idea of “validation.” How does a pramāa provide valid knowledge? The three parts as Dignāga explained them were, these commentators believed, inadequate for providing such proof, a witness that the witness itself is telling the truth.
“Verifi cation” here would seem to mean something like ‘getting the object right,’ replicating mentally exactly the features of the ālambana. Darśana may or may not get the nimitta right, and hence it may or may not be pramāa, an “accurate” cognition. But, according to BBh-U, inner perceptions by the two inner bhāgas are invariably accurate, and hence pramāa. It is important to keep in mind that while some notion of “validity” seems to be implied by the term pramāa, no criteria for validity of a perception is offered here, nor, for that matter, in Dignāga. That burden has not yet been placed on pratyak•a-pramāa, but here we are moving in that direction. Dignāga, just like Asaga, merely asserts that erroneous perceptions such as hallucinations, sense-organ defects, mental confusions, etc., are, by defi nition, not pratyak•a-pramāa.55 Aside from getting the object right, no other ‘truth value’ is asserted for pratyak•apramāa by Dignāga.56 Dignāga offers us no theory of error; he merely acknowledges that there are perceptual errors.
To state this hypothesis more clearly, it is conceivable that for Dignāga and for these BBh-U commentators writing just before Dharmakīrti’s revisions of Buddhist pramāa-vāda emerged, the concept of a pramāa had not yet acquired the sense of ‘justifi cation,’ though it may have been developing a weak sense of ‘validity.’ It is weak because it offers no intrinsic criteria by which a perception could be “verifi ed” as a correct cognition within the perception itself. To validate it outside the perception — for instance, intersubjectively, or by proposing a theory of reality to which it theoretically complies — brings anumāna, and hence kalpanā into perception, which, for Dignāga and these commentators following him, would be a fatal category error. Each pramāa must be a unique means to a type of knowledge that can be acquired from no other pramāa. Without providing a clear test to determine the validity of a perception, no justifi cation is possible. Justifi cation is so lacking in Dignāga’s treatment of pratyak•a and in this BBh-U discussion, that one could argue that the four bhāga theory is not a validation theory at all, but merely an attempt to construct a model that satisfi es certain theoretical requirements, viz. acknowledging that all cognitions must have an object (ālambana) in order to be considered cognitions; that for a cognition to see itself, it must objectify itself to itself; that infi nite regresses are fatal; and so on. This hypothesis can quickly be tested: Is there anything in this BBh-U passage that would contradict this thesis? That is, is anything explicitly (or even implicitly) stated which provides any justifi catory tools or criteria? I contend there is not. By differentiating pramāa from apramāa — as BBh-U does — or acknowledging that some perceptual events may be abhrānta (erroneous) — as do Asaga and Dignāga — these pre-Dharmakīrti Buddhists show that they are aware that perception does not automatically and nonproblematically present its contents to a perceiver. This text, like most Buddhist texts, assumes there is something inadequate or false in the usual way we see and understand things, and that advancement on the Path consists of improving and correcting those ways of seeing. But BBh-U does not advance a theory of perceptual error, aside from the standard problems, such as attachment, the presumption of selfhood, etc. By not articulating a theory of error that provides criteria by which a perceptual event can test, verify, and justify itself — or be justifi ed — they show a lack of concern for the problems of justifi cation, as if they were unaware that any such problems exist, much less that such issues need to be addressed.
Returning to the passage, the idea that each citta and caitta is aware of itself is not taken up as primarily a problem of validity, but as an attempt to account for how mental components can know themselves, ‘selfilluminate.’ Since they can’t know themselves without making themselves an object to themselves, they are distinguished into partitions that treat each other as objects, until the BBh-U feels there are enough partitions to have a complex mental act know itself as well as an external nimitta. One watching the other, suggests an infi nite regress.
How does BBh-U avoid the infi nite regress? Weakly. Simply by asserting that the four bhāgas offer the full picture. This, of course, is mere assertion, not an adequate argument.
By this reasoning, although [[[cognition]]] is a single event, it is a composite of many parts that are neither identical nor separate. The inner and outer [components that constitute a cognition], being altogether known, there is no fallacy of an infi nite regress.
And now the proof-text:
The verse is explained as follows:
The idea of this verse is that the nature of the mind of sentient beings is a composite of two parts. Whether [directed] internally or externally, all [[[cognitions]] are] intertwinings of grasped and grasper. [Particular acts of] seeing (darśana) the plethora (of perceptual objects) may be either pramāa or apramāa. [One sees] the multitude of distinct differentiations either (directly via) perception or (indirectly via) inference. 此頌意言。眾生心性二分合成。若內若外皆有所取能取纏繞。見有種種或量非量。或現或比多分差別。
Now the text reminds us that we are discussing not just cognition in general, but uncontaminated cognition, i.e., cognition in terms of the Four jñānas. The mental components of the Four Cognitions, even though they have many parts, are nonetheless all classifi ed as uncontaminated valid perception (anāsrava-pratyak•a-pramāa). This idea has been elaborated elsewhere (in this treatise). The idea is that while [cognition’s] activities (用 yong) are divided into many, [[[cognition]]] has no difference in itself (體 ti). This is just like the one Dharma being differentiated into a plethora of ideas such as suffering, impermanence, etc., while [the Dharma] itself is one.
四智心品雖有多分。然皆無漏現量所攝。此義廣如餘處分別。義用分多非體有異。如一法上苦無常等。種種義別而體是一。 So the commentators, at least as expressed in Chinese,58 follow Dignāga in taking these different aspects as ultimately parts of a single event of consciousness, but rather than expressing this, as Dignāga does, as a single act of consciousness, a ti-yong 體用 distinction is introduced that suddenly gives each "part" its own activity (yong), united now by a single “body” (ti). Having provided a proof-text, the BBh-U next offers “reasoned argument” to buttress this proof:
Next, as to what was said about the mental components associated with the Four Cognitions as having an image part, a seeing part, and so on, there defi nitely is a seeing part that illuminates (prakāśa) and a cognitive-object (vi•aya) that is illuminated. [That is obvious to everyone’s experience.59] There is a self-aware part that illuminates both the seeing part and the being aware of being self-aware part, since the being aware of being self-aware part illuminates the selfaware part [and validates it].60 [The latter two parts] also defi nitely exist, since if they didn’t exist, differentiated in this way into three parts, then there would be no cognitive-support (ālambana) and they wouldn’t be called cognitions (jñāna).
The main argument for why consciousness, in addition to having a cognitive object such as a sense-object, must be divided into three additional cognizers, is given in the last line: “If they didn’t exist, differentiated in this way into three parts, then there would be no cognitive-support (ālambana) and they wouldn’t be called cognitions (jñāna).” That is, in order for a cognition to be called a cognition (jñāna), the cognition must have an ālambana, an object-support. That which illuminates must illuminate something else, something which can refl ect back in order to be cognized. The seeing part only sees what is other than itself. Hence each component of consciousness must be seen by something other than itself. A consciousness cannot be its own object-support, so, in order for all the cognizing aspects of consciousness to be, in turn, ālambanas as well as cognizers, the mathematical permutations require three distinct parts, or so these commentators reason. Darśana-bhāga has as its ālambana the nimitta, but it cannot be its own ālambana, so it requires an additional cognizer, the svasavitti. The svasavitti takes the darśana as its ālambana. But, by this reasoning, the svasavitti cannot act as its own ālambana either, and the darśana only looks outward toward nimittas. So the svasavitti requires a third cognizer, the svasavitti-savitti which takes the svasavitti as its ālambana. But who takes the svasavittisavitti as its ālambana? That task falls back to the svasavitti as well. Since, unlike the darśana, it already is gazing inward at inner mental events, it has the requisite ability for this task. So the svasavitti-savitti cognizes the svasavitti as its ālambana, and the svasavitti does likewise to the svasavitti-savitti. In short, the fi rst partition serves as the ālambana for the second. The second is the ālambana for the third, and the third is ālambana for the fourth. To avoid the need for a fi fth partition (and so on ad infi nitum), the third does double-duty, taking not only the second as its ālambana, but the fourth as well. Hardly an elegant solution!
To restate this: The argument here appears to be arguing that (1) each consciousness or cognition (jñāna) must, by defi nition as just stipulated, have a double role, viz. as cognizer and as cognized. (2) What is cognized is an ālambana. (3) Since each of the three (darśana, svasavitti, and sva-savitti-savitti) must have both roles, they must exist, (4) or else one could not take one’s own cognition as an object (ālambana), which would make self-refl ection impossible, even though it is part of everyone’s experience.
As this relies on accepting the formal defi nition of a consciousness or cognition necessarily having both parts, as well as on the necessity of adding a fourth part (svasavitti-savitti) to the fi rst three — which has not been logically demonstrated here — this argument begs the question. One might concede that if there are four parts, they could reasonably be construed this way, but nothing in the argument compels us to accept the claim that there are indeed four parts. Moreover, even if one concedes a certain plausibility to the relations between the four as described, one might nevertheless question whether either the third or the fourth are necessary at all since it is equally conceivable that each of the three consciousness bhāgas also takes itself as its own object. If, for instance, darśana-bhāga has the capacity to svasavitti itself — which one could assert with the same conviction as one claims that cognition has four parts, since, as was argued earlier, a lamp lights both other things and itself without requiring additional lamps, wicks, etc. — then the third and fourth bhāgas become superfl uous. Similarly, if the third can svasavitti itself — since it is svasavitti by nature and name — then the fourth is entirely redundant. If the third is not svasavitti in that sense, then what does the sva- 自 in sva-savitti 自證 refer to?
We now arrive at the most fascinating part of the passage. In the portion just concluded, the problem was how to turn cognizers into ālambanas. The three perceiving bhāgas have been discussed, but the status of the fi rst bhāga, the nimitta, was skipped over, since it is not a cognizer but, in terms of cognition, only an ālambana. This neglect is now rectifi ed. This portion of the passage is fascinating because, rather than spin out a defi nitive theory on how consciousness constructs its objects, or something of that sort — which is what many would expect from a “Yogācāra” text — it begins by telling us that the status of the nimitta is “inconclusive.”
The [[[Wikipedia:status|status]] of] the image part is inconclusive. [There are three theories.] 相分不定。
In Chinese buding 不定 can render several Sanskrit terms, but if the commentators are continuing to draw on Dignāga's vocabulary, then buding here means aniścita (Asaga and Śakarasvāmin use the term anaikāntika for this). Anaikāntika is a type of argument that results in neither a true nor false determination, but remains “inconclusive.” So the status of the image is inconclusive (a point Dignāga himself makes, e.g., near the end of Ālambana-parīk•ā 61).
We are now given three distinct theories on the status of the image. It is important to remember that all three are Yogācāra theories, though the second will look surprising to those who assume that Yogācāra inviolably embraces idealism. In the process, each theory tries to explain what key words in the section of the Buddhabhūmi sūtra being commented on — such as nirvikalpa, nirākāra, acintya, etc. — refer to, each theory assigning those terms different referents. There is a theory: Since there is no obstruction between real things (tattvas) and the uncontaminated mental components, [the components] directly/immediately illuminate the objects that are before them, without having to pursue them.62 The mind turns itself into a replica of the image of the objects that are before it. The term “imageless” (*nirākāra) refers to the uncontaminated mind, since it doesn’t imagine (nirvikalpa), and “non-conceptual” (acintya) refers to the cognitive object (lambana-vi•aya).63
As with ordinary perception, this theory posits a prakāśa type of uncontaminated perception. Unlike ordinary perception, this uncontaminated perception is “non-obstructed,” directly illuminating whatever is before it such that the mind immediately and accurately forms a replicate image. That image is called “imageless” because it lacks conceptual, imaginative distortions (kalpanā, etc.). The cognizer is devoid of imagination (nirvikalpa) and the cognized object is “nonconceptual” (acintya).
Another theory: [For the cognition of] real things (tattvas), uncontaminated mental components also have an image part. What are called the ālambana for [the uncontaminated] citta and caittas is the appearance of cognitive-objects that discloses them as dharmatā [i.e., just as they are]. This is not like pincers, etc., actively grasping things, nor is it like lamps, etc., whose light radiates to illuminate things.64 [The cognition] is like a bright mirror, etc., in which the refl ection of illuminated things appear.65 The term “non-obstructed” [indicates] that the replicas66 of the cognitive-objects in perception are clearly seen, illuminated and discerned.
The term “imageless” [indicates] that [these mental components] neither attach to nor schematize [their objects]; and the term “nonconceptual” [indicates] the nondiscriminative [[[cognition]]] whose wondrous functioning is diffi cult to calculate.
It is not that it doesn’t perceive images (pratibimba).67 If one says there are no images (ākāra), that would entail there is no image part (nimitta-bhāga). If one says there is no discriminating [of images], then there would be no seeing part [either]. If both the image and seeing parts were entirely nonexistent, that would be like empty space (ākāśa), or [like proposing nonsensical chimeras] like “the horns of a rabbit”; [that sort of nonsense] shouldn’t be called “cognition,” since [uncontaminated jñāna] lacks such [fallacious] attachments and schematizations.
As to the phrase “the image is devoid of grasper and grasped, etc.” it is not that [this cognition] lacks the function which illuminates intentional objects (artha), [[[mentally]]] replicating cognitive objects (vi•aya) from those cognitive conditions (ālambana). If uncontaminated citta was entirely devoid of an image part, then Buddhas wouldn’t perceive bodies and fi elds, etc., nor [would they perceive] the plethora of images (*pratibimba). That would contradict what the sūtras and śāstras say in many places.
有義。真實無漏心品。亦有相分。諸心心法,法爾似境,顯現名緣。非如鉗等動作取物。非如燈等舒光照物。如明鏡等現影照物。由似境現分明照了名無障礙。不執不計說名無相。亦無分別妙用難測名不思議。非不現影。若言無相則無相分。言無分別應無見分。都無相見應如虛空。或兔角等應不名智。無執計故。 言無能取所取等相。非無似境緣照義用。若無漏心全無相分。諸佛不應現身土等種種影像。如是則違處處經論。 (似＝以【宋】【元】【明】【宮】。)
This theory expends great effort to refute any notion that uncontaminated cognition lacks images, i.e., it is making a strong sākāra case against a nirākāra opponent. Since this is one of three positions collected here by Bandhuprabha in the BBh-U, and we fi nd here no passage laying out the nirākāra purvapak•in arguments, we may assume that this compilation is not a complete rendering of the three commentaries being drawn from, but rather an edited selection from those three. In that sense, it may have served as the model for Xuanzang’s own Cheng weishilun, in which he selectively stitched together a number of Triśika commentaries as well as other texts, including the BBh-U.
This second theory makes several interesting points. One sees things just as they are (dharmatā), immediately, without anything interceding or impeding. The term dharmatā puts the focus and emphasis on the cognitive object. One doesn’t have to pursue and “apprehend” an object, nor does one have to shine a light outward to chase down and illuminate the object. One’s mind becomes receptive, so that the objects replicate in one’s mind, as images in a mirror. The objects effortlessly come to one. Cognitive objects are fully and accurately replicated and discerned. There is no attachment nor any distortive mental projections, unlike what is usually the case in ordinary (= contaminated, sāsrava) perception.
The theory then mounts a refutation of a nirākāra opponent whose arguments are not provided. The nirākāra theory would have been that uncontaminated cognitions have no images (nirākāra) at all. None of the three interpretations presented in BBh-U argues for nirākāra as the sheer absence of any images whatsoever, though we know that some in India did understand it that way. The second BBh-U theory sets out to demonstrate that the idea that cognition lacks any image-content is absurd. If there is no object, then there is no perceiver either, in which case there would be no perception at all. Instead of a description of cognition (jñāna), this would be a description of blank, void space (ākāśa), lacking not only images but cognitions. To argue otherwise, this theory tells us, would be to indulge in chimeras and conceptual nonsense, something which, by defi nition, uncontaminated jñānas specifi cally don’t do. To make the perceiver and the perceptual object in any way unreal would undermine Buddhism itself, since “Buddhas wouldn’t perceive bodies and fi elds, etc., nor [would they perceive] the plethora of images. That would contradict what the sūtras and śāstras say in many places.” The Cheng weishilun reiterates this.68 Is this as realist as it sounds? This theory continues:
If one who overturns the basis of the rūpa-skandha doesn’t attain69 rūpa, then overturning the basis of the [other] four skandhas would [result in] being without consciousness, etc. [[[Thinking]] like that] would be to commit a great error. 轉色蘊依不得色者。轉四蘊依應無識等。則成大過。
This passage leaves no doubt that the second theory is not only rationalist (i.e., opposed to chimeric formulations), but unabashedly realist as well. It is probably worth noting as well that one of the things such discussions in the BBh-U demonstrates is that at that time there was no fi xed doctrine of āśraya-parāv•tti (“overturning the basis” from which one cognizes, changing from contaminated to uncontaminated cognition), but rather the term elicited a number of competing notions. 70 Finally, a third theory is given:
And another theory: The mental components associated with uncontaminated, nondiscriminative cognition71 [are to be explained as follows]. Because they are non-imaginative (nirvikalpa), the ālambana is [seen] just as it is (tathatā), since [[[seeing]] things exactly as they are means there are] no separate [[[imaginary]] images intermediating between the cognition and] the thing itself, just as when illumination of the self-nature (svabhāva) [of a lamp arises] it is not separate from the image part.
If [an uncontaminated cognition] has discriminations (savikalpa), the mental components are associating with post-attainment cognition (p••halabdha-jñāna72) [and not nirvikalpa cognition]. Since ālambanas and cognitive-objects (*vi•aya-gocara) sometimes are separate from the things themselves, [in such cases this is] just like when a contaminated mind perceives a replica of the image of a cognitive object by clearly seeing (*vispa•a) and illuminating the ālambana.
若後得智相應心品有分別故。所緣境界。或離體故。如有漏心 似境相現 分明緣照。
This third theory also explains that because this cognition is devoid of imaginative construction, its ālambana is perceived just as it is, this time using the term tathatā rather than dharmatā to make that point. While dharmatā put the emphasis on the cognitive object, tathatā puts the emphasis on the proper functioning of the entire cognitive operation. This theory also stresses the absence of any mediating factors. This immediate form of cognition is compared to the lamp, as described above when refuting the objection that consciousness, like lamps, doesn’t illuminate other things as well as self-illuminate. There, after pointing out that the light from lamps illuminates other things by eliminating the obstructive darkness, making those things visible and available for perception, the lamp’s self-illumination was explained thus: “When [a lamp’s] self nature [to illuminate] arises, the encompassing obstruction of darkness is cleared away, making it visible to perception; therefore this is called ‘self-illumination’.” Uncontaminated cognition operates without the aid of intermediate imaging or imaginings.
The issue this theory will try to address is the degree to which an image of something is identical to or separate from that thing itself. In ordinary perception, it seems to suggest, the ālambanas and cognitive objects that appear in cognition are not exact replicas of something real. However, even in ordinary perception, on occasion one might perceive something as it is in itself; uncontaminated cognition is like that all the time. This third theory now introduces a distinction not mentioned by the previous theories, but which recurs elsewhere in the BBh-U, namely differentiating nirvikalpa-jñāna — generally understood in these theories as a cognition devoid of imagination [could 無分別智 here stand for “absence of kalpanā” rather than the usual nirvikalpa-jñāna?] — on the one hand, from post-attainment cognition (p••halabdha-jñāna 後得
智) on the other. "If [an uncontaminated cognition] has discriminations [[[savikalpa]], parikalpa, kalpanā?],” this theory informs us, then “the mental components are related to post-attainment cognition, [and not nirvikalpa].” Why? “Since ālambanas and cognitive-objects (*vi•ayagocara) sometimes are separate from the things themselves, [in such cases this is] just like when a contaminated mind perceives a replica of the image of a cognitive object by clearly seeing (分明, *vispa•a) and illuminating the ālambana.” If an image, which is not the thing itself, appears in perception, then this is either ordinary contaminated perception or uncontaminated post-attainment cognition. The latter, in other words, operates much as ordinary perception. Why the sudden distinction between nirvikalpa-jñāna and post-attainment cognition? This is an attempt to answer the argument in the second theory that insists there must be distinct, determinate objects and images in order to account for Buddhas and other enlightened beings who, while perceiving uncontaminatedly, see and interact with distinct things. According the third theory, when the cognition of a Buddha is nirvikalpa, it is devoid of “images” or imaginings. Buddhas, however, can interact with things and beings while engaging in post-attainment cognitions. Hence, when sūtras speak of Buddhas doing such things, that is what they mean.
However, this third theory admits, by implication, that images of things and the things themselves are sometimes “separate.” Now an objection based on this is raised.
[Objection:] If an uncontaminated mind takes as its ālambana [something that] is separate from the object itself, [it could happen that] the image [one perceives] bears no resemblance to that [[[object]]], and yet one apprehends an ālambana.73 [That would be a problematic cognition, since an uncontaminated jñāna should provide true knowledge]. 若無漏心緣離體境。無似彼相74而得緣者。
This is an interesting objection. Doesn't the image (ākāra) that appears in perception need to be a replication of some thing that it appears to represent? Wouldn’t this even be the case more so for an uncontaminated mind whose perceptions should be true, accurate, and apprehend reality as it is? An ālambana must convey its own image to the cognition in order to be an ālambana. If a resultant cognition is not an exact replication of something, how is it that one nevertheless apprehends ālambanas? Wouldn’t that imply that the ālambana is not really an ālambana, and thus one is engaged in a false, illusory perception? If what appears in cognition is solely a product of one’s own mind, exclusively generated by one’s own mental seeds and corresponding to nothing else, then what distinguishes a delusion from a correct or valid cognition? Delusions are also produced by one’s own mind — in fact, that is precisely what makes them false and delusory.
It cannot be the case that only the two inner-looking mental components — svasavitti and svasavitti-savitti — are valid, since we were explicitly told that darśana — the component that exclusively looks outward — is sometimes pramāa, “accurate,” “valid.” Svasavitti does not validate the content of what darśana sees, only that it is seeing. This is analogous to acknowledging that someone believes X to be true, regardless of whether X is actually true or not. Svasavitti sees darśana observing certain nimitta, but, since svasavitti is exclusively confi ned to inner introspection, it cannot go “out” to check on the status of the nimitta. BBh-U has already fully conceded this by admitting that the status of the nimitta is inconclusive. As to the objection raised by the opponent, viz. that what appears to be an ālambana may bear no resemblance to the underlying object, this theory responds as follows:
[Reply:] [According to Dignāga’s] Ālambana-parīk•ā one shouldn’t say that because the image of atoms does not appear in the fi ve consciousnesses that therefore there is no ālambana (at all). In this way, the image of the cognitive-object is identical to the uncontaminated mind. Uncontaminated seeds arise. Even though they resemble contaminated dharmas, nonetheless they are not contaminated, just as a contaminated mind may [have cognitions that] resemble an uncontaminated image, though they are not uncontaminated.
觀所緣論 不應說言 五識上無似極微相 故非所緣。如是境相同無漏心。無漏種起。雖有相似有漏法者。然非有漏。如有漏心似無漏相非無漏故。
This ends the elaboration. 且止廣論。
Referring to another of Dignāga's works, the Ālambana-parīk•ā, this theory correctly concludes that while Dignāga refutes the possibility that either atoms or collections of atoms can satisfy the criteria for an ālambana, nevertheless for Dignāga cognition does involve ālambanas.
As I understand this third theory, it is not claiming that there is no object, and only mental production — which would make this type of cognition parikalpita (false imagining), and not parini•panna (consummate comprehension) — but rather that all cognitive distance, all "obstructions," etc., have been eliminated so that objects appear directly and immediately just as they are. For Dignāga, one of the tasks an ālambana must perform in order to be an ālambana is that it must convey its own image (svarūpa) to the cognition. According to the third theory, the ālambana needn’t convey an image from the object to the mind, since the mind automatically and instantaneously gives rise to an impression of the object that is exact and accurate in every detail. No middle man or mediating process between mind and object is required. One sees things just as they are because the mind has ceased to impose imaginary constructions. One’s own mental seeds — since now no longer contaminated by distorting hindrances
心). Uncontaminated cognitions can "mirror," i.e., appear the same as, contaminated cognitions, and vice versa. What distinguishes them is the type of seed from which they arise. Uncontaminated seeds will give rise exclusively to uncontaminated cognitions, whether the resultant cognition resembles a contaminated cognition or not. What is crucial is how one sees, not what one sees.
The distinction this theory draws between uncontaminated savikalpa (有分別) and uncontaminated nirvikalpa (無分別) cognitions, such that the former would be p••halabdha cognitions, and not strictly nirvikalpa, deserves some additional comment. First, we need to keep in mind that the Chinese, wufenbie 無分別, is used for both nirvikalpa and kalpanāpoha, as mentioned above. Dignāga defi nes perception as kalpanāpoha, ‘devoid of conceptualization.’ If, as would appear to be the case, the BBh-U’s discussion is being guided by Dignāga’s ideas, then the savikalpa type of cognition would be one in which ‘conceptualization’ remains involved. A wufenbie cognition would be a Dignāgan-type ‘perception.’ An analogy would be: One looks at a colorful painting. The actual seeing of the colors, shapes, etc., is a wufenbie cognition. This is simply being absorbed in visual perception, without judgment, analysis, or any sort of conceptualization. As one views the painting, one begins to identify the colors — too much red, nice use of blue, etc. — and discerns and distinguishes shapes — wellbalanced composition, interesting use of brush and palette knife, etc. That would be, if stemming from uncontaminated rather than contaminated seeds, p•halabdha cognition. That is why the third theory states that there is no perceptual difference between p••halabdha cognitions and ordinary (contaminated) cognitions. Such p••halabdha cognitions would not only be on the order of art connoisseurship, but would see the full range of causes and conditions entailed by the painting — how it relates to an audience, human culture, the art market, the construction of meaning, and so on. Thus, elsewhere, BBh-U insists on attributing omniscience to Buddha’s cognitions. Perhaps that alone would differentiate a p••halabdha cognition (i.e., a post-awakening cognition) from ordinary contaminated cognition.
To these three theories, the BBh-U now adds a crucial qualifi cation.
Such distinctions (vikalpa) are only from the conventional point of view, as explained logically. They are not from the [[[Wikipedia:perspective|perspective]]] of ultimate meaning; the ultimate meaning is apart from words and deliberation. From the perspective of the imageless (nirākāra-d••i 無相見), one already is incapable of speaking of citta, caittas, and so on.75 Those are beyond fi ctional-proliferation (prapañca 戲論) and incapable of being conceptualized (acintya 不可思議). 如是分別但就世俗言說道理。非就勝義。若就勝義離言絕慮。既無相見。不可言心及心法等。離諸戲論不可思議。
All talk of citta and caittas, eight consciousnesses, various partitions of consciousness, etc., are only “conventionalisms.” Ultimately (paramārtha), there are no separate images, partitions of consciousness, and so on. To speak of such things is to still dabble to some extent in prapañca and conceptualisms (cintya). “Beyond prapañca” (the Chinese gives prapañca in the plural) refers to the cognitive capacity; acintya (nonconceptualizable) refers to the object, a nonconceptual vastu. Ultimately uncontaminated cognition is devoid of prapañca and its cognitive-object is nonconceptual (acintya). This is tathatā, cognizing things as they are.
We now turn to a related passage from the Cheng weishilun, to note quickly some differences.
From Cheng weishilun76
Some of this will look similar to what we just saw in the BBh-U, but there are two notable differences. First, a clearer distinction is made concerning ālambana, specifi cally, a distinction between ālambana 所緣 and ālambaka 能緣, i.e., the cognitive-support and the taking of a cognitivesupport. This may refl ect a distinction in the original Sanskrit passage from which Xuanzang drew, or it may have been his own "clarifi cation" (i.e., the Sanskrit may have only said ālambana while Xuanzang differentiates these two senses in his translation). The second, and more important difference from the BBh-U is that after laying out the four bhāga theory, Cheng weishilun then reduces them all to the second bhāga, darśanabhāga, which would put it in line with what Kuiji identifi es as Sthiramati’s theory, namely that all are reducible to the second bhāga.77
Sometimes the seeing part is not classifi ed as a pramāa [i.e., it sometimes has erroneous cognitions]. Due to this, the seeing part doesn’t ‘verify’ (or isn’t aware of) the third, since to verify itself it would necessarily have to perceive [itself]. Of these four components, the fi rst two are external, and the latter two are internal. The fi rst is only an ālambana; the other three are both
That is, the second part has only the fi rst for its ālambana. Sometimes it is a valid cognition (pramāa), and sometimes an invalid cognition (apramāa). Sometimes [it cognizes its ālambana] by perception (pratyak•a), and sometimes by inference
The third takes the second and fourth as its ālambana [i.e., the darśana and svasavitti-savitti are the ālambana for the svasavitti]. The svasavitti-savitti only has the third as its ālambana, but not the second, since it lacks that function.
The third and fourth are both classifi ed as 'valid perception' (pratyak•a-pramāa). 第三第四皆現量攝。
Thus, that citta and caittas consist of these four parts is established. [Since this is the] full [account of the relation between] ālambaka and ālambana, there is no fallacy of infi nite regress. Neither the same nor different, they are established by reason to be consciousness only (vijñapti-mātra).
This covers the same ground as BBh-U. Note that the label vijñapti-mātra has been brought into the discussion. One reason for this is that this description of the four partitions is not concerned with uncontaminated cognition or post-āśraya-parāv•tti cognitions. It concerns, as the next passage shows, ālambana and ālambaka as entanglements of grasper and grasped. According to the Cheng weishilun and Kuiji, an ālambanapratyaya necessarily always is cognitive (所慮).80 Cheng weishilun now offers the same proof text from the Ghanavyūha Sūtra, with the same explanation.
What this verse intends to say is that the nature of the mind of sentient beings is a composite of two parts. Whether [directed] internally or externally, all [[[cognitions]] are] intertwinings of grasped and grasper. Seeing (darśana, d•śya) has many types. Sometimes [[[seeing]]] is valid knowledge (pramāa), and sometimes invalid knowledge (apramāa). Sometimes [it cognizes its ālambana] by perception (pratyak•a) and sometimes by inference (anumāna). It differentiates into many parts. Among these, “seeing” is the darśana-bhāga (seeing part).
Now something new is introduced, with an additional proof text.
[Reducing the number of bhāgas]
In this way, the four parts may be grouped as three, since the fourth category gets included in the svasavitti part. Or they may be grouped as two, since the nature of the last three is to be a cognizer (ālambaka). So all (three) are classifi ed as darśana-bhāga. The meaning (artha) of the word “seeing” (darśana) is “cognizer” (ālambaka).
Or they may be grouped as one, since there is no separation between them.
That which is seen (d•śya) does not exist.
由自心執著 心似外境轉彼所見非有 是故說唯心
In this way, in every place and situation, we say there is only a single mental event (ekacitta-mātra). This term “single mental event” also includes the caittas. Hence, the defi ning activity (ākāra) of consciousness (vijñāna) precisely is discerning (vijñapti). Discerning is precisely the seeing part (darśana-bhāga) of consciousness.
The Lakāvatāra passage states that attachment to one’s own mind compels the sort of self-interest that sets out (sapravartate) to satisfy itself by seeking, looking outward (bahirdhā) for things to gratify and assure itself. Mind projects based on its self-attachment. These mental projections do not exist (nāsti) as they seem; what is seen (d•śya) is only mental projection (citta-mātra). Cheng weishilun, drawing on d•śya (the seen) as cognate to darśana (seeing), equates consciousness (vijñāna) with seeing (darśana), since the defi ning activity (ākāra) of consciousness is to discern (vijñapti), i.e., to see. However and wherever consciousness is operating — motivated by these inner compulsions — it basically does one thing: it sees, it cognizes. Wherever cognition occurs, there is a singular, compulsive, mental act — projection based on a desire to seek, to see. The Cheng weishilun therefore comes back into line with Dignāga, stressing that the differentiation of consciousness into four partitions should not distract from the fact that wherever consciousness is acting, it is only doing one thing: seeing. Consciousness, in this sense, is singular in what it is and what it does (svabhāva and ākāra). Consciousness is defi ned by what it does. What it does is discern (vijñapti), make known. And discerning is the defi ning activity of the darśana-bhāga. As mentioned previously, this brings Cheng weishilun back in line with Sthiramati as well.
What of the fact that darśana-bhāga is sometimes apramāa? Since the context of this passage is vijñapti qua attachment and contaminated cognition, and not the uncontaminated cognitions being discussed in BBh-U, darśana-bhāga’s capacity to be either pramāa or apramāa is not at issue. Cheng weishilun here is not concerned with “validating” cognition, but with how unenlightened cognition works.
As for BBh-U, it is worth noting that all three theories assumed some correspondence between what appears in consciousness and the object being cognized in this way. All three theories call this mental image a “replica” (sād•śya 似). The BBh-U gives three theories on what postāśraya-parāv•tti cognition involves. The fi rst theory depicts it as producing a replica (sād•śya) of the actual thing, similar to how non-enlightened cognition works, except it is devoid of attachment and grasping. The second theory is similar, except it explicitly insists that rūpa itself is cognitively obtained, directly contacted by the mind without any form of apprehending or the need to illuminate objects. The third is similar to the fi rst, except it emphasizes that the replica (sād•śya) is produced from anāsrava-seeds.
If what appears in cognition is a replica, of what is it a replication? One possibility is that the replica marks a correspondence between the cognitive-object and the actual thing. As mentioned, Dignāga insists that an ālambana must convey its svarūpa. Of the three theories, only the third seems to preclude that one perceives something other than what one is predisposed, by accumulated seeds, to see. It only allows that uncontaminated jñāna is doing its job more clearly, with uncontaminated predispositions. The fi rst and second theories shorten the distance between the cognizer and the cognized, while the third theory claims that this distance is never really breached. Even this cognition produced from uncontaminated seeds, however, is called a replica in the third theory. Another way of understanding the third theory is that wufenbie cognition is simply self-illumination, as a lamp self-illuminates, while p••halabdha cognition is when the lamp illuminates other things, “just like when a contaminated mind perceives a replica of the image of a cognitive object by clearly seeing (*vispa•a) and illuminating the ālambana.”
On the other hand, if there is no presumed correspondence, how would objects produced through one's own cognitive construction be any different from false imaginings, parikalpa? Is this supposed to be some sort of non-false imaging? Since ordinarily one has such false cognitions arising from one’s mind, and, as some interpretations of Yogācāra tell us, mind alone is “real,” why aren’t these mental products real? Are the cognitions arising from uncontaminated seeds only real when one doesn’t mistake the products of one’s own mind for external things? Why wouldn’t that be a case of narcissistic parikalpita, a solipsism into which Buddhas, teachers, Dharma, etc., could not arise from anywhere but one’s own imagining. Is this only parikalpa if one presupposes some sort of selfhood or non-emptiness to one’s imagining? Does this all boil down to forms of dogmatic labeling of one’s imagination? If so, that too would still be parikalpa. Xuanzang’s rendering of parikalpita as 偏計所執 indicates that he understands it as entailing the calculative (計), ubiquitous (偏) grasping (所執) of things. Non-parikalpa is not simply coming into correspondence with an actual thing (the point of Yogācāra is not simply ontological attainment), but recognizing the processes and structures at work, the attitudes they engender, and the correction of cognitive errors. Attachment is a cardinal error, but not just attachment. Mistaking one’s projections for things themselves obstructs our ability to see actual things (vastu) as they are — rather our cognitive-object should correspond exactly with what is there.
Another Buddhabhūmy-upadeśa passage
While there isn’t time in this paper to explore fully the many passages in BBh-U and other Yogācāra texts that bear on this, nor to trace out how consistently and coherently each of the contributing commentaries in the previous BBh-U passage present their respective positions in the rest of BBh-U, one later passage from the BBh-U may be instructive. While there was no Tibetan counterpart to the previous passage, the Tibetan Buddhabhūmi commentary attributed to Śīlabhadra parallels most of the Chinese version of the following passage (though the Chinese adds a discussion on the meaning of vijñapti-mātra and applies trisvabhāva to its analysis, while these are missing in the Tibetan). I will fi rst present the Chinese version, which is more detailed, followed by the Tibetan version, which differs on a number of points from the Chinese.
Here is the passage, followed by a translation. 82
佛地經論》卷4：「若無實影圓鏡中生云何為喻。有質有鏡和合為緣。如是相現故得為喻。謂諸有情顛倒執著影像。熏習成熟力故。鏡面為緣 自識變異 似面影現。由是世間起增上慢。謂我鏡中見其面影。以無別影鏡中生故。經但說言眾像影現。不言生起。如是應知。一切境相 皆是自識 變異顯現 非別實有。以識勝故。但言唯識。非無心法。亦不說言唯有一識。以諸有情各有八識及心法故。一切色等雖各有種。皆是 自識變異 熏習識上。功能差別為性故。變現時還不離識。就世俗說別有心法。非真實義。以就勝義 諸法皆無定別性故。乃至真如雖非識變。亦不離識。識實性故。識上二空無我共相所顯示故。此唯識言。但遮愚夫橫計一切心心法外定性色等遍計所執。不遣不離諸心心法色等諸法。依他起性圓成實性。非無有故。由平等故。此二平等。是故說言平等平等。世間圓鏡如來智鏡[or鏡智]俱無分別。皆能現影無有差別。由是因緣名圓鏡智。」
If the refl ection (pratibimba 影) in the mirror is not a real [thing], why [would we offer such an] analogy? This is the way the image (ākāra 相) appears [or is perceived; 現]; the conditions are: When [i] there is a bimba 質 and [ii] there is a mirror [and, iii. these two] intermix. Hence the analogy obtains.
That is, sentient beings, by upside-down thinking (*viparyasa 顛倒), attach to refl ected-images (pratibimba 影像) [instead of seeing the bimba directly or understanding its conditions,] because of the maturation of habitual forces.
A face in the mirror (*adarśa-mukha 鏡面) as condition [is analogous to] the differing alterations (vipariāma 變異) of one's own consciousness [in which] a replicate-appearance (*sād•śya, *pratibhāsa 似) of a face (mukha面), i.e., the refl ected-image (pratibimba), appears [or is perceived]. From this, worldlings give rise to great arrogance (adhimāna 增上慢83), [falsely believing that this refl ection is the thing itself, that is, that the refl ection is their own face]. That is, [they think] “It is me in the mirror” when they see a refl ection of their face. This is because they do not distinguish [themselves] from the refl ection that arises in the mirror.
The sūtra only says that the various refl ected images appear [in the mirror]; it doesn’t say that they are produced [by the mirror].84 From this you should know that all the images of cognitive- objects (vi•aya-nimitta 境相) are displays85 of the differing alterations of one’s own consciousness, not separate dravyas, because consciousness is the dominant (condition) [i.e., the adhipatipratyaya,86 以識勝故].87
But the term "consciousness only" (vijñapti-mātra) does not [mean] there are no caittas, nor does [the sūtra] say that only a single consciousness exists, since sentient beings each have eight consciousnesses and their caittas. Although all rūpas, etc., each has [its own distinct] types,88 all are differing alterations of one’s own consciousness [infl uenced by] the habits in consciousness, since the capacity to differentiate is considered its [i.e., consciousness’] nature. At the moment when the alterations appear [due to the infl uence of] returning [[[vāsanās]]], [the images] are not apart from consciousness.89 From the standpoint of sav•ti, one says that the caittas exist separately [from consciousness], but that is not a truthful expression (真實義 *tattvārtha; *yathārtha90), since, from the standpoint of paramārtha, dharmas all lack a defi nitive separate nature. This extends up to suchness which, although it is not an alteration of consciousness, it is not apart from consciousness either, since it is the actuality of consciousness (識實性故), and since it is revealed in consciousness as the general feature (sāmānya-lak•aa) of selfl essness of the two emptinesses [of persons and dharmas]. The term "consciousness only" is only used to refute (但遮) the mistaken notion of ignorant people that the nature of rūpa, etc., is defi nitely external to all the cittas and caittas, which is false imagining (parikalpita). It does not reject (不遣) dharmas such as rūpa, etc., that are not apart from citta and caittas; those are paratantra and parini•panna, and hence not inexistent, since they are equal (sama 平等). These two being equal, they are therefore said to be equal in every way (sama-sama 平等平等). The worldling's round mirror and the Tathāgata's mirror cognition are both nonconceptual, nondiscriminating (nirvikalpa 無分別). [In them,] refl ections can all appear [or be perceived] without differentiation. Due to this cause-and-condition [the analogy calls it] "the round mirror cognition."
This passage seems to have attracted wide attention in Chinese Buddhist circles, since we fi nd parts of it quoted in other sources.91 And, once again, it reassures us that Buddha's cognition and our own are not so different after all. We worldlings can learn from our mirrors. When we recognize that the rūpa contributing to our cognition, at the moment of pariāma, is not separate from consciousness, and that obviously remembered, imagined and anticipated rūpa, as well as the concept of rūpa, and any theory of rūpa, are never apart from consciousness, then that rūpa is accepted as real. Rūpa and mind become equalized; a double equality (sama-sama) in which, because mind becomes rūpa and rūpa becomes mind, they are the same; nirvikalpa and acintya. As the Heart Sūtra says: “Form itself is emptiness, emptiness itself is form.”
Unlike the previous passages which lacked any parallels in the Tibetan, this passage does have a parallel, though with some differences. Without attempting to confl ate the Chinese and Tibetan versions into a diplomatic rendition (since they seem to differ in several details), I offer the following tentative translation of the parallel Tibetan passage, with some comparative remarks in the notes:92
But if [something’s] own-refl ection (rang gi gzugs brnyan; *svabimba; *sva-pratibimba) is not another entity (dravya) that is other than the thing it is refl ecting, then how can it be a suitable example?
It is because as follows: The image (nimitta) of something in a mirror will appear as that very thing (bimba). For that reason, moreover, in the same way, it is because of the power of the coming to maturity93 of vāsanās of attachment (abhiniveśa) that appearances (*ābhāsate) abide in consciousness. This is why ordinary people (loka) develop pride, thinking “I see [my own] refl ection (pratibimba);” they don’t say “This is [another being] being produced.” Similarly, the object (jñeya94) is otherwise than the mirror, but the jñeya appears to be inside [it], and is seen that way.
Same (‘draba)-sameness (mnyampa) is sameness (mnyampa).95 [An ordinary] mirror and the circle (*maala) of the Tathāgata’s mirror-like knowledge are only images (nimitta-mātra) from which refl ections (pratibimba) arise, and are equally nonconceptual (nirvikalpa). “Therefore” refers to these three reasons.96
Therefore, the conventional designation (vyavahara) “mirror-like knowledge” obtains.97 This is how one should understand the term “mirror-like knowledge.” Are the additional portions found only in the Chinese — and not in the Tibetan — editorial supplements provided by Xuanzang himself, or commentarial interpolations from one of BBh-U’s other commentar ies, or merely speckled refl ections in the mirror of the main text? Whose face is shining back at us?
Select Bibliography and Abbreviations
MW Monier-Williams, Monier. (1986) A Sanskrit-English Dictionary. Delhi: Motilal Banarsidass, rpt. of 1899 Oxford University Press ed. T Taishō shinshū daizōkyō. Takakusu Junjirō, Watanabe Kaikyoku, et al. (eds.), 100 vols. Tokyo: Taishō Issaikyō Kankōkai, 1924-1932. XZJ Xu Zhang Jing 續藏經. [Supplement to the Canon]. Taipei: Xinwenfeng Chubanshe, 1975.
Bhattacharya, Kamaleswar, E.H. Johnston and Arnold Kunst. (1986) The Dialectical Method of Nāgārjuna: Vigrahavyāvartanī. Delhi: Motilal Banarsidass. 2nd ed. Cook, Francis H. (translator). (1999) Three Texts on Consciousness Only. Berkeley: Numata Center. Garfi eld, Jay L. (1995) The Fundamental Wisdom of the Middle Way: Nāgārjuna’s Mūlamadhyamakakārikā. Oxford and NY: Oxford University Press. Hattori, Masaaki. (1968) Dignāga, On Perception. Cambridge, MA: Harvard University Press.
Keenan, John (translator). (2002) The Interpretation of the Buddha Land. Berkekey: Numata Center. Keenan, John. (1980) A Study of the Buddhabhūmy-upadeśa: The Doctrinal Development of the Notion of Wisdom in Yogācāra Thought. Unpublished Ph.D. Dissertation, University of Wisconsin.
Lakāvatāra sūtra. (1963) P.L. Vaidya, ed., Darbhanga: The Mithila Institute. Lusthaus, Dan. (2002) Buddhist Phenomenology: A Philosophical Investigation of Yogācāra Buddhism and the Ch’eng wei-shih lun. London and New York: RoutledgeCurzon.
Randle, H.N. (1981) Fragments from Dinaga. Delhi: Motilal Banarsidass, rpt. of 1926 Royal Asiatic Society ed. Steinkellner, Ernst, Helmut Krasser, and Horst Lasic (eds.). (2005) Jinendrabuddhi´s Pramāasamuccayaika, Chapter 1, Part 1: Critical Edition. Part 2: Diplomatic Edition. Beijing: Verlag der Österreichischen Akademie der Wissenschaft. TACHIKAWA, Musashi. (1971) “A Sixth-Century Manual of Indian Logic (A Translation of the Nyāyapraveśa),” Journal of Indian Philosophy 1, 111-145. Tat, Wei (translator). (1973) Ch’eng Wei-shih lun: Doctrine of MereConsciousness. Hong Kong: The Ch’eng Wei-shi lun Publication Committee. Tola, Fernando and Carmen Dragonetti. (1982) “Dignāga’s Ālambanaparīk•āv•tti." Journal of Indian Philosophy, 10, 2, 105-134. Tucci, Guiseppe. (1929) “Buddhist Logic Before Dinaga (Asaga, Vasubandhu, Tarka-śāstras).” Journal of the Royal Asiatic Society, 451-488, 870f. Williams, Paul. (1998) The Refl exive Nature of Awareness: A Tibetan Madhyamaka Defence. Surrey, England: Curzon. Notes
1. Nanhai jigui neifa chuan 南海寄歸內法傳 (Buddhist Monastic Traditions of South and Southeast Asian): 法稱則重顯因明, "Dharmakīrti then revitalized hetu-vidyā” (T.54.2125.229b20). The description 重顯 zong xian (“revitalize”) suggests that hetu-vidyā studies among Buddhists had been in a less than robust condition prior to Dharmakīrti.
2. As I will try to show on another occasion, many of Dignāga's supposed innovations are already to be found in the hetu-vidyā discussions of Asaga (in the Śrutamayī-bhūmi of the Yogācārabhūmi, the Xianyang shengjiao lun [[[顯揚聖教論]], T.31.1602], and concluding section of Abhidharmasamuccaya). Tucci already began to note this (Tucci, 1929), but there has been little follow-up since. The restriction of possible pramāas to only two — pratyak•a and anumāna — was already a Vaiśe•ika tenet, and already partially accepted by Asaga. Dignāga, I believe, was deeply infl uenced by Vaiśe•ika, though this is hard to gauge with precision, since aside from the Vaiśe•ika Sūtra no pre-Dignāga Vaiśe•ika works survive (I am in agreement with those who hold that Praśastapāda was responding to Dignāga, not the other way around; see n. 12, below). It is another sign of Dignāga’s radical impact that almost all pre-Dignāgan hetu-vidyā texts have not survived, and that those which do — such as Caraka-sahitā, Nyāya-sūtra, Vaiśe•ika sūtra, Jaimini’s Mīmāsa-sūtra, and Śabara-bhā•ya — either offer very general statements that could be reconstrued by subsequent generations, or show signs of stratifi ed accretions that likely post-date Dignāga, in effect, responding to the challenges Dignāga’s thought initiated. Vātsyāyana, the earliest extant commentator on Nyāya Sūtra, was believed to have been pre-Dignāga, but recently that too has been called into question. Cf., e.g., Eli Franco and Kārin Preisendanz, “Bhavadāsa’s Interpretation of Mīmāsa-sūtra 1.1.4 and the Date of the Nyāyabhā•ya,” Berliner Indologische Studien (BIS) 8.1995, 81-86. I believe the reason the pre-Dignāgan materials have disappeared is that they came to be considered atavistic and obsolete, even embarrassing, by their adherents once the import of Dignāga’s logic sunk in. The notable surviving exceptions are Bhart•hari and Śabara, the former due to his own profound infl uence on subsequent thinkers (including Dignāga), and the latter because his commentary on Jaimini remained the root text for the surviving Mīmāsa schools of Kumārila-bhaa and Prabhākara, who reinterpreted its statements on pramāa to accord with contemporary thinking.
3. Some fragments have been discovered since Randle (1981 rpt. of 1926), but that was largely the extent of known Sanskrit material for Dignāga that we possessed until the unveiling of the Jinendrabuddhi commentary on Pramāasamuccaya (Steinkellner, et al., 2005), the fi rst volume of which has already appeared, with more to follow. The commentary contains major portions of Dignāga’s root text, perhaps as much as 70 percent. This promises to change dramatically how we understand Dignāga.
4. A major exception is Śakarasvāmin’s Nyāyapraveśa. For Romanized Sanskrit text and English translation, see Tachikawa (1971). 5. Ernst Steinkellner, in a paper based on a lecture given in Beijing in 2002 entitled "The Buddhist Tradition of Epistemology and Logic (tshad ma) and Its Signifi cance for Tibetan Civilization” (in Andre Gingrich and Guntram Hazod (ed.), Der Rand und die Mitte. Wien: VOeAW, 2006, pp. 193-210) sums up the state of the fi eld thus (p. 199):
The most important philosopher after Dignāga / Phyogs-glang is Dharmakīrti/ Chos-grags (ca. 600–660 CE). It is Dharmakīrti's interpretation of Dignāga's theories that has become authoritative for the whole subsequent period of the school. This is already refl ected by the respective transmission of their texts. All of Dharmakīrti's texts have come down to us, either in their original Sanskrit, or in excellent Tibetan translations, while Dignāga's 'root-text' seems to be lost in the original and is available only in two quite insuffi cient Tibetan translations, one from the late eleventh or early twelfth century (Vasudhararak•ita with Seng-ge rgyal-mtshan) and the other from the late fi fteenth century (Kanakavarman with Mar-thung Dad-pa'i shes-rab).
6. The tabulation appears in his Biography by Huili 慧立 and Yancong 彥悰, Datang Daciensi sanzang fashi zhuan 大唐大慈恩寺三藏法師傳 (T.50.2053.252c). Xuanzang's translations: Nyāyapraveśa = T.32.1630; Nyāyamukha = T.32.1628.
7. There has been much speculation among scholars concerning possible "logic" or "debate" texts that may have been written by Vasubandhu, though little remains of them. The Rushilun 如實論 (T.32.1633), translated by Paramārtha in 550, and better-known by an implausible reconstructed Sanskrit title, Tarka śāstra [rushi usually translates tattva, bhūta, yathā-bhūta, etc., not tarka], is sometimes attributed to Vasubandhu; the Chinese only preserves three sections of what must have originally been a larger text; no vestige of this text survives in any other language. Tucci offered his translation of Rushilun into Sanskrit in G. Tucci, Pre-Dināga Buddhist Texts on Logic from Chinese Sources, (Baroda, 1929). Brendan Gillon writes: “Vasubandhu…wrote at least three works on debate: the Vāda-h•daya, the Vāda-vidhāna and the Vāda-vidhi. No Sanskrit original survives of any of these, though Sanskrit fragments of the last have been collected by Frauwallner [[[Wikipedia:Frauwallner|Frauwallner]], Erich. 1957. “Vasubandhu’s Vādavidhi.” Wiener Zeitschrift für die Kunde Süd — und Ostasiens 1: 104–146].” (B. Gillon, “An Early Buddhist Text on Logic: Fang Bian Xin Lun,” Argumentation, Volume 22, Number 1 / March, 2008, 15-25). A portion of the Vāda-vidhi is preserved in Tibetan; translated in English by S. Anacker in Seven Works of Vasubandhu, Delhi: Motilal Banarsidass, 1984. Frauwallner identifi ed fragments of the Vāda-vidhāna in Uddyotakara, etc.; cf. Frauwallner, Erich (1982). Kleine Schriften. Hrsg. Gerhard Oberhammer und Ernst Steinkellner. Wiesbaden 1982.
10. Xuanzang translated one Vaiśe•ika text, a treatise by Candramati (or *Maticandra) 慧月, entitled Shengzong shi juyi lun 勝宗十句義論 (Treatise on the Ten Padārthas of Vaiśe•ika), T.54.2138. A Sanskrit version of this is unknown in India, as is a Vaiśe•ika system enumerating ten padārthas instead of the more usual six (Vaiśe•ika Sūtra) or nine (Praśastapāda and all subsequent Vaiśe•ikas). That this was the only non-Buddhist text Xuanzang translated suggests the importance he attached to Vaiśe•ika, perhaps precisely because of its relevance for Dignāga.
11. For instance, ideas in the Vaiśe•ika Sūtra (and possibly unknown pre-Praśastapāda Vaiśe•ikas) to which Dignāga is indebted include: Contending that only two pramāas — perception and inference — are legitimate; proofs that 'sound is impermanent' (sometimes considered a signature Dignāga position) (VS 2.2.24-42); Katsura has argued that Dignāga's apoha theory presupposes Vaiśe•ika ontology (Shoryu Katsura, “The apoha theory of Dignaga”, Indogaku Bukkyōgaku Kenkyū, 28.1, 1979, 16-20); etc..
12. Among the more obvious changes he introduced to the system: Vaiśe•ika sūtra proposes six padārthas (elemental factors of reality), while Praśastapāda adds an additional three, raising the total to nine; Vaiśe•ika sūtra lists seventeen qualities, while Praśastapāda expanded the list to twenty-four; perhaps most importantly, prior to Praśastapāda, Vaiśe•ikas had the reputation of being atheists, aggressively challenging the theological claims of their fellow Hindus — Praśastapāda added God to the system, making it respectable in the eyes of fellow Hindus.
13. Nanhai...: 因明著功。鏡徹陳那之八論 (一觀三世論。二觀總相論。三觀境論。四因門論。五似因門論。六理門論。七取事施設論。八集量論也). (T.54.2125.230a6-7).
15. Paramārtha's 真諦 tr: 無相思塵論 T.31.1619; Xuanzang's tr: 觀所緣緣論 T.31.1624; Yijing's tr, with Dharmapāla's commentary: 觀所緣論釋 T.31.1625. 16. The Tibetan tradition attributes this short text to Āryadeva, not Dignāga, but both Chinese versions assign it to Dignāga. It discusses the rope-snake analogy, correct cognition, etc. An English translation (from the Tibetan), accompanied by a reconstructed Sanskrit version, the Tibetan and two Chinese versions was published by F.W. Thomas and H. Ui, " 'The Hand Treatise,' A Work of Āryadeva," Royal Asiatic Society of Great Britain and Ireland, April 1918, 267-310.
17. (1) Sāmānya-lak•aa-parīk•ā 觀總相論頌, T.31.1623, tr. by Yijing in 711; (2) Ālambana-parīk•ā (see n. 15); (3) Nyāyamukha 因明正理門論 T.32.1629, tr. by Yijing in 711, previously translated by Xuanzang in 650 因明正理門論本 T.32.1628 (Yijing adds some fi ller at the beginning, otherwise his translation is almost a verbatim repeat of Xuanzang's tr.); (4) Upādāya-prajñapti-prakaraa 取因假設論, T.31.1622, tr. by Yijing in 703. 18. Two additional texts (or one text and its –v•tti) were translated early in the eleventh century by Dānapāla. (1) Prajñāpāramitāpiārtha-sagraha-kārikā 佛母般若波羅蜜多圓集要義論, T.25.1518 (tr. in 1011); (2) Prajñāpāramitāpiārtha-sagrahav•tti 佛母般若波羅蜜多圓集要義釋論, T.25.1517. The fi rst text consists of verses summarizing Prajñāpāramitā thought, and the second is a commentary on the verses. Sanskrit exists for the kārikās, as does a Tibetan tr. (P #5870; D #3809)..
20. For a full accounting of the known works of Dignāga, see M. Hattori (1968) Dignāga, On Perception, Cambridge, MA: Harvard University Press, pp. 6-11, though this was before the recovery of the Jinendrabuddhi commentary on Pramāasamuccaya. Hattori points out (p.10) that Kamalaśīla in the Tattvasagraha-pañjikā quotes two short passages from the Hetumukha, providing us with confi rmation of its Sanskrit title as well (p. 10, n. 56). 21. Fodijing 佛地經, T.16.680, translated by Xuanzang 玄奘 in 645.
22. That this text offers three opinions on most matters has been recognized from early on. E.g., to explain the word "Bodhisattva" 菩提薩埵, Huilin 慧琳 writes in Yiqiejing yinyi: 一切經音義:「菩提薩埵 (依佛地論云 親光菩薩 以三義釋 菩提薩埵 具如彼論中) 。」"'BODHISATTVA’ (As in Bodhisattva Bandhuprabha’s Buddhabhūmy-upadeśa, there are three explanations of the meaning of ‘Bodhisattva’). (T.54.2128.449b11). 23. Peking edition, no. 5498; Derge n. 3997: ‘phags pa sangs rgyas kyi sa’i rnam par bshad pa. The title of this text is usually Sanskritized as Buddhabhūmi-vyākyāna.
24. On this, see Keenan (1980), who summarizes the arguments of KATSUMATA Shunkyo 勝又俊教. Lusthaus (2002) chapter 15, argues against the Dharmapāla attribution for several reasons, including the fact that Śīlabhadra was Dharmapāla’s student and it would be highly unusual for a teacher to write a commentary on his own student’s work. There are also serious incongruities in Kuiji’s account of the transmission. Most citations of the BBh-U in other texts refer to its author either as “Bodhisattva Bandhuprabha” or Bandhuprabha, etc.”
Kuiji, in fact, seems to suggest that Dharmapāla was a contributor to the Fodijing lun in one of his Cheng weishilun commentaries, mentioning him and “Bandhuprabha, etc.”: 《成唯識論述記》卷1：「佛地一師。作如此解 護法．親光 等云。」 (T.43.1830.230b18-19). However, Huizhao 惠沼, in his Cheng weishilun commentary, cites Wŏnch’uk (圓測 = 西明) as indicating that Bandhuprabha, in the BBh-U, expressed an opinion that is different from that expressed by a position Kuiji identifi es as by
Dharmapāla in the Cheng weishilun: 《成唯識論了義燈》卷1：「又二障體。佛地論云。煩惱障體。一根本煩惱及隨煩惱。二所發業。三所得果 若此論下一百二十八根本煩惱及隨煩惱以為體性 西明云。如此二論障體不同者。護法就正障說根本．等流以為障體。親光通據障及眷屬故障．業．果為障體性 若准此釋亦未盡理。護法初出障體且據勝說。非不取業及果。以不說唯故。又下明十障中說斷惡趣雜染愚。云應知愚品總說為愚。明知障品亦說為障。」(T.43.1832.672b8-17).
Lusthaus, ibid., raises strong suspicions against taking attributions by Kuiji in his Cheng weishilun commentaries uncritically, since it is easily demonstrated that he was unaware of which positions derived from Sthiramati’s commentary, the only source commentary to survive independently. Kuiji himself suggests elsewhere that Dharmapāla, Bandhuprabha and Sthiramati were contemporaries (without mentioning Śīlabhadra): 《成唯識論料簡》卷2：「同時即有 護法，親光，及 安慧，等」 (CBETA, X48, no. 806, p. 362, a6-7 // Z 1:76, p. 479, c9-10 // R76, p. 958, a9-10).
25. SAKUMA Hidenori has very kindly provided me with his parallel critical edition of this section, hopefully soon to be published. It confi rms the absence of any Tibetan parallel. Prof. Sakuma also provided his parallel version of a second BBh-U passage that I discuss at the end (see n. 92, below). That passage does have a Tibetan parallel though there are interesting differences between the Chinese and Tibetan versions, as will be discussed when we get to it. Notably that passage does not offer alternate theories.
26. Hattori (1968) offers a translation of the fi rst chapter of Dignāga’s Pramāasamuccaya on perception (pratyak•a), though the Sanskrit commentary by Jinendrabuddhi now being made available should revolutionize Dignāga studies and how we understand this chapter. A translation and study of Ālambana-parīk•ā is Tola and Dragonetti (1982), based on the Tibetan. No Sanskrit has been discovered, though Chinese versions, which vary from each other and the Tibetan, are available (see n. 15).
27. A slightly abridged but generally reliable translation of Vigrahavyavārtanī is Bhattacharya (1986), which also contains the romanized Sanskrit text. A philosophically astute translation of Mūla-madhyamaka-kārikā made from the Tibetan version, not the Sanskrit, is Garfi eld (1995). A Chinese tr. of Vigrahavyavārtanī by Vimok•aprajñā••i and Gautama Prajñāruci — Huizheng lun 迴諍論 T.32.1631 — is of mediocre quality.
28. A Sanskrit equivalent for 證自證分 is not attested. Prof. KATSURA Shoryu has suggested to me these possibilities: svasavedana-savedana-bhāga, svasavit-savid-bhāga, or svasavitti-savitti-bhāga. Katsura agreed with a point I raise later, that, since bhāga is not attested for any of the four components in any extant Sanskrit work, it may be wise to drop the term “bhāga” from our treatments. The Chinese 分 fen (which means ‘division, part’) may have been added by Xuanzang as a topic marker to reinforce that nimitta, darśana, svasavitti, and this fourth aspect are ‘partitions’ that distinguish facets of consciousness, i.e., the nimitta ‘part,’ the darśana ‘part,’ and so on. Though it may just be a marker added in Chinese to which nothing in the Sanskrit text corresponded, I will continue to translate fen in what follows since, whatever its justifi cation or implication, fen does appear in the Chinese text. I am grateful for the many helpful comments Prof. Katsura offered on an earlier draft of this paper.
29. More precisely, Buddhists largely abandon it, but something comparable, using different terms, does appear later in some Hindu, especially Nyāya, formulations. Cf. Williams (1998) chs. One and Two which trace out various developments of what corresponds to three of the four bhāgas from Dignāga through Śāntarak•ita and subsequent Buddhist thinkers, often raising arguments whose roots are found in this section of BBh-U.
30. Cheng weishilun (T.31.1585) is traditionally held to be a compendium of ten Indian commentaries on Vasubandhu’s Thirty Verses (Triśikā) compiled and translated by Xuanzang in 659. I have argued elsewhere against uncritically accepting that it is composed from ten commentaries, with Dharmapāla invariably providing the “correct” interpretation. Kuiji is our main source for that idea, not the Cheng weishilun itself, and his attributions are problematic. 31. On distinguishing these two senses of ākāra, cf. Kuiji 成唯識論述記 T.43.1830.500b19501a6. The meaning of ākāra as the physical qualities of an object, which Kuiji associates with Hīnayāna, is considered by him 相狀 sasthāna (size, shape, etc.), rather than ākāra proper.
32. Tadākāratā 帶彼相; cf. Abhidharmakośa-bhā•ya, Pradhan ed., p. 474; Xuanzang’s tr. T.29.1558.157b23-24; Dignāga’s Ālambana-parīk•ā, T.31.1624.888b8-12; Cheng weishilun T.31.1585.4b26 and 49c23-24. Since Chinese uses 相 xiang for both ākāra and nimitta, and uses 似 si for a number of terms (sād•śya, ābhāsa, nirābhāsa, etc.), unfortunately it is not always clear which original Sanskrit term underlies specifi c occurrences of the Chinese equivalents in other contexts. Some consideration of context and related literature can help make guesses more “educated” than capricious, but uncertainties remain. See n. 74. 33. T.26.1530.303a26-c28.
34. I am translating 證 zheng here as “aware,” though, as will be discussed in more detail below, it also means “verify,” etc. With the latter sense, this line would mean: “Cittas and caittas verify themselves.” This implication will come into play later in the passage. However, in the light of the objection raised in the next line (see below), the meaning of “aware” best fi ts the present context, confi rmed by the Sanskrit version of this statement. Zheng here is the Chinese equivalent for savedana.
35. Prakāśa, in Ch. 照 zhao; both the Chinese and Sanskrit terms mean “to illuminate, to shine a light on, to make visible.” This theory held that perception was not a passive reception by sensory organs of sensory data, but rather it entailed an active intentional probing of the environment by the sense organ. Vision, for instance, consisted of a light shining out from the eye, illuminating objects, which are thereby illumined and thus perceived. This beacon theory of perception is what the term prakāśa presupposes in this context. The reader may want to substitute “perceives” for “illuminates” in the translation to get a clearer sense of the meaning, but I will translate it literally despite its semantic awkwardness if for no other reason than to highlight the prakāśa theory of perception and to remind the reader of the active, intentional dimension of perception presupposed by Buddhists and others during this period.
36. This line was taken by the subsequent East Asian traditions as if it were an actual translation from the Pramāasamuccaya, and not a paraphrase. E.g., it is cited as a quotation of the Pramāasamuccaya in Wŏnhyo’s 元曉 commentary on the Awakening of Faith, 大乘起信論別記 at T.44.1845.236b3-5 and in Zhizhou's 智周 Cheng weishilun commentary 成唯識論演祕 at T.43.1833.866b24-25.
37. This was fi rst pointed out to me by S. Katsura.
38. Bhikkhu KL Dhammajoti (2007), Sarvāstivāda Abhidharma, Hong Kong: University of Hong Kong, pp. 25-26. The two quotations from the Mahāvibhā•ā in this citation occur at T.27.1545 405c and 65a, respectively.
39. Dhammajoti's Abhidharma Doctrines and Controversies on Perception (2007, University of Hong Kong) presents much abhidharma material relevant to this issue; it also provides pertinent contextual background for many of the issues BBh-U engages in the passage that is discussed below. See especially ch. 10, “Ākāra, sākāra-vijñānavāda, nirākāra-vijñānavāda.” The debates recounted there and passim on the way various Buddhist schools included “judgement” (adhyavasāya, vyavasthā, etc.), “ascertainment” (avasīyate) — i.e., the part played by a mental determination in sensory and mental pratyak•a that was neither kalpanā nor anumāna — as well as some differing theories on what an ākāra is and how it participates in cognition, are especially germane to the BBh-U’s discussion.
40. Most scholars have presumed that svasavitti must have some veridical function, since they assume a pramāa can only provide “valid” knowledge. This, again, is a Dharmakīrtian presumption, derivative of Nyāya and Vaiśe•ika theories of pramāa (not initially shared by others). Dignāga himself never states that, nor does he defi ne a pramāa as entailing “truth” in any form. It makes arthas (meanings, referents, things) known. Only at the initial moment of discovery can a pramāa be called a pramāa. Once something is already known, the means of sustaining such knowledge can no longer be called pramāa according to Dignāga. Note that an implication of this requirement of radical novelty is that inference is only a pramāa when it is causing something to be known or discovered for the fi rst time (if I convince you by inference of something I consider true, my argument is a pramāa for you, but it is not a pramāa for me). However, since perception is always novel, perception, in whatever aspects exclude conceptualizations (kalpanā), is always a pramāa. 41. Cf. PS 1:8cd-9ab:
savyāpārapratītatvāt pramāa phalam eva sat || 8 || svasavitti phala vātra tadrūpo hy arthaniścaya | 9ab We call [[[cognition]]] a pramāa, though it is really the effect.
Or else, the svasavitti is the effect, since it essentially is that (tadrūpa) which determines the artha. 42. Ernst Steinkellner (April 2005), "Dignāga's Pramāasamuccaya, Chapter 1,” PDF available at www.oeaw.ac.at/ias/Mat/dignaga_PS_1.pdf. We should note—as a caution against the tendency to treat Chinese translations as transparent word-for-word windows onto Sanskrit originals — that if Steinkellner's Sanskrit is accurate, the 相 in 能取相 would be ākāra, not nimitta, though this arises during the Cheng weishilun’s discussion of nimitta-bhāga (相分), thus blurring the line between nimitta and ākāra (a recurrent and persistent ambiguity in texts like Cheng weishilun). For that matter, the 相 in 似境相 would be gratuitous with no Sanskrit counterpart at all, and thus not nimitta either. The 體 in the last pada (此三體無別; traya nāta p•thak k•tam) is also a potentially misleading — because of the glorifi ed connotations with which the term 體 has been imbued by the Chinese Buddhist tradition — interpretive gloss added by Xuanzang that lacks a Sanskrit equivalent. In most other respects, the Chinese and Sanskrit match up well.
43. Hattori 1968, 29.
44. Both Dignāga and Dharmakīrti derive these four types from Asaga’s discussion of pratyak•a in the hetu-vidyā section of the Śrutamayī-bhumi in the Yogācārabhūmi (T.30.1579.357c19-29; cf. Alex Wayman, A Millenium of Buddhist Logic, Delhi: Motilal Banarsidass, 1999, pp. 17f for his Sanskrit transcription). Asaga designates the four as: rūpīndriya pratyak•a 色根現量; (2) mano ‘nubhavapratyak•a 意受現量; (3) lokapratyak•a 世間現量; and (4) śuddhapratyak•a 清淨現量, which he subsequently explains in more detail.
Dignāga's version (Pramāasamuccaya-v•tti I.5-6; Cf. Nyāyamukha 16 and v•tti) are: (1) rūpam indriyagocara… pañcendriyaja pratyak•a-jñāna nirvikalpam, (2) mānasam ... anubhavākāra-prav•tta, (3) rāgādi•u ca sva-savedanam indriyānapek•atvān mānasa pratyak•am, and (4) yoginām apy āgama-vikalpāvyavakīram arthamātradarśana pratyak•am.
Dharmakīrti (Nyāyabindu I.8-11) denotes them thus: (1) indriya-jñānam, (2) sva-vi•ayânantaravi•aya-sahakāriendriya-jñānena samanantara-pratyayena janita tan-manovijñānam, (3) sarvacitta-caittānām-ātma-savedanam, and (4) bhūtārtha-bhāvanā-prakar•a-paryantaja yogi-jñāna. When comparing each of these lists, it is the third item that varies most, which may provide clues to key differences between their respective systems. Following such an analysis is beyond the scope of the present paper, except to note that both Dignāga and Dharmakīrti explicitly associate the third type with svasavedana / ātma-savedana.
45. Cf. Nyāyamukha 16 and v•tti.
46. There has been some discussion in recent years, especially among some Japanese scholars, as to whether Dharmakīrti's dates should be pushed somewhat earlier based on possible mentions of or quotations from him retained in Chinese translations. On this, see FUNAYAMA Tōru, “Two Notes on Dharmapāla and Dharmakīrti,” Zinbun (2000), 35, 1-11, and the sources he cites. He mentions the line we are discussing on pp. 5-6. One of the possibilities entertained in such discussions is the infl uence Dharmapāla’s hetu-vidyā writings — which are not extant — may have had on Dharmakīrti. On this, see also Tom Tillemans, “Pre-Dharmakīrti Commentators On Dignāga’s Defi nition of a Thesis,” in Tillemans (1999), Scripture, Logic, Language, 53-66.
47. BBh-U merely asserts this here, without explaining why self-awareness is a necessary prerequisite for memory. PS I makes a similar assertion, but it also fails to offer a clear and compelling reason for why that would be the case. The closest it comes to formulating a reason is PS I.11ab-v•tti, which Hattori translates as: “Further, [if the cognition had only one form, either that of the object or that of itself,] then the object which cognized by a preceding cognition could not appear in a succeeding cognition. Why? Because that [[[object]] of the preceding cognition does not exist when the succeeding cognition arises and] could be the object of the latter. Hence it is proved that cognition has two forms.” (p. 30) This is not much of a reason, and seems to beg the question rather than providing a proof.
49. I will use svasavitti in the remainder of this paper to indicate the range of terms — svasavitti, svasavedana, ātma-savedana—since the Chinese term 自證 zizheng does not allow us to differentiate between them.
50. Cf. Nyāyapraveśa: 現量謂無分別 = pratyak•a kalpanāpoha. T.32.1630.12b28.
“What is perception? It is not unobserving; not prior consideration nor consideration of subsequents; and undistorted.” Note that Asaga is already employing a type of apoha for this defi nition since all the terms are negative exclusions. His subsequent explanation of anabhyūhitam anabhyūhyam (非已思應思) prefi gures Dignāga's kalpanāpoha. Note also that Asaga includes avibhrānta in his defi nition, which not only anticipates Dharmakīrti’s emphasis on this as part of the defi nition of pratyak•a, but is also mirrored in a terse manner by Dignāga at PS I:7cd-8ab: bhrāntisav•tisajjñānam anumānānumānikam || smārtābhilā•ika ceti pratyak•ābha sataimiram.
52. Cf. Aguttara Nikāya I, PTS p. 199, etc. I am currently working on a paper detailing this. For a glimpse of the argument to be developed there, see n. 60, below.
53. The arguments that follow are attempts to answer and refute Nāgārjuna's arguments in Mūla-madhyamaka-kārikā 7:8-12 and especially Vigrahavyāvartanī 31-51, that light does not illuminate itself or other things (Āryadeva uses the same “light doesn’t illuminate itself or others” in his Catu•atika; cf. 百論 T.30.1569, starting at 168c25). The discussion in our text echoes statements made there. Nāgārjuna’s argument is rigorously logical but metaphoric. Light illuminating itself and others is an analogy to consciousness knowing itself and others as our text is claiming as well as good vs. evil, etc. His arguments are also arguably more in line with our understanding from modern physics of how light works (light-rays in space do not illumine themselves; it is only when an object ‘rejects’ the rays, that the rejected/refl ected rays can be seen). In Vigrahavyāvartanī Nāgārjuna employs these arguments specifi cally in the context of pramāas, i.e., what validates a pramāa, and the discussion in our text echoes statements made there.
55. Asaga, Yogācārabhūmi T.30.1579.357c1-18 (cf. Wayman, Millennium, 16-17), describes two models: (1) a list of fi ve bhrānti and (2) a list of seven, which adds two bhrānti-s to the fi ve, claiming they pervade the fi ve. He offers descriptions, not criteria. For Dignāga, see PS I.7cd-8ab and v•tti; and Nyāyamukha 16-v•tti which only stipulates that a pratyak•a-pramāa must be kalpanāpoha, and hence excludes “cognitions such as memory, inference, longing, doubt, delusions and mirages, etc., from base desires, etc.” 憶念 比度 悕求 疑智 惑亂 智等 於麤愛等 皆非現量, T.32.1628.3b26-27, again, with no criteria on how to distinguish within a perception itself — or otherwise — an accurate perception from one that is 'erroneous.'
56. Even what is conventionally true (sav•ti-sat) or proven true by inference are excluded from pratyak•a-pramāa. This is not simply a matter of maintaining the integrity and separation of the categories of pratyak•a and anumāna, but impinges on what one means by truth value in the case of perception since, as many pramāa-vādins later argued, the validity of a perception is best confi rmed by comparing a previous perception with later perceptions (to see if the object really was there and appears as it initially seemed). Such comparison is, strictly speaking, an inferential act, not simply perceptual. While secondary treatments of pramāa-vāda often ignore or side-step this, the ābhidharmikas were more rigorous, side-stepping it in ingenious ways that would have been familiar to Dignāga. Cf. e.g., Dhammajoti, Abhidharma... Perception, 172-73.
Grasper and grasped,
Which are the mind perceiving the multitude of things.
58. There is a possibility that ti-yong here does refl ect something in Sanskrit, such as svabhāva and k•tyā. But there is an equal or better chance that it does not. In other works, as in his translation of the Abhidharmakośa-bhā•ya, Xuanzang sometimes introduces tiyong distinctions in places where nothing corresponds in the Sanskrit.
60. 證 zheng has a number of meanings. It can mean “to witness”; it also means “to realize, to have a realization.” It can also mean “to verify, prove, validate.” Dignāga’s notion of svasavitti is derived, I believe, from a set of terms that mean “to attain a realization and be aware that one has attained that,” terms which already appear in the early Pali texts, but increase in frequency in the later strata of Pali materials. 自證 zizheng, used here by Xuanzang for svasavitti, was used earlier by translators for Pali sacchikiriyā (realization, experience), related to “to see with one’s own eyes, to experience for oneself” (sacchikaroti); “(able) to be experienced” (in four ways: by kāya, sati, cakkhu, and paññā, i.e., by body, mind, eye, and intellective cognition) (sacchikaraīya). In the Yogācārabhūmi Xuanzang uses zheng to render pra-√āp, anuprāpt, prāpnoti; adhi√gam, adhigama, adhigacchati; abhi-sa-√budh, abhisabudhyate; avetya, upapatti, samudāgama, sāk•in, sāk•āt-karoti, sāk•āt-k•ta (which have similar meanings and implications to the Pali terms cited above); and he uses zizheng to translate svayam abhijñayā sāk•āt-k•tvpasapadya (attaining direct higher knowledge for oneself). So zizheng means both “to be aware of itself” (or “to experience for oneself”) and “to verify itself, to be self-validating.” According to BBh-U, cognition in toto verifi es itself by being composed of parts that treat other aspects of cognition as cognitiveobjects, thus verifying them the way that directly perceiving a sensory object verifi es its presence. While I am translating zheng as “aware” (“the self-aware part” 自證分 and "being aware of being self-aware part” 證自證分), the reader should keep in mind the additional sense of "verifi cation" which this passage has put into play, but with the caveats discussed previously. What I am suggesting, speculatively, is that this additional sense of verifi cation was not part of Dignāga’s original understanding of the term pramāa, but only developed subsequently, in fact, precisely in the environment that felt it necessary to add a fourth bhāga for the express purpose of providing such “verifi cation,” something which the BBh-U commentators obviously believed Dignāga himself had not adequately dealt with. In this text, which clearly comes after Dignāga since it cites him, pramāa had acquired some sense of “valid” or “validating” cognition, a sense that arguably it does not yet contain in some of the pre-Dignāgan pramāa literature. Dignāga’s inclusion of svasavitti as an integral component of pratyak•a was attacked by others, for instance by Kumārila Bhaa in his Ślokavarttika, who argues that a consciousness cannot know itself, i.e., be its own object. It may be in response to this sort of criticism whether from Kumārila or earlier Mīmāsikās, etc. that this four bhāga system, which tries to explain how parts of consciousness become objects (ālambana) for other parts of consciousness, arose. This is a complex topic that I plan to take up elsewhere.
61. After disproving that either an atom or a collocation of atoms can be or produce an ālambana, Dignāga resists jumping to an idealist conclusion. The penultimate sentence of Ālambana-parīk•ā reads: 根境二色與識一異或非一異。隨樂應說。 "As to whether the two rūpas — viz. indriya and vi•aya — and consciousness are the same or different, or whether they are neither the same nor different: One can say according to one’s wishes (隨樂, yathā-āśaya).” (Xuanzang’s version; T.31.1624.889a8-9). Paramārtha’s tr. is slightly different, but basically carries the same meaning: 識者或異二或不異二或不可說。"Some [claim that] consciousness is different than those two; some [claim] it isn't different from the two; some are unable to say (one way or another)." (T.31.1619.883b21-22) The Tibetan is close to both: rnam par shes pa las de gnyis gzhan nyid dang gzhan ma yin pa nyid du ci dgar brjod par bya’o. The Tibetan ci dgar, like Xuanzang’s 隨樂, means "as you wish," (*yathā-āśaya and comparable expressions).
63. 執計 here could be taken as an unusual shorthand for 偏計所執 parikalpita, though the reversal of the order of 計…執 suggests otherwise. 計 by itself means gaana, “reckoning, counting, calculation; considering, supposing.” Xuanzang also uses it for terms that mean to believe in or trust in something (truly or falsely), as well as for several terms that mean to imagine, or imaginatively construct something. Rather than take 執計 here as a compound (for a √kp term), I am treating them as distinct terms: 執 = attachment, holding (a position); 計 = calculatively imagining, i.e., schematizing. That 執計 are meant to be taken as distinct terms, and not as a single compound, is confi rmed a little further in the text where we fi nd the expression 不執不計, which clearly treats 執 and 計 as distinct. This passage, in fact, may hold a key to why Xuanzang introduced a new Chinese rendering — 偏計所執 — for the term parikalpita that previous translators (and occasionally Xuanzang himself) tended to render as 分別, so that parikalpitasvabhāva had been rendered 分別性. 64. Not only is the “appropriating” aspect of cognition absent (“not like pincers”), but the standard prakāśa theory is also being rejected (“not like a lamp…”). A different prakāśa theory will be offered shortly in its place (“it is not that [this cognition] lacks the function which illuminates intentional objects [[[artha]]], [[[mentally]]] replicating cognitive objects [vi•aya] from those cognitive conditions [[[ālambana]]].”). Cf. Xuanzang’s disciple, Pu Guang’s 普光 commentary on the Abhidharmakośa-bhā•ya, 俱舍論記 T.41.1821.26b-c and 49a11-25. This model is explicitly applied to the Tathāgata's cognition later in BBh-U, 317a25-c9.
65. This could also be translated “perceiving the refl ections of illuminated things.”
Some of the diffi culties of such passages in Chinese is that several key Sanskrit terms come to be represented by the same Chinese term. 似, as mentioned, can represent sād•śya (and related forms), ābhāsa, pratibhāsa, etc. 相 can render nimitta, ākāra, lak•aa, etc. Sometimes distinguishing between the Sanskrit options is vital for keeping the meaning clear, a task for which the Chinese can be unhelpful. Kuiji, for instance, in his commentaries on the Cheng weishilun and Yogācārabhūmi, when discussing the bhāgas, seems, on the one hand, to be aware that the Chinese term 相 is being used in the sense of ākāra. Nonetheless, misled by the Chinese, he also insists that “for Mahāyāna” 相 in this context is 相分 (nimitta-bhāga), as if the two 相 refer to the same term, whereas 相分 renders nimitta-bhāga. The problem of confl ation of terms, or rendering different terms by a single word in translation, extends to English, in which ābhāsa, pratibhāsa, nirābhāsa and several other terms are diffi cult to render as anything other than “appearance”; 相, 影, 影像, etc., end up as "image,"; and so on. 67. 影 means "refl ection" or "refl ected image" (as I translated it in a previous sentence), but it also means "image." It is one of the Chinese equivalents for pratibimba, especially when a distinction is being made between a bimba (the original thing of which one perceives an image) and pratibimba (the “refl ected” or echo version of the bimba). Sometimes
in Sanskrit literature the terms bimba and pratibimba are used interchangeably, and sometimes they are differentiated in terms of the primacy of bimba. One way that Xuanzang indicates this differentiation when he is so inclinedis to render bimba as 質 or 本質 and pratibimba as 影 or 影像. That distinction does not seem to be in play here, since the term 質 does not appear. In this passage, the sense of pratibimba as signifying both “image” (as the images which appear in a nimitta-bhāga) and the sense of pratibimba as “refl ection” (what appears in the “bright mirror”) are evoked.
68. 經說佛智 現身土等 種種影像 如鏡等故 (T.31.1585.46a5-6).
69. Adhigata = 得, found, obtained, acquired; gone over, studied, learnt. On this type of āśraya-parāv•tti, cf. Cheng weishilun T.31.1585.50b25-26. 70. For instance, Kuiji and Wŏnch'uk, in their extant commentaries, such as in their Heart Sūtra commentaries, debate at which stages in the practice, and in conjunction with which of the eight consciousnesses, āśraya-parāv•ttis take place. The Cheng weishilun delineates eight types of āśraya-parāv•tti. Applying it to the overturning of each the fi ve skandhas is already an idea developed in Abhidharma literature. See also Xuanzang’s tr. of Mahāyānasamgraha T.31.1594.149c1-2; Vasubandhu’s Mahāyānasagraha-bhā•ya T.31.1597.371c24-25; and Asvabhāva’s –īka on these, T.31.1598.437c22-23.
71. 無分別 = nirvikalpa. Vikalpa can mean (and is usually translated as) “discrimination,” so the negative version, nir-vikalpa, would mean “non-discimination.” Vikalpa, deriving from the √kp root, can also at times be used as a synonym for kalpanā, kalpita, etc., all implying some sort of mental or imaginative construction; Paramārtha tended to translate all such kp terms as 分別, and on occasion Xuanzang does as well. It may be this parikalpa meaning that is at play here, or just as likely kalpanā; i.e., for this theory, the mental components do not “imagine.” They simply see things as they are (tathatā). If so, one might translate this thus: “Because they are devoid of imaginative construction, their ālambanas are tathatā [things just as they are], since that is inseparable from [nondiscriminative cognition] itself.” That is, non-imaginary cognition is, by its nature, tathatā, and can only see things as they are, since it fails to superimpose erroneous mental constructions. Its cognitions (and cognitive objects) share the same “just-as-itis nature.” One might be tempted to read the passage as claiming that tathatā itself is the object of such cognitions, but Yogācāra literature is divided on whether tathatā is or isn’t an “object” properly speaking, some texts saying yes and others no. Different sections of BBh-U lend themselves to one or the other position. For Yogācāra, terms with –tā suffi xes, even important terms such as śūnya-tā, dharma-tā, anitya-tā, etc., are considered abstractions, generalized conceptualizations of discrete particulars. E.g., there is no ontological status for “impermanency” (anityatā), only a world fi lled with distinct, particular impermanent things that arise and cease due to causes and conditions (and “impermanence” is not one of those causes, but only a conceptual label for the conceptualized consequences of that causal process). Similarly, tathatā is a conceptualized expression of things just as they are, not a transcendent principle or entity that either subtends or supertends them.
72. 後得智 houdezhi, Skt. p••halabdha-jñāna are the types of cognitions (jñāna) an enlightened being has subsequent (p••ha) to attaining (labdha) Awakening, which, according to some theories, may be qualitatively different from the immediate seeing of things-as-they-are as one would during the experience of Awakening. The third theory uses this distinction to account for how Buddhas, etc., can still make necessary distinctions, engage in the conventional world, and experience the plethora of things that appear in the image part (nimitta-bhāga) without undermining the fact that, in some sense, Buddhas, etc., have transcended the cognitive obstructions (jñeyāvaraa) that usually limit cognition to only seeing the world that way. 73. This would be an absurdity according to Dignāga's defi nition of an ālambana in his Ālambana-parīk•ā. According to Dignāga, an ālambana must satisfy two criteria: (1) it must cause a cognition, and (2) it must convey its own image to the cognition. The objection raised here is that it would fail the second criterion.
74. The phrase 似彼相 is reminiscent of the important phrase 帶彼相 (or 帶己相) that one fi nds in Xuanzang's translations of the Abhidharmakośa-bhā•ya, Ālambana-parīk•ā, and Cheng weishilun, the latter phrase(s) meaning “conveys its own image.” The Sanskrit corresponding to 帶彼相 in the Kośa is tadākāratā (tad = 彼, ākāra = 相, tā = 帶). The phrase here, 似彼相, also occurs in the Cheng weishilun (T.31.1585.4b7-9, 13a2, and 39c15) and in Xuanzang’s translation of Dharmapāla’s commentary on Āryadeva’s Catuśataka (大乘廣百論釋論, T.30.1571.215a2). Yijing also uses the phrase in his translation of Dharmapāla’s 成唯識寶生論, a commentary on Vasubandhu's Viśatikā (T.31.1591.79b13-14; 79c20; 96b09-12). 似彼相 occurs three times in BBh-U, the other two times signifying Buddha successfully conveying the crux of his teachings to others (cf. T.26.1530.292a3 and 327c26).
75. This fi nal tag is crucial. It is not extolling an ineffable reality, but making clear that the basic components of Yogācāra doctrine, such as mind (citta), mental associates (caittas), etc. are all only vyavahāra, conventional descriptive terms, not the names of ultimate realities, much less anything absolute.
77. While that is a plausible interpretation of Sthiramati's thought, nowhere in his Triśikābhā•ya is there any discussion of these bhāgas. Kuiji attributes the theory of a fourth bhāga to Dharmapāla. But the Cheng weishilun itself is content to roll the third and fourth bhāgas into darśana-bhāga, i.e., giving (allegedly) Sthiramati the fi nal word.
83. "Declaring falsely that one has realized the ultimate truth and is in possession of supernatural abilities. Believing that one has great virtue when one does not. This is the fi fth of the seven kinds of pride 七慢, and the fourth of the eight kinds of pride 八慢." DDB.
84. Keenan (2002) 127 tries to suggest an idealist reading here by supplying a different bracketed addition: "But this Scripture on the Buddha Land says only that those images appear and not that they are produced [by any self]” (square brackets by Keenan). While it is always possible to mount cases for alternative interpretive glosses on passages that seem to require some fi lling out, Keenan’s reading is strongly opposed by this passage, especially as formulated in the Tibetan version (see below). The Tibetan interpretation places great emphasis on the fact that the arrogant error made by ordinary people is to mistake the refl ection of their face in a mirror for themselves. That is, the object being refl ected in the mirror (one’s face) is not actually in the mirror, only its appearance is. Mistaking the refl ection (pratibimba) for the thing being refl ected (bimba), i.e., misidentifying oneself (or reality) with a refl ected appearance, is the cardinal cognitive error. The refl ected face is in the mirror, while the actual face itself is not. “Cognition,” however, takes place when the mirror, the object that provides its image, and the refl ected image all collude. The Chinese text, immediately preceding the section I am translating above and thus contextualizing it, even in Keenan’s own translation, makes explicit the point I draw on for my translation of this line: “…when [[[mirror wisdom]]] arises it is able only to be the enabling cause [for these results]. It is not an eternally creative agent… it is not to be understood as the non-Buddhists understand Maheśvara, who is falsely thought to be the creative agent of the world and whose nature is held to be eternal” (ibid). What Keenan translates as “enabling cause” in this passage is adhipati-pratyaya 增上緣. Yogācāra texts offer a variety of explanations of adhipati-pratyaya, the most basic being a condition which is one of a number of synchronic, simultaneous conditions of different types, cooperating to produce an effect. In short, the mirror cognition is a contributive condition in tandem with other conditions that are distinct from it. If by “any self” Keenan means something Maheśvara-like, then his translation is justifi ed but misleading, since it seems to imply that the mirror, rather than a self (person) is causing the image. The mirror alone is no more responsible for the refl ection that appears in it than is the face by itself.
86. It is important to keep in mind that the passage is claiming that images appearing in unenlightened cognition are fl uctuations in consciousness because consciousness is the dominant condition (adhipati-pratyaya) in their construction, not their cause, whereas, as it states at the end of the passage, the “round mirror cognition” is the cause (hetupratyaya) of the ability to perceive such images without differentiation. The authors were being more careful with their terminology than have most subsequent interpreters.
87. It is important to be clear about what is being claimed here and what is not being claimed here. The images that appear in perception are “alterations” (pariāma) of one’s own consciousness. The Cheng weishi lun consistently uses bian 變 (pariāma) to indicate how consciousness responds to stimuli from without by projecting an image based on one’s accumulated habits (seeds and vāsanās). Ordinarily one cannot “see” past one’s own superimpositional projections. One reduces the object that is causing the cognition (nimitta) to one’s image of it. This, the text tells us here, is called pariāma of consciousness because consciousness is the dominant (adhipati) factor in the constitution of that image. It is the dominant factor, not the exclusive factor. The passage does not mean, as a casual reading might suggest, that consciousness is creating the actual things outside one’s consciousness that are being subject to the pariāmic reduction.
88. 有種 here could also mean "has its distinct seeds (bīja),” which would set up the contrast that is about to appear between these “seeds” and “habits” (vāsanā), though that distinction, if it is indeed being alluded to, is not developed here. Zongmi 宗密
refers to this passage in his commentary on the Perfect Enlightenment Sūtra, 圓覺經大疏 Yuanjue jing da shou:「然內外四大，雖各有種，外起現行 必由內變 四世間諸 法。」(XZJ:no.243.295b13-14), "Now as to the Four internal and external Mahābhūtas, although each has its own type (of seeds), the arising of the external (mahābhūtas) in actual perception necessarily derives from the internal alterations [of consciousness], those four being the worldly dharmas.” On the various types of dharmas each having their own “types,” or their own types of “seeds,” cf. Asaga’s Mahāyānasagraha ch. 1 which, e.g., differentiates ‘external seeds’ (such as wheat, rice, etc.) from internal seeds (the ālaya-vijñāna’s latent tendencies). Some Yogācāra sources, such as Cheng weishilun, argue that each of the eight consciousnesses has its own seeds which, though stored in the ālaya-vijñāna, are distinct from the eighth consciousness and do not originate there (cf. discussion of the āśrayas and pratyayas in the Manas section of CWSL).
89. A less likely, but also possible translation: Taking 還 as niv•tti, pratiniv•tti, pariv•tti, etc. — which are attested equivalents — it can mean not only "return," but to "desist," to come to an end. In the latter sense, this sentence would mean: "When the present alteration ceases [to be perceived], it is not apart from consciousness," i.e., it is remembered, it becomes a further vāsanā. 還 might also refer to the light that bounces back from the object in refl ection, à la prakāśa. In short, while I added “vāsanā” in brackets in the translation above as one possibility for whence arises the imaging-potential, the text does not actually say that, and it could just as well be from some object out there. That indeterminacy is a recurrent aspect of Yogācāra texts (and an indeterminacy explicitly pursued by Dignāga in Ālambana-parīk•ā and Pramāasamuccaya). It is signifi cant that the passage emphasizes that the nonseparation of consciousness from its images occurs “at the moment when alterations appear,” i.e., when they enter one’s conscious domain.
90. 真實義 = a truthful expression, one that expresses a contextual truth. As DDB explains:
The expression of truth, of which four kinds are listed in the Púsà shànjiè jīng: The expression of truth in everyday speech 世流布眞實義, wherein sentient beings, when seeing the earth, call it earth, and when seeing fi re, call it fi re, without confusion.
The expression of truth that cleanses affl ictive hindrances 淨煩惱障眞實義, by means of which śrāvakas 聲聞 and pratyekabuddhas 緣覺 use uncontaminated practices to destroy affl ictions and attain unobstructed wisdom.
義. Although śrāvakas and pratyekabuddhas may have attained unobstructed wisdom, they are still not able manifest the truth of the middle way. Through the expression of this truth, bodhisattvas and buddhas eliminate the cognitiv e hindrances and naturally manifest the truth of the middle way. ［菩薩善戒經, T 1582.30.968a-b) 91. For example, in addition to Zongmi's paraphrase cited earlier, quotes also appear in Wŏnch'uk's 圓測 Jie shenmi jing shu (Commentary on the Sadhinirmocana Sūtra) 解深密經疏, fasc. 6: (XZJ:no.369,.852a6) and in Zhizhou's 智周 Cheng weishilun Yanmi (Expounding on the Implicit within Cheng weishilun) 成唯識論演祕, fasc. 6, (T.43.1833.934c27).
92. I gratefully thank H. Sakuma for providing me with this section of his hopefully soon to be published critical parallel edition of BBh-U with the Tibetan Buddhabhūmi commentary (I have omitted the variants he includes) and M. David Eckel for going over the Tibetan text with me. The Tibetan reads: ‘o na ni rang gi gzugs brnyan rdzas gzhan med pa’i ltar na | de ji ltar dper rung ste | ‘di ltar rgyu mtshan du gyur pa ma long yod na gzugs nyid de lta bur snang ngo zhe na | de la yang mngon par zhen pa’i bag chags yongs su smin pa’i dbang gis rnam par shes pa nang na gnas pa nyid de ltar snang ste | des na ‘jig rten gzugs brnyan mthong ngo zhes de snyam du nga rgyal skye’o || de nyid kyi phyir snang ngo zhes bshad kyi skye’o zhes ni ma bshad do || de bzhin du shes bya gzhan yang shes bya nang gi yin par lta ba la de dang ‘dra bar lta’o ||
des na zhes bya ba ni rgyu’i gsum pa yin te | de’i phyir me long lta bu’i ye shes zhes bya ba’i tha snyad ‘thob bo || 93. The Tib. has yongs su smin pa (= paripāka, paripākva; pariāma), while the Chinese has 變異 (= vipariāma, anyathātva). The Chinese also provides more information on the example of what the ordinary person sees in the mirror, viz. his own face. The Chinese also attributes the two quotes to the sūtra (i.e., Buddhabhūmi-sūtra), whereas the Tib. seems to treat them as what people say (Ch: “The sūtra only says that the various refl ected images appear [in the mirror]; it doesn’t say that they are produced [by the mirror]”; Tib: “’I see a [i.e., my own] refl ection (pratibimba).’ But they don’t say ‘This is [another being] being produced’.”). Both readings make sense, though they make different assertions. The Chinese appears to be the correct interpretation, since the sūtra passage being commented on reads: 《佛地經論》卷4：「經曰。復次妙生。大圓鏡智者。如依圓鏡眾像影現。如是依止如來智鏡。諸處境識眾像影現。唯以圓鏡為譬喻者。當知圓鏡如來智鏡平等平等。是故智鏡名圓鏡智。」
(T.26.1530.309a19-22). "Next, O Well-born, the Great Mirror Cognition: just as a mirror is the basis for the appearance of a multitude of refl ected images, just so is the mirrorcognition of the Tathāgata a basis for the appearing of a multitude of refl ected images of the āyatanas, vi•ayas, and vijñānas. ‘Mirror’ is only used as an analogy. You should know that the mirror and the Tathāgata’s mirror-cognition are the same-same (samasama). Hence the (Tathāgata’s) cognitive mirror is called ‘Mirror Cognition’.”
95. This passage makes little sense here, perhaps because the Tib omits a lengthy excusus in the Ch on vijñapti-mātra for which this offers a culminating conclusion. The underlying Sanskrit terms for ‘dra ba and mnyam pa are possibly sād•śya and sama (or samatā), respectively. However, one of the attested equivalents for ‘dra ba is pratibimbaka, which is intriguing. In some ways, the Chinese is no clearer, having 平等平等 (= sama-sama) here, which it brings in as the tail-end of a discussion on how citta, caittas and rūpa (when the latter is not separate from citta and caittas) are “equal,” or “the same” from the perspective of paratantra, and that paratantra and parini•panna are “both the same” (由平等故。此二平等); since they are the same, they are equally the same (sāma-sāma, 平等平等).
96. What constitutes the "three reasons" is unclear. Possibly: (1) Though the object being refl ected is distinct from the mirror, it appears as if inside the mirror; (2) the thing (bimba) and its replica (pratibimba qua sād•śya) are, for all intents and purposes identical; (3) the way a mirror works and the way the Tathāgata’s mirror-like cognition works are the same. The Chinese text, however, in the preceding discussion (not translated here — see T.26.1530.309a23-309c27; Keenan (2002) 123-27) discusses alternate theories of the "three" factors implied by āyatana, vi•aya, and vijñāna, i.e., whether this is simply the eighteen dhātus as ordinarily involved in perception, or whether the Buddha’s omniscience entails some sort of difference. The BBh-U rejects the theory that Buddha knows all things only sequentially (as might an ordinary mirror), arguing instead that it knows or refl ects all things at once (頓). Both theories of omniscience are rejected later by Dharmakīrti and Śāntarak•ita as incoherent and mutually exclusive, with Dharmakīrti rejecting the very idea of omniscience (sarvajñā) altogether (proposing, instead, that Buddha is dharmajñā, ‘a knower of Dharma’), and Śāntarak•ita, hesitant to jettison the idea of Buddha’s omniscience altogether, arguing instead (in the fi nal chapter of Tattvasagraha) that one should treat it as an open possibility.